Welcome back everyone. Summer’s over, whether we agreed to it or not, and you can feel everyone with impact returning to their desks. It shows up in the inbox, on the timeline, and most of all in the bond market.

Market Update📈📉

Warsh gave his Jackson Hole keynote and dropped the pretence:

“we have work to do.”

Core PCE is running 3.3 percent against a 2 percent target, headline 3.7 on oil, and the market has genuinely priced a September rate hike into play.

That’s a stunning shift from March. You can see it in the ten-year, in gold, and especially in small caps, which carry the highest financing costs and get hurt first when the tail of the curve rises.

IMPORTANT MARKET STRUCTURE CHANGE: While the Fed talks tough, the Treasury has been quietly running a yield-cap operation, buying down the long end of the curve, because mortgage rates ripping higher into a midterm cycle is a political problem.

The Fed is actively tightening while the Treasury tries to loosen, two arms of the same government pulling against each other, and the Fed chair doing the tightening is the one Trump picked himself. And the Treasury is losing. Yields sit above where they were when the operation started.

Why?

Go back to the Ben Thompson question we covered two episodes ago: are we going to have enough money?

For eighteen months the bottlenecks were GPUs, optics, power and the market has worked diligently toward solving each one.

The new bottleneck is capital.

For the first time in roughly twenty years, the hyperscalers are issuing serious amounts of debt, and per Cembalest’s new Eye on the Market piece, their 2026 issuance is about $320 billion, which works out to roughly 70 percent of all new long-duration Treasury borrowing this year.

When Washington goes to raise money, it now competes with Microsoft and Google for the same buyers, and frankly, on the financial metrics, some of those companies are maybe more creditworthy than the government. That competition is what’s pushing the long end up, and the long end is what sets your mortgage rates.

The AI infrastructure layer, the biggest contributor to S&P earnings growth for two years, has ground sideways to down, with many names off nearly 40 percent from the June peaks.

Beating earnings and guiding up wasn’t enough:

Earnings season is over and essentially all of the bottleneck tech companies beat and guided higher. It didn’t matter. When the cost of capital rises, your multiple compresses, and beat-and-raise stops being enough. The market wants perfection.

Is the spending justified? The revenue says yes. Google Cloud’s backlog jumped $52 billion in a single quarter to over half a trillion dollars. AWS grew 37 percent, Azure 43, Google Cloud 82. Earnings revisions are at the high end of anything we’ve seen in thirty years. Two asterisks belong on all of it, though. First, capex and R&D from these companies went from $460 billion in 2024 to nearly $900 billion over the last twelve months, so some of the market’s earnings strength is the spending itself. Second, an unusually large share of S&P earnings right now comes from unrealized gains on stakes in Anthropic, OpenAI, and SpaceX, positions marked up from raises done at $200 to $500 billion valuations. Anthropic is expected to go public in about eight weeks at something like $2 trillion. Those are real gains on paper, and paper is doing some of the market’s lifting.

My honest read: none of this is fatal.

The proverbial trains are full and running on time, and demand for the railway has never been more justified. Expect a choppy six to eight weeks while the bond market finds and appropriate yield level, and remember that these rate limiters are healthy.

Bubbles without governors are the ones that end in 2008. Bubbles with them tend to leave behind the railways and the fiber, the things that define the next century.

I think AI ends up delivering something like the railway plus the internet combined. But from here, you have to know exactly what you own.

It won’t be as easy as it was twelve months ago.

Canada note: the Bank of Canada is expected to hold at 2.25 for a seventh straight meeting. Our headline CPI of 2.8 is gasoline-driven and our core is bang on target at 2 percent, so there’s no case to hike. The loonie sits at 1.39, banks don’t see a Canadian hike until 2027, and anywhere between 1.30 and 1.40 the economy is happy. Meanwhile Alberta’s Q1 fiscal update landed with oil averaging roughly $90 against a $60.50 budget assumption. As Albertans, we are on the right side of nearly every force squeezing everyone else.

Why this isn’t the dot-com rerun

The question every client asks: is this a bubble yet? Here’s the framework that finally made it click for me.

The internet had one dimension of exponential growth: user count. Everyone paid a flat fifty or a hundred bucks a month, and using it ten times more didn’t cost ten times as much. One curve, and once penetration hit about 50 percent, it saturated. That was March 2000.

AI has two dimensions which is different than the Internet/fiber era.

Penetration crossed 50 percent a few months ago, right on schedule. But usage per person is just leaving the station. Median enterprise AI spend is about $12 per employee per month today. The plausible destination is around 10 percent of a white-collar salary, call it $1,200 a month.

And it’s already visible: a16z said this week that their portfolio founders are running at multiples of the median.

The internet stopped at fifty bucks a month. This doesn’t stop at twelve.

And the unit economics of the infrastructure have changed underneath us. A gigawatt costs $40 to $60 billion to build, roughly 70 percent of it flowing to Nvidia. But look at SpaceX’s filings renting compute to the labs and the software companies: in some cases a gigawatt is generating on the order of $100 billion in revenue. Put out $40 to make $100 in a year and tell me that’s malinvestment.

The usual caveat stands, all exponential growth is an illusion before the ceiling arrives. But this ceiling has an extra dimension holding it up.

The backlash, and the window of opportunity

The next leg of risk isn’t chips or money. It’s politics.

The blanket hatred of data centers on social media is real and spreading: New York has its moratorium, Virginia is wobbling, and now Texas, the build-anything state, just imposed a temporary pause.

Mel’s framing deserves to be quoted at length.

This is an arms race, the first our generation has seen since the space race, and the question isn’t whether the world is going there, it’s who arrives first and captures the upside. Skepticism about water and power is legitimate, and we have institutions and regulators built to answer exactly those questions. But ask who benefits when North America blocks its own growth. It’s China, period, end of story. And a society that can be talked into fighting itself over obvious opportunities is a society that can be divided and conquered without anyone setting foot on a battlefield.

Her challenge to listeners was the best moment of the episode: whenever a TikTok take hardens your politics, ask yourself whether TikTok would permit that same discourse about the Chinese government inside China.

You know the answer. So why is it happening here?

For Alberta, every moratorium elsewhere makes our window wider and more valuable than it was six months ago. Bring-your-own-generation, anchor tenants paying their own freight, ratepayer bills going down. We have the model. The only question is whether we defend it before someone imports the fear.

CUSMA/USMCA

The escalation, quickly: on August 25 Canada suspended negotiations and announced a dollar-for-dollar response to US 50 percent tariffs on $27.6 billion of Canadian exports, effective September 8, with groceries largely carved out, alongside a $7.5 billion support package.

Trevor Tombe’s estimate in The Hub is worth reading carefully: the aggregate impact is small, about 2.5 percent on the average export rate, but the aggregate hides the real concentration of impact...

JOBS: Roughly 90,000 jobs are at risk, 36,000 of them in Ontario, 9,000 in Alberta.

Mel’s reframe: this is not an economic conversation, it’s a political one, on both sides of the border. And the real strategy is about 2036, not 2026.

Trump is using tariffs and industrial policy to rebuild American primary aluminum, with an Oklahoma smelter that would double US production, because aluminum is autos, aerospace, defence, and infrastructure. The CHIPS Act taught the lesson: control of critical industrial capacity beats cheap imports. Canada can eventually negotiate tariffs down. What we cannot negotiate back is American demand once the plants are built and the supply chains reroute around us. And the unfortunate reality is that this outlasts Trump; Gavin Newsom has said as much on Canada. The next Democratic administration may be more polite about it, but wont be any easier to deal with.

The uncomfortable truth: successive Canadian governments sold voters a lie that we could have our cake and eat it too, and the first step out of the lie is telling the truth about how dependent we actually are. Admitting you’re not in charge, ironically, is what starts giving you back control. Meanwhile the US Treasury Secretary went on CNBC and joked that Canada might sic the two submarines from West Edmonton Mall on them. They’re mocking us and we’re walking into it. (Joke’s on them. We don’t have submarines at the mall anymore.)

CLOSING THOUGHTS: Keep the human brain on top of the robot legs

We closed with Lord of the Rings and Tolkien, of all things, via an essay that reads The “Lord of the Rings as a technology parable”. Sauron pours his power into an external object so it can act at scale, and the amplified power is no longer his. We’ve already handed our sense of direction to the GPS and our memory to the search bar; writing and thinking are next in line. Tolkien’s definition of the Machine was precise: external devices used as a substitute for developing your own inner powers. His test is the one to carry into the agent era. Is the tool helping you grow, or offering to grow for you?

The centaur versus the reverse centaur.

A human head on machine legs is the goal. A machine head on human legs is the trap. Mel’s practical guardrail was the right note to end on: use AI intentionally for commerce, for the government report you have to summarize. Don’t use it to find out what happens at the end of Pride and Prejudice. Some things are the journey, and the journey is where the joys of life are.

And yes, Mel has never seen The Lord of the Rings. She has promised to watch it over the holidays and report back to Jay.

New episodes every two weeks. Keep learning, keep earning, and keep investing.

Podcast & YouTube Recommendations🎙

* All Eyes on the 10yr Yield

* A16Z on the new Token Economics

Episode Article Links:

* Michael Cembalest’s “Rear Window” (Eye on the Market, Sept 1), the source for the Treasury twist and the hyperscaler-borrowing stat, plus a masterclass in grading your own calls in public

* Trevor Tombe in The Hub on what the new tariffs actually cost, read past the headline

* The Tolkien technology essay that framed our close

* And Ben Thompson on Invest Like the Best from two weeks back, if you missed it, since this whole episode was downstream of his “enough money” question

* Canada’s Aug 25 countermeasures release and the Sept 8 counter-tariff product list

* Meta’s $16.7B settlement — CNBC plus the Al Jazeera explainer confirming your on-air detail: $5.3B (30%) is contingent on TikTok and YouTube adopting matching guardrails

* Bessent’s “two submarines” clip — Mediaite and CTV Edmonton’s hometown angle

* The Norton late-capitalism book from Mel’s notes

Welcome back everyone.

Market Update📈📉

Start with one number that has nothing to do with AI.

The 30-year Treasury hit 5.31% this week. That’s the most expensive long-term money since 2007. I was in high school the last time capital cost this much. Canada’s long bond is at levels last seen in 2010, and it’s the same story in Germany, France, and Italy. Everywhere you look in the developed world, the cost of borrowing for thirty years is breaking out, and it’s happening at the exact moment the biggest companies on earth need to borrow more than any group of companies has ever borrowed.

You might assume these companies are so rich it doesn’t matter. That’s no longer true. On top of the capex you can see on their income statements, nine large tech companies are carrying roughly $3 trillion in commitments that don’t show up on their balance sheets, most of it tied to AI and much of it longer-dated than people assume. Notably, on the calls this quarter, the CEOs themselves stopped comparing this moment to the dot-com era.

They compared it to the railroads.

Ben Thompson was on Invest Like the Best this week and said a thing I haven’t been able to stop thinking about since. Everyone argues about whether we’ll have enough compute or enough electricity.

He thinks the harder question is simpler: are we going to have enough money?

For three years the market has treated capital as infinite and chips, power, and memory as the bottlenecks. Each of those has been a bottleneck. But watch the walk down the capital ladder. First the buildout ran on big tech’s cash flows. Then it moved to debt and off-balance-sheet structures. Now Nvidia is partnering with Apollo, Brookfield, Blackstone, and KKR on a $500 billion vehicle to reach into pension funds and insurance floats, the deepest and most conservative pools of capital on the planet. To get that money, the private equity shops have to underwrite data centers like real assets: how long do they run, what cash do they throw off, what’s the honest depreciation schedule. And Nvidia effectively had to underwrite its own CUDA moat to give them the confidence to do it.

Which brings us back to the railroads… In the 1870s the world literally ran out of money mid-buildout. Nearly every railroad went bankrupt, many more than once. BUT, the trains kept running. The railroads opened the West and powered the economy for a century and a half. The technology worked and the investors got wiped, at the same time. The great returns on the rails didn’t really show up until Buffett started buying them a hundred years later. The returns on this buildout need to show up a lot sooner than that, or we’re in for a world of hurt.

So is there real revenue?

Yes, and the numbers are hard to process. Anthropic is running at roughly $65 billion in annualized revenue and is expected to go public within about ten weeks. They believe they exit this year at $100 billion or better, and there are credible investors who think $400 to $500 billion by the end of 2027 is on the table. If the data centers are the railroads, Anthropic and OpenAI are the biggest trains on the track, and the freight is real.

One more distinction: a data center is really two businesses.

* Training, done on the newest chips, is a loss leader. You can’t sell a training run.

* Inference, serving your prompts, is believed to carry margins in the 90s for Anthropic.

The training bill is why the leading labs are barely profitable despite those inference margins. And that explains the strategy underneath the open-source push. The hyperscalers don’t want to spend trillions building infrastructure only to serve two customers who keep all the margin.

So Nvidia, Meta, Google, and Amazon are backing open-weight models, Kimi K3 and Muse Spark among them, that now run three to six months behind the frontier. Their goal is to compete the frontier labs’ margins down and pull that profit back toward the infrastructure layer. That’s a healthier market, and it’s why the picks-and-shovels names have rallied so hard these past few weeks.

Canada, America’s farm team:

Mel’s segment this week deserves your full attention. Her opening argument was blunt:

“Canada does not have the leverage it thinks it does, and the longer we spend beating our chest, the longer we delay fixing our own house.”

Trevor Tombe’s piece in The Hub made the rounds as a good-news jobs story, complete with a victory lap tweet from the Prime Minister. However, if you read the actual article the employment gain was within the survey’s margin of error, and Tombe’s real conclusion was that Canada is returning to normal growth “along a permanently lower path.” Our peer countries grew about 1.7% while we contracted. Mexico, heavily exposed to the same US trade risk, outgrew us. So did the EU. So did post-Brexit Britain. With commodity prices this friendly, that’s not bad luck; that’s just horrific policy choices coming to bear.

Canada has become the farm team for the American economy. There are now 517 US-based companies with Canadian founders that have together raised $414 billion. Fifty-six of them are worth more than a billion dollars each. That’s roughly the size of our entire federal budget, built by people we educated and then lost. And it’s not an accident. The Americans built a conveyor belt for our talent: the Thiel Fellowship and 1517 Fund for high schoolers, Neo Scholars and Emergent Ventures and YC’s apply-and-defer for students, the Founders Fund residency and SPC for new grads, then Y Combinator and the a16z speedrun. Each rung lifts a kid from Waterloo or the U of A up and out, funds them, houses them, and celebrates them. We spend twenty-five years raising a great entrepreneur, and then they pay taxes in Delaware, California or Texas.

Part of this is culture… We have no better country on earth in which to be average, and I mean that as a genuine compliment. But we treat the top one percent as the enemy, and they are listening.

Mel put it well: “nobody’s asking you to bring out the violins for millionaires, but a good start would be to get off their necks.”

The Fraser Institute’s new study says the average Canadian family now spends 42% of its income on taxes, more than housing, food, and clothing combined, and the tax bill has grown faster since 1961 than any of them. If we want the infrastructure, the services, and the rising standard of living, we need the people who pay for it to want to stay. Elbows up is a privilege. It means you’re already post-economic. The people still trying to build don’t have that luxury.

The new status symbols

We closed on a lighter thread that I think is a real investing signal. The perfectly curated life is dead as a status marker. When everyone’s kitchen is Instagram-ready and anyone can buy the bag, its no longer signal... That has become a base case.

We should watch what’s replacing it: refinished shoes instead of new ones, thrifted furniture, and increasingly, not posting at all. Mel’s list of the new status goods was better than mine: muscle you visibly worked for, clear eyes and real rest, the ability to sit still with a book for an hour, a family, and the quiet flex of not needing a job. (agree to disagree on this one)

My wife and I have been watching Cape Fear on Apple, the Scorsese production with Javier Bardem and Amy Adams. Best show I’ve seen in a year. But the moment that stuck with me was Adrianna saying she loved how Amy Adams looks because she looks real. Six months ago we watched a Nicole Kidman movie and her comment was that nothing in her face moves.

When the tastemakers start prizing real over done, that eventually shows up in earnings, the same way new entrants showed up in Lululemon’s and Nike’s numbers. Hollywood is still how America exports its culture, and culture is where consumer trends are born. Keep an eye on the aesthetics-procedure complex. The volumes may quietly tell the story first.

New episodes every two weeks. Keep learning, keep earning, and keep investing.

Podcast & YouTube Recommendations🎙

* Greed is good?

* My two favorite tech thinkers colab on Invest Like The Best:

* Gear Aquisition Syndrome

Welcome back everyone.

Two weeks ago I wrote that Thursday was a warning, not the flush. The flush coincidentally came with a face and a situationally ironic name.

Leopold Aschenbrenner and his Situational Awareness fund.

Market Update📈📉

The AI trade compressed for five straight weeks, names down 40, 50, 60 percent from their highs in a straight line. Meta was down 10 percent on the year at the lows, Microsoft down 20, Amazon and Google flat while the Nasdaq held up 13. The proximate cause was sitting in plain sight: the Nasdaq topped the day of Kevin Warsh’s first FOMC meeting. Long-term inflation expectations barely moved, but real rates, the way markets price long-duration assets, jumped about fifty basis points. Warsh is a balance-sheet hawk. He wants the Fed out of the mortgage and long-Treasury market, less forward guidance, and real price discovery. The longest-duration trade in the market, the AI buildout, paid the toll first. Meanwhile, for the first time in a year, the index’s built-in AI hedge failed: the capex payers got punished and the capex receivers didn’t rally enough to catch the tape.

Then came the capitulation event. Situational Awareness, one of the most followed AI-concentrated funds in the world, blew up. Roughly $46 billion of equity levered about three times, north of $125 billion of gross exposure, concentrated in exactly the photonics, memory, and GPU capex-receiver names that had been cut in half. When your book drops 50 percent on three turns of leverage, the banks don’t wait. Ken Griffin flew back from Monaco for a day, bought the public equity book for somewhere between $20 and $30 billion depending on who’s reporting, and flew back to his vacation. One of the largest hedge fund bailouts in history, done between dinner reservations.

Forced selling from a leveraged whale is the kind of thing that marks bottoms, and so far it has. The capex receivers have rebounded hard. Microsoft went from down 20 on the year to up five. Meta recovered. Amazon just became the fifth $3 trillion company on earth after an incredible AWS quarter. And in the great rotation tell, Apple, which spent a year as the safe haven for people who wanted tech without AI, gave the crown of world’s most valuable company back to Nvidia.

Here’s why I think the rebound has fundamentals under it rather than just short covering. GPU rental prices have doubled in the past year.

If these were bad investments being depreciated into a glut, the rental market would be telling you. Instead the infrastructure is in such demand that the hyperscalers are demonstrably under-earning on the assets they already own. Microsoft and Amazon’s quarters showed real return on the cloud investment, and the frontier labs are doubling revenue every six months. We’ve never watched businesses go from startup to a hundred billion of annualized revenue this fast. The market is having a genuinely hard time valuing it, and the next test comes when OpenAI and Anthropic hit the public markets, likely around year-end.

The market is once again believing the spend is justified.

Volatile bottoms don’t mean the volatility is over. But the burden of proof has flipped back onto the bears.

No influencers allowed

A basket store on Nantucket hung a sign that says “No influencers,” and Mel and I spent longer on it than we planned, because it’s secretly a markets story. Scarcity is the oldest luxury strategy there is. Rolex runs it, Hermes runs it, Porsche runs it, and now an expensive basket shop is running it against the very people who normally manufacture demand. When access gets democratized, exclusivity becomes the product. You’re seeing the same physics in collectibles, where a PSA-10 grade takes a $500 Charizard to $45,000. Authentication, uniqueness, and real-life experience are becoming the scarce assets in an economy of infinite reproduction. It’s why I keep a long-term bucket in portfolios for real-world experiences, the Four Seasons and airline-travel end of the consumer. It’ll never be as sexy as the picks-and-shovels trade. It also never has a Situational Awareness moment.

The West, and the rest

Mel’s framework this episode is the one I’d ask you to sit with. The left/right spectrum you grew up with, mostly an economic argument about markets versus intervention, is dead as a map. The axis that actually predicts where politicians land in 2026 is this: do you believe Western civilization is worth defending and improving, or do you believe it needs to be dismantled for something presumed better? Data centers, pipelines, billionaires, even the Billy Bishop airport fight all sort cleanly along that line, and almost none of it sorts along the old one.

Her sharpest point was about the scarcity loop. An abundance mindset says the pie grows; when someone else wins, that’s evidence you can too. What a chunk of our politics runs on instead is a grift cycle: make people fearful there isn’t enough to go around, pass policies that ensure there isn’t, offer the state as the solution, collect the votes, repeat. After enough cycles, people stop believing the private sector can deliver anything, which makes the state more necessary, which makes the pie smaller. Scarcity is a policy choice. So is the exit.

WestJet, and the party that walked away

The WestJet strike resolved into a tentative deal, but the political tell stuck with me. The Trudeau Liberals spent a decade courting unions. This Liberal government watched 4,400 flight attendants shut down summer travel over unpaid ground work and had essentially nothing to say. Not intervention, not support, just indifference. Mel’s distinction matters here: the old Liberal enthusiasm was mostly for public sector unions, which grow the state, and about 30 percent of employed Canadians now belong to a union of some kind. When public sector compensation stops involving trade-offs entirely, you get what she called a hostage situation rather than a negotiation, and the public’s sympathy erodes. Ask the postal workers. The deeper irony is the one we keep hammering: the biggest thing happening for blue-collar wages in this country right now is the data-center and megaproject buildout, and the party that historically claimed those workers is ambivalent about both the projects and, apparently, the picket lines. The workers have moved. The parties haven’t caught up.

Alberta, it’s time to build 🏗️

This week I published the longest thing I’ve written this year, and it’s the piece I most want you to read and share. It makes the case that Alberta is about to run the last great cycle of the old energy economy and the first great cycle of the new one, simultaneously.

The short version. On July 2, the Prime Minister and the Premier stood together and announced a million-barrel pipeline to the coast. Six days later Meta committed $13 billion to a one-gigawatt data center in Sturgeon County, its largest anywhere outside the United States. The pipeline is the biggest version of Alberta’s hundred-year-old model: pull energy out of the ground and ship it somewhere else to become valuable. The data center is the first serious version of a new model, where the gas stays home and gets refined into the most valuable commodity on earth right now. A data center is a token refinery. It converts BTUs into intelligence and sells it globally at software margins.

The piece walks through the gas math (every gigawatt of gas-fired data center load is roughly 0.16 bcf a day of new, flat, creditworthy, in-basin demand, the structurally better customer AECO has waited decades for), why one campus becomes five (hyperscalers copy each other’s site audits, and Alberta just passed Meta’s), the TSMC-Phoenix precedent ($12 billion became $265 billion and forty follow-on projects), the Strathcona County tax lesson (industrial assessment is why Sherwood Park pays 30 to 45 percent less property tax than comparable cities, while Edmonton households just absorbed a 29 percent increase in four years), and who actually gets paid: the electricians, pipefitters, module yards, brokers, lawyers, and accountants of Edmonton, St. Albert, Sherwood Park, and Fort Saskatchewan. The commodity gets the headlines. The infrastructure gets the early returns. The services get the multiples.

It closes on the Abundance argument, because central Alberta is now the closest thing North America has to a live test of Klein and Thompson’s thesis that scarcity is self-inflicted. Greenlight went from announcement to a $4.6 billion final investment decision with a four-year build. In most US markets, the interconnection queue alone takes longer. Our advantage isn’t capital. It’s the willingness to say yes quickly, and it’s measured in one number: months from application to token.

Share it with your MLA, your favourite councillor, and the smartest skeptic you know.

Podcast & YouTube Recommendations🎙

* Four of the sharpest minds in geopolitics predicting how the world ends

* A must listen from RenMac and their special guest - Stephen Moran

* One of the best interviews on Consumer Trends with the CEO of Doordash Tony Xu

Welcome back everyone.

Market Update📈📉

For most of the past year the index has carried an almost built-in AI hedge. When hyperscalers raised capex, the companies receiving the money rallied, because the spending became their revenue. When hyperscalers hinted at spending less, their own stocks rallied on better free cash flow and more room for buybacks. One side could get hurt, but the other side caught enough of the money to stop the index from falling far. That arrangement made SPX remarkably hard to push down, and it’s why all the violence underneath kept disappearing at the index level.

Thursday was the first time that hedge failed.

Alphabet reported good operating numbers and raised 2026 capex guidance again, to $195-205B. Free cash flow came in at negative $5.9B for the quarter, and the stock fell 7%. The payer got punished, and the recipients didn’t rally enough to replace it. The old reflex of buying anything that might receive a dollar of hyperscaler capex was nowhere near as reliable. The largest index weights fell, and the hedge on the other side was too narrow to hold the whole thing up. Nasdaq dropped 2% while Brent pushed through $100 and the 10-year traded above 4.7%.

Alphabet wasn’t the first warning, just the first one big enough to matter at the index level. Over the prior two weeks, ASML, TSM, Texas Instruments, MaxLinear, Intel and BE Semiconductor all traded lower despite beats or raised guidance. Intel reported a blowout quarter Thursday night and the market yawned. Markets don’t top on bad news. They top on good news, when people stop paying attention to it.

Friday bought time without fixing anything. Reports of possible US-Iran talks knocked Brent back down, yields eased, and SPX finished flat. Nasdaq still closed down 1%, so the market didn’t rediscover its love for AI. Macro pressure simply eased. Even the support looked defensive rather than bullish: Apple outperformed by nearly 5%, and names like AT&T had their best week relative to the market in decades. My read: Thursday was a warning, not the flush itself.

Where the money is going

Money is never destroyed in a momentum unwind. It flows somewhere else.

Right now the tell is real yields, which are breaking out globally: the US, France, the UK, Bunds, JGBs. The price of money is rising, and investors are rotating out of long duration, the stocks priced to perfection and to grow forever, into financials, energy, REITs, insurance, and select healthcare.

It rhymes with the pre-GFC regime, when fundamental compensation for risk actually started to matter again.

The macro backdrop explains why.

Kevin Warsh’s Fed is facing three inflation drivers at once: relentless AI-driven demand for equipment and electricity, a reignited tariff regime, and a re-emerging energy shock out of Iran, with shipping lanes blocked, the SPR largely depleted, and Brent above $100. I think the market is underappreciating the risk of a hike as soon as this week. I don’t expect one, because Warsh still appears to view the energy shock as transitory. Which, again, is a dangerous word.

Semis are deeply oversold and violent bounces are coming. The momentum playbook says you sell into them.

Dimon is playing a different game than you are…

The headline of the week was “Jamie Dimon says don’t buy stocks,” and most investors read that as sell everything. Listen to the actual interview and he said “I don’t like prices today,” which is not the same statement. CEOs steward capital on time horizons that have nothing to do with yours. The last time Dimon made headlines like this was 2022, before the Fed hiked 500 basis points, and he wasn’t calling a top then either. He was describing risks to a bank that will outlive him. Meanwhile JPMorgan just broke out of an eighteen-month base to all-time highs. Your portfolio doesn’t care about headlines, and neither does he. Follow your own process.

Data centers are the new pipelines

At home, Canada has said almost nothing on AI policy while the two superpowers set the rules. Mel’s reframe on the podcast stuck with me: Carney isn’t silent by accident. He has no political incentive to touch it. He doesn’t win that argument with investors, and he doesn’t win it with the public. So the conversation doesn’t happen, and the comment sections under every Edmonton data-center headline supply their own answers. I’m having the same reset conversation at men’s league golf and in client meetings: intelligent, well-meaning people convinced these buildings are the devil. We did the same thing with pipelines for 35 years. Bad communication became bad policy, and the economy paid for it.

The tariff file makes the stakes plain. CUSMA didn’t get renewed, so the agreement now faces annual reviews instead. Trump’s threatened 50% tariff on certain Canadian goods hits BC (13.7% of its US exports), Quebec (10.8%) and Ontario (9%) hardest, per Trevor Tombe’s estimates. Alberta: about 1%. The likely next step is pressure on Alberta to tariff energy exports, the only lever that actually bothers Washington. Expect that to go over poorly in a province that has spent 25 years giving more to Confederation than it gets back.

The saving grace: natural gas

Here’s the positive note I’ll close on. US electricity demand is rising far faster than supply, natural gas production can’t keep up, and the political will to build nuclear at scale isn’t there. Sometime before 2030, I think the Americans won’t ask for Canadian natural gas. They’ll demand it, pipeline and all. Alberta has quietly built the blueprint: infrastructure ready for data centers, abundant gas, and a likely surplus this year with oil trading nowhere near the $56 the province modeled. The US spent 20 years as an energy exporter. That era is ending, and we’re the obvious solution next door.

New episodes every two weeks. Keep learning, keep earning, and keep investing.

Podcast & YouTube Recommendations🎙

* One of my favorite podcast episodes from ILTB:

* We’re not the biggest All-In fans but i really liked this perspective from Friedberg

* A great listen if you’re interested in better understanding datacenter + Frontier Lab economics

Alberta is so back.

Stampede wrapped, and while the rest of us were nursing a ten-day social hangover, Alberta quietly had the best economic week it’s had in a long time.

Four announcements, potentially more than a hundred billion dollars of investment if they all land.

For the first time in my adult life, this province is about to build the entire stack: pull energy out of the ground, use it to make something the world wants, sell that, and tax it right here.

Let me start with the one that matters most, because it’s the one people are complaining about.

The data center comes to the energy

On Wednesday last week, Meta announced a thirteen-billion-dollar data center in Sturgeon County, about an hour northeast of Edmonton. Its first in Canada, its thirty-third globally, and its largest anywhere outside the United States.

Put one gigawatt in perspective. The entire city of Edmonton, every house, every hospital, every rink, runs on about 1.4 gigawatts. This one campus, roughly the size of 33 CFL fields, will pull close to two-thirds of what the whole city next door uses.

The incoming script:

Community meetings… angry residents…

“Big Tech is going to spike my power bill.”

Giant new load shows up, prices go up, Econ 101. Except the data mostly says the opposite, and it’s worth understanding why.

A power grid is a fixed-cost machine. You pay for the transmission lines, the substations, the transformers whether electrons flow through them or not. Across a full day, most grids run at maybe 55 to 60 percent of capacity. A data center runs at 90 to 95 percent, around the clock, every day. It’s the customer who buys a seat on every single flight instead of leaving half the plane empty. Fill those seats and the cost per seat falls for everyone.

EPRI, about the most sober outfit in the industry, found that for every 10 percent bump in data center capacity in a state, residential rates fell about 0.4 percent, and in the states that grew capacity fastest, bills came down around 6 percent.

Then there’s the fiscal side, which is the fun part. Loudoun County, Virginia, is the densest cluster of data centers on Earth. They sit on about 4 percent of the commercial land and throw off 38 percent of the county’s general-fund revenue. Loudoun has cut its residential property tax rate every year for a decade, from $1.145 down to $0.805 per hundred dollars of value. A 2026 analysis found that without the data centers, that rate would nearly double, roughly $5,800 a year for a typical homeowner. If you live in the Edmonton region, you already understand this intuitively.

It’s the same reason Strathcona County homeowners pay less: the refineries carry the bill. Server racks are about to do the same thing for Sturgeon County.

Here’s what Mel rightly kept flagging.

The legitimate criticism of US data centers is cost allocation. When a hyperscaler plugs into an existing grid and the utility builds new generation to serve it and socializes that cost across everyone’s bill, ratepayers really can get hurt before the benefits show up.

Alberta solved that.

The province lets data centers generate their own power, and it structured the tax so that if you draw from the grid you pay more, and if you bring your own generation you pay less. Meta isn’t showing up to drink from the existing pool. Pembina, Morgan Stanley Infrastructure Partners, and Kineticor are building the Greenlight Electricity Centre right next door: 932 megawatts of new gas-fired power, $4.6 billion, permitted to double. New load, new supply, financed by the customer.

The subsidy fight poisoning these projects in Ohio and Georgia basically doesn’t exist here.

And the talent piece is the part outsiders miss.

I’ve heard genuine horror stories about how hard these things are to build in the US, because the electrical systems are brutally complex and the skilled trades are stretched thin. Albertans are unusually good at exactly this. Drive north to Grande Prairie, Fort McMurray, Cold Lake, or walk the refineries around Edmonton, and you’re looking at some of the most complex energy systems on the planet, built and run by people who do this for a living.

Meta figured out it can build a state-of-the-art closed-loop, liquid-cooled facility here, with more consistent power, for something like half what it costs in Texas. If they can do it for thirteen to twenty billion instead of double that, I can genuinely see a path where they build ten of these.

So how do you actually make money on it?

A friend asked me the obvious question: how do we invest in this?

And the intuitive answer, “just buy Meta,” is probably the wrong one locally.

Everybody races to the hyperscaler at the top of the stack.

The value you can actually reach shows up at the bottom. The electrical contractors who wire these things. The land developers. And above all, the power. Meta and the next nine tenants can’t always build their own generation, which means power purchase agreements, which means the companies that can sell them electricity, whether that’s natural gas, wind, or solar, just found a twenty-year, investment-grade customer.

Watch the merchant power names and the midstream players like Pembina, who just found a growth vertical that has nothing to do with pipeline-egress politics. And it’s not a coincidence any of this landed in the lowest-taxed province in the country.

The pipelines:

Two pipelines hit the table this week.

The big one is the West Coast oil line. On July 2 Alberta submitted it to the federal Major Projects Office, and the news that mattered was a private proponent: Pembina, taking a 10 percent economic interest with the option to go higher.

It’s a million-barrel-a-day line from the Edmonton area to the BC coast, and it lays on the Trans Mountain corridor for most of its length(85-90%).

Notably, it’s a southern route, not the shorter northern one that would sit closer to Asia and take bigger tankers, because the northern tanker ban is not going away. Carney has said as much. We should hear by October 1 whether Ottawa lists it as a project of national interest, with shovels possibly in the ground by late 2027. All of it plays out against an Alberta referendum on October 19.

The second line, a proposed Alberta-to-Ontario route announced July 6, is still at the ideation stage. I wouldn’t put a number on its odds. It reads less like a construction plan and more like a signal: provinces cooperating, “build in Canada, stay in Canada,” which is a pointed message from a UCP government that critics love to call separatist.

Talk to politicians and the public and everything sounds fantastic. Talk to the people who actually allocate capital to mega-projects, and the needle has barely moved.

Because the fundamentals haven’t changed. The tanker ban is still there. The C-69 consultation uncertainty is still there. The carbon tax is still there. What changed is that we have a Prime Minister personally willing to champion a project. And that’s the risk hiding inside the good news. When one individual, through his own leadership, is making the exceptions and pushing the project through, you still don’t have a framework. You have a man. If Mark Carney weren’t Prime Minister, Canada would not be in a structurally better position than it was under Trudeau, because the institutions that would give an allocator certainty still don’t exist. It rhymes, uncomfortably, with investing in a jurisdiction where the rules depend on who’s in the chair rather than what’s written down.

It’s good news. It’s real.

But you can’t run a resource economy on catching a few project interests once a decade because the right person happened to be in office. The work of locking in durable investment certainty is still ahead of us.

The tape, and the geopolitics underneath it

The market gave us a perfect illustration of where AI is in its cycle. Meta launched “Meta Compute,” selling its excess AI capacity against AWS, Azure, and Google, and the stock popped more than five percent. But the same headline gutted the picks-and-shovels names underneath it: Coherent, Wolfspeed, Modine, Teradyne, Micron all red. After two years of pricing AI infrastructure as permanently scarce, the biggest buyer just admitted it has enough spare capacity to become a seller. That cracks the scarcity premium. You saw the same rotation when a report that DeepSeek is building its own inference chip knocked Nvidia while AMD ripped almost eight percent. Money isn’t leaving AI. It’s changing seats inside it. And it’s all sitting on a hawkish Warsh-led Fed with inflation at a three-year high and the ten-year at 4.55%, which keeps punishing the highest-duration corner of the market.

That inflation is partly energy, and energy is where the geopolitics bites. The US-Iran ceasefire collapsed this week after Iran struck commercial tankers in the Strait of Hormuz, and that risk premium is what’s keeping oil in the seventies even on soft demand. Mel’s frame on the wider picture is the right one: two things can be true. The US under Trump is an unpredictable and at times unreliable partner, and we still don’t actually have the option of walking away from that relationship. CUSMA certainty isn’t there, and part of why is a foreign policy that seems built around avoiding Washington at any cost. Carney in Saudi Arabia this week, the first Canadian PM to visit since 2000, signing thirteen agreements worth about a billion dollars, is the same instinct. Not objectively wrong. The Saudis want minerals and we have them, plus the oil. But if the motivation is “anybody but the US,” that’s not obviously a long-term win for Canada. Diversifying because you’re building strength is smart. Diversifying to spite your biggest customer is something else.

Pull it together and the theme of the whole show is this. For twenty years our problem was that we produced energy nobody would let us ship. Now the demand is coming to us, the capital is coming to us, and even the Saudis are coming to us. The only thing still standing in our way is whether we can actually build, and whether we’ll turn one good week into a framework instead of a fluke.

See you in two weeks.

West Coast oil pipeline (submitted to the Major Projects Office, July 2)

* Alberta.ca — West Coast Oil Pipeline

* Major Projects Office (Canada.ca) — West Coast Oil Pipeline project page

* CBC — Alberta pitches southern route, price tag of $35B or more

* BOE Report — West coast pipeline, private partner, oilsands production (Key facts: ~1 MMb/d, Edmonton region → southern BC coast; southern route (tanker-ban avoidance); $35.2B–$43.7B; Pembina 10% economic interest, option to +10%; Gazette listing decision by Oct 1, 2026; FID Q1’28–Q4’29; shovels as early as 2027.)

East–West pipeline (Alberta → Ontario, July 6)

* Alberta.ca — East-west pipeline announcement (Ideation stage; framed as cooperative federalism / “build in Canada.”)

Policy backdrop (the “framework vs. one man” point)

* Canada.ca — Building a stronger economy: sovereignty, exports, emissions (One Canadian Economy) (Context for Bill C-5 / Major Projects Office; the tanker ban (C-48) and C-69 uncertainty that still constrain investor certainty.)

⚡ The data-center-economics case (for the monologue)

* EPRI research — Data centers pushed average U.S. residential rates down, 2015–2024 (+10% DC capacity → −0.4% residential rates; fastest-growth states → ~−6%.)

* Loudoun County, VA (official) — Data Centers: Tax Revenues and the County Budget

* Progress Chamber — Data centers cut Loudoun property taxes ~$5,800/yr (DCs on ~4% of commercial land → 38% of general-fund revenue; rate cut from $1.145 to $0.805 over a decade.)

Two weeks ago I sat in this chair and told you to stop arguing about whether AI is a bubble and watch two things instead: the bond market, and the war in the Middle East. We got our answer for both.

So let’s grade the tape - starting with the one that should humble every expert who opened their mouth this spring, including me.

The humble pie nobody’s eating

For four months we had a shooting war with Iran. For 90% of it, the Strait of Hormuz was effectively closed. A fifth of the world’s oil moves through that strait, and in the eyes of basically every energy analyst alive it’s the single most important piece of infrastructure in the global economy…

And every single person in the energy market had the identical opinion. Oil’s going to a hundred and twenty, maybe a hundred and fifty. Inflation shocks everything. Batten down the hatches, the Nasdaq sells off twenty-five percent.

It was unanimous.

None of it happened.

This past week tankers started transiting the strait again, Iran is winding down the closure, they’ve signed onto a memorandum of understanding — and instead of oil staying high, it collapsed back to roughly five to seven percent above where it sat before the first missile flew. WTI is down in the seventies. After all of that. A war, a closed chokepoint, a quarter of the year with twenty-plus percent of global energy trade taken offline — and crude is basically flat on the year.

Nobody is doing the postmortem. Nobody is eating the pie.

So let me eat mine, because being wrong in public is the only honest way to get better. I was sure energy stayed higher for longer. I was wrong.

Why?

Three reasons:

One — China drew down its own inventories instead of panic-buying. They had far more in reserve than anyone modeled, which tells you how little we actually know about how other countries operate.

Two — China could flex its demand down. Their build-out of EVs, electrified transit, and solar has quietly made oil far more elastic than the textbooks assume; the idea that price has to stay high for long is now genuinely in question.

Three — dark transits. There is a whole shadow logistics system moving barrels outside the view of the people who price this stuff, and it’s bigger than anyone admits. Put those three together and you have the only real explanation for why every commentator got it wrong.

The part that matters for your portfolio: a meaningful chunk of that Chinese demand reduction in transport is going to be permanent. That’s not a wartime blip you get back. That’s a structural dent in the long oil story. Separate Alberta’s very real cyclical windfall from the secular demand picture — they are not the same trade.

Warsh changing the feds math while everyone watched the war

Kevin Warsh ran his first meeting as Fed chair, and he is a more hawkish animal than Powell — which is gloriously ironic, because he’s exactly who Trump pushed for. Two things came out of that meeting.

First, he told the market he doesn’t love dot plots and isn’t going to telegraph the next three moves. Markets hate uncertainty, and they let him know it.

Second, and bigger: he made clear he’s not cutting this year. The market repriced instantly — from pricing a possible December cut to roughly two-thirds odds of a rate hike in September. That is a massive move in the discount rate sitting under every equity attached to the USD.

That happened Thursday. Everyone went on vacation. We came back to two red days, the Nasdaq and S&P each down one to two percent, and a market that had been up ten, twelve, thirteen percent on the year handed back a third of it to sit around seven or eight.

A Canadian wrinkle I can’t ignore. A Fed that’s cleared out 2026 cuts and is flirting with hikes bleeds straight into us. I was poking through realtor.ca data for fun, and inventory is climbing across the country — including here in Alberta, in Calgary and Edmonton, the fastest-growing part of the country. This housing market cannot handle rate increases. Ontario and BC certainly can’t. Which brings us to the thing that genuinely annoyed me this week.

The condo bailout, and the courage we don’t have

On June 18, Mark Carney and BC Premier David Eby cut a deal. There are thousands of finished condos in BC sitting empty — over two thousand units — as demand cooled with slower population growth, and the developers who built them are staring at insolvency. So Ottawa and the province are co-funding a package north of three billion dollars, roughly split between them, to buy those condos at their list prices and convert them into affordable housing.

Mel’s take was sharp, and I think it’s the right starting point. Why are we intervening at all?

When demand falls, price is supposed to fall with it. That’s the whole mechanism. Let the market clear these units at a lower number, and if government wants affordable housing, take the money and build it.

Instead we’re rescuing a sector at list price. Her deeper point cut harder: this is preservation, not progress. An outsized share of Canada’s economy is real estate and everything stapled to it, the wealthy and the older cohort are heavily invested in keeping those values up, and protecting them is politically expedient. We have a brilliant economist running the country who surely understands supply and demand — but apparently those laws get suspended for the one golden-goose industry nobody wants to touch. She called it a failure of political courage, and the line that stuck with me: be the one who decides, or stop pretending you can’t.

I pushed back, because I’m more sympathetic to the occasional bailout than she is, and here’s why. Force a fire sale and you don’t just punish wealthy developers — you wipe out the contractors, electricians, drywallers, and painters underneath them who only get paid if these projects get made whole. Let the majors fail and you send a signal that Canada isn’t investable, that cap rates need to double, and that contagion can run from condos into the entire commercial and industrial real estate complex. Central bankers — and Carney is still one at heart — look at this through 2008 and 2009 and refuse to risk a run. I understand that instinct.

But Mel’s rebuttal landed. If we care this much about the jobs in this sector, why don’t we extend that same care to the natural resources workers we’ve ignored for a decade — workers who, conveniently, don’t swing ridings in the GTA and Vancouver?

A central banker’s job is to read the room and pull the levers available. A leader’s job is to change the board. Carney is one of the few people on earth who could actually change the board, and he’s choosing to defend the old one.

Europe 2031, and the choice in front of Canada

Europe just moved to block US big tech — Microsoft, Amazon, the hyperscalers — from selling software into government agencies, a de facto rejection of American influence dressed up as driving homegrown innovation. And it’s framed perfectly by a piece making the rounds called Europe 2031.

It’s a fiction story of two characters living out a a five-year scenario. Its written by serious EU AI-policy people as a warning.

The setup: January 2025, DeepSeek’s R1 drops, Nvidia falls twenty-odd percent, memory names down thirty, photonics down forty to fifty, and Europe takes a victory lap — see, AI is solved and cheap, we don’t need America’s hundred-billion-dollar data centers, we’ll buy intelligence on the discount rack.

Then the gap doesn’t close. It explodes.

By the end of the story the US runs seventeen times Europe’s compute and hosts eighty percent of the world’s AI capacity, because Europe bet that compute was a commodity and that “sovereignty” would make it stronger. It made it weaker. The instinct to reject the technology became the cause of its irrelevance.

The piece names Canada explicitly, and this is the part that scared me a little. It lists the middle powers who actually hold leverage, the ones sitting on real bottlenecks. The Netherlands has ASML, the only company on earth that builds the ~$350M EUV machines that print every advanced chip; no EUV, no GPU, no CPU, no memory.

Germany has robotics. France and Norway have AI talent. The UK has finance. Japan controls the chemicals the whole data-center build-out depends on. South Korea has the memory - Hynix, Samsung, Micron’s footprint. And Canada has the energy, the critical minerals, world-class AI labs like Amii right here in Edmonton, and the geographic luck of sitting directly on top of the largest consumption engine on the planet.

We do not have to become Europe. We could be something better — a blend of what makes the US and Europe each work. But the warning in the piece is that we share Europe’s exact disease: we’d rather study a hard problem to death than pour the concrete. The investment lesson is blunt: if you allocate capital into jurisdictions that reject progress on principle, your return on invested capital will be atrocious.

Policy is not background noise. It’s the thing that decides whether you get ten-percent compounding or a low-growth trap. Mel framed the whole posture as a luxury belief - like the person who’s gone to the gym for ten years and decides they’re just naturally healthy, forgetting it was the daily reps.

And right on cue: a hundred billion dollars of proof

If you want evidence the gap in Europe 2031 is real and not science fiction, look at what dropped this weekend. The two leading American AI labs, OpenAI and Anthropic, now have combined annualized revenue north of $100 billion. Sit with that.

The Anthropic line alone is the wildest chart in business right now: roughly $1 billion of ARR in January 2025 to about $30 billion by April 2026, when it passed OpenAI in revenue for the first time - and climbing into the high-forties since.

OpenAI is doing about two billion dollars a month and filed confidentially to go public earlier in June. Eighteen months ago most people couldn’t name either company. Today they jointly clear nine figures of recurring revenue.

That is the whole Europe 2031 thesis in a single data point. While one continent debates whether to permit American software, the American model layer is compounding revenue faster than any software business in history. This is what “the cat is out of the box” looks like on an income statement. AI is here, it’s reshaping the economy in good ways and bad, and the value is pooling exactly where the compute, the capital, and the willingness to build already live. If you’re a Canadian allocator and that doesn’t reframe how you think about ROIC and where the next decade of returns gets made, I don’t know what will.

SpaceX, and giving Elon his flowers

Which brings me to where I wanted to land: SpaceX. Last episode we talked about the wealth it was about to create. It priced at $135, raised an extra fifteen percent to pull in close to $85.5 billion — an absurd sum — popped nineteen percent on day one, and closed around $163, the level that actually matters off an IPO. It ran to $225 the day Elon briefly became the world’s first trillionaire. Today it trades around $150 to $154. That round-trip, top to bottom, is a pretty honest tell about where this market is right now.

But the number I keep coming back to isn’t the price. It’s the people. I don’t have enough fingers to count the individuals I personally know who’ve been made wealthy by that one company — and nothing remotely like that wealth-creation engine exists in our country.

Now, longtime listeners know I was the original Elon hater. I despised the way he’d stretch the truth on earnings calls. Brock asked Mel if I’d forgotten that. I haven’t forgotten — I just understand it differently now. He weaponizes belief. He convinces brilliant people that something outlandish is achievable, and in doing so makes it more likely to actually happen. That’s how you get a SpaceX prospectus claiming a $28 trillion total addressable market and nobody laughs. Pair that with a genuine ability to take on big, hairy engineering problems — build the first successful new American car company in a century, land rockets and reuse them, fly a constellation that beams down cheap internet, all under the banner of colonizing Mars — and you get something rare.

Here’s the real gift. For the first twelve years of my career, all we did was optimize bits — myopic little efficiency tweaks to software after the iPhone ate the world. We forgot how to build atoms. I read Where Is My Flying Car? and that’s the whole indictment: we stopped building real things. Elon dragged the culture back to building. And now stocks that haven’t worked since the 1990s are leading the market, because the whole economy remembered how to attempt big, crazy, physical things again. There’s a humanoid robot in my five-to-ten-year future folding my laundry. That should be astonishing, and it’s downstream of one guy convincing everyone the impossible was just a schedule.

So I wanted to give him his flowers. Tune out the political noise, watch what he builds, and we’d all be better off. Mel put the bow on it: being a hater is lame. Have your debates, fine — but if you’ve never tried to build anything, even a community garden, you don’t get a seat in the conversation about the people who do. The next phase of this economy needs people with skin in the game. Get yours.

Positive-sum mentality. We’d all be better off with more of it.

See you in two weeks - when we finally take on the social media question.

Podcast & YouTube Recommendations🎙

* Did China Match Claude? - Not really.

* Maverick Capital on Opportunities in AI

* The Bond King Talks About the New Fed

For three years the only question anyone asked me was whether AI is a bubble. Is it real, who wins, what inning are we in. I’ve been saying inning three or four the whole time, and I’ll say it again. But it’s the wrong question, and last Friday’s tape is the reason why.

Here’s the better one. Where is the money actually flowing inside all of this, and what happens to every high-flying name if the bond market becomes the story?

The jobs report nobody wanted

Last Friday the US printed a jobs report that was too good. Stronger hiring, plus upward revisions to the prior two months that pushed the three-month average to its highest since early 2024. Job openings reversed from shrinking to growing.

That’s a hot economy.

For two and a half years the market has priced one thing: rate cuts.

Money was going to get cheaper. Friday said the opposite. When the cost of money goes up, the multiple you’ll pay for long-duration, high-beta, fast-money stuff comes down, and that’s exactly what AI momentum names are.

The compensation you demand for risk has to rise with the risk-free rate. So the two-year Treasury broke out to a fresh twelve-month high, the ten-year sits at 4.57%, and stocks sold off hard. Down two and a half percent on Friday, with more bleeding into this week as we await todays inflation data.

All of it is happening with a Fed in the middle of a leadership change, a President running a war in the Middle East, an inflation mandate he hasn’t delivered on, and a midterm cycle bearing down. That’s a lot of headwind in one frame.

Sit with how stark the shift is. The market spent two years rewarding one kind of positioning, and the regime just flipped under everyone’s feet.

There’s a Canadian wrinkle I can’t ignore. If the data stays hot and the Fed can’t cut, the Bank of Canada can’t comfortably cut either, and we are not in the same economic shape. The US is running hot. Europe is choking on input costs. Most of Canada outside the export-heavy West already looks like it’s in a recession. The exception is Alberta, where oil above eighty dollars has handed us the leverage we’ve wanted for a long time. I’ll be honest. As an Albertan, I haven’t had this much fun investing in eighteen months. That’s an uncomfortable thing to say when a lot of the country is hurting, but it’s true, and that divergence is going to define the next year.

The math says don’t buy stocks. The history says buy the right ones.

Now look under the hood. The ten-year yields 4.57%. The S&P’s earnings yield is 4.56%. Those are the same number. On the classic risk-adjusted math, why would you own stocks at all? You’re not being paid to.

This is where everyone needs to understand one piece of research, because it reframes the whole thing. Hendrik Bessembinder studied 64,000 global stocks over thirty years, a window that includes the dot-com bust, the financial crisis, and four recessions. Every bad thing that’s happened in my lifetime is in the data. What he found is that one to two percent of stocks generated all the net wealth created above Treasury bills. The other 98 or 99 percent, added together, netted out to roughly the return on T-bills. Somewhere between 55 and 57 percent of stocks actually underperformed T-bills over three decades.

Bring that down to the S&P 500 and the point gets sharp. Out of five hundred names, maybe ten or twelve do the real work. If you don’t own them, you didn’t just lag. You got wrecked for all the risk you took. And the cruel part is those names are usually the controversial, hard-to-hold ones. The last few years it was easy. Apple, Microsoft, Amazon, Alphabet, Meta, Tesla, Walmart. The next thirty years won’t be the same companies. But I’d bet my house the shape is identical. One or two percent of the names deliver the entire prize. If you won’t invest into where the paradigm is going, you miss it. Full stop.

The IPO supercycle, and the lie about it draining the market

And the paradigm is going public. Over the next six to nine months we get Anthropic, OpenAI, Anduril, and Stripe, the biggest private payment processor on earth. SpaceX prices this week. Google just raised eighty billion in equity anchored by Berkshire. The greatest entrepreneurs of the last fifteen years are all reaching for liquidity at once. So much for “sell in May and go away.” The people who make their living raising capital can’t go anywhere this summer.

The question every client asks me: won’t all these IPOs suck the money out of the market and trigger a correction? It’s the intuitive fear, and it’s wrong.

Here’s why. The largest pools of capital on the planet, the Fidelitys and T. Rowe Prices and Manulifes and Sun Lifes, mostly have no mandate to own private companies. Where they can, it’s capped around 15 percent. They were locked out of Anthropic, locked out of SpaceX, locked out of the best compounders of the decade. Once these names are public, those funds have to allocate. That’s where the bid comes from. The global equity market is north of a hundred trillion dollars. It can absorb four hundred billion in new equity without blinking. If there’s a problem, it’s a short one.

Think about what that liquidity does on the other side. Ask which Canadian city has minted the most millionaires recently and the answer is Ottawa, because of Shopify. The stock went from a sub-billion valuation to the second most valuable company in the country and turned a generation of early employees and investors into a tech hub’s worth of capital. Now run that forward across the zip codes where SpaceX, Anthropic, OpenAI, and Anduril live. That’s the American flywheel. Liquidity creates wealth, wealth funds the next wave of founders, and the cycle compounds.

SpaceX: the $1.8 trillion magic trick

Which brings me to SpaceX, where I’ll disclose up front that my firm, RBC, is one of the underwriters. They’re raising roughly $75 billion at a $1.77 trillion valuation, $135 a share, listing on the Nasdaq this week under SPCX. That raise is triple the size of Alibaba, the previous record. It’s the largest IPO in history, and it came in double-subscribed. Finding the capital was never the problem.

Is $1.8 trillion outrageous? Of course. But read the prospectus and it gets funny. SpaceX claims a $28.5 trillion total addressable market, the largest TAM in human history. And that number is the whole tell about how Musk works.

There are three businesses stapled into one ticker. Rockets, Starlink connectivity, and xAI-plus-X, and yes, X is Twitter. The Twitter business is a dumpster fire. xAI, for all practical purposes, is a dumpster fire. Before Musk merged them in about six months ago, the core was profitable. Fold xAI and X back in and you get a negative-cash-flow company posting a $4.9 billion loss on $18.7 billion of revenue. So why do it? Because rockets and Starlink are only about six to ten percent of that headline TAM. The other ninety-plus percent is “AI,” and the valuation of AI right now is, conveniently, infinity.

Here’s the part where I’ve changed my mind on the man more than once. Musk’s actual superpower isn’t engineering. It’s that he never forgets his first job as a steward of capital is to make his investors money. He bought Twitter for $25 billion when Biden’s regime more or less forced his hand, pulled his closest friends in to fund it, and three years later that albatross gets merged into the hottest IPO ever. His backers get their liquidity through SpaceX. Along the way he built Colossus, the biggest data-center and chip-manufacturing project in the US, and used it to expand his story from a $2 trillion space-and-connectivity TAM to $28.5 trillion. That’s not signaling. That’s the discipline of putting returns first, and he’s the best in the world at it.

Mel’s frame on this is the one I want people to sit with. For fifteen years a chunk of society convinced itself that if we just clicked our heels and thought about capitalism differently, we could rewrite the rules. We can’t. The person who makes the gold makes the rules, and the rules of free markets outlast whatever Overton window the politics of the moment is trying to drag around. Being a hater is not an investment strategy.

The cleanest proof is Ontario Teachers’. In 2019 the pension put $300 million into SpaceX. That stake is now worth about $16 billion, call it a 5,200% return in seven years. That single gain is more than half the size of Alberta’s entire Heritage Savings Trust Fund. And the point of it isn’t to make anyone rich. It’s to fund stable, defined-benefit pensions for hundreds of thousands of Ontario teachers for decades. Meanwhile the Premier of that same province, Doug Ford, canceled a Starlink contract on principle. Intention and impact are not the same thing. When fiduciary duty gets hijacked by vibes, the people who lose are the teachers and nurses on the other end of the cheque. Have your politics. Vote your conscience. Just don’t confuse a boycott with a strategy.

Canada’s AI “strategy,” and the data-center problem nobody wants to name

About ten days ago Carney’s government tabled its national AI plan, “AI for All.” Six pillars running from protecting democracy to scaling Canadian champions to building global partnerships, wrapped around a $2 billion sovereign-compute commitment and headline promises of $200 billion in GDP and 250,000 jobs. The pillars are fine. They’re all good things to want.

The number is so small relative to the ambition that it reads as a signal more than a strategy, and Mel’s take is the right one. You can’t take a press release to the bank. This is a government with a track record of announcing important things and not executing them at the speed business actually moves. And here’s the line that matters. You do not have an AI strategy if you don’t have a data-center strategy. No data centers means no compute, which means no champions, full stop. That requires energy policy, environmental and infrastructure policy, and a corporate-tax regime that actually rewards the behavior that builds the next Shopify. Right now data centers are becoming the new pipelines in this country, the thing we’d rather argue about than build.

The deeper issue is incentives. As Munger said, show me the incentive and I’ll show you the outcome. When the government is the venture capitalist of last resort, the cheques come with handcuffs, performative conditions, and a capital stack that doesn’t serve shareholders. So you get founders who either pass or get hobbled. Mel and I land in the same place. Smaller hand, less winner-picking, less regulation, and let the dollars we save go to the healthcare, education, and social services we actually want the state running. Government doesn’t need to be good at picking winners. It needs to build the conditions where the private sector can.

Apple’s quiet bet: skip the CapEx war, own the trust

Apple held its WWDC keynote yesterday, and it matters because I’d bet 95 percent of the people reading this operate their lives on iOS.

Apple is the most valuable consumer company on earth and it has badly fumbled the AI ball. It promised “Siri AI” in 2024 and didn’t ship it. Vaporware, which used to be Microsoft’s reputation, not Apple’s. That broke a twenty-year track record of delivery, and it’s been a headwind on the stock for two years. Yesterday they rebranded it “Siri AI” again, showed working demos this time, and pointed it at iOS 27.

Second time looks like the charm.

But the investment case isn’t about whether Siri is best in class. It’s that Apple is the one mega-cap not betting the balance sheet on AI capex. While Meta spends $145 billion this year, Apple is effectively renting compute, leaning on Nvidia chips running in Google’s data centers, and paying Google roughly a billion a year to use Gemini under the hood. They’re happy to be second, third, sixth to the party as long as they ship something better.

That posture is a hedge. If all this capex goes sideways, Apple didn’t pour the money in.

This is also where Anthropic was right and OpenAI was wrong. OpenAI built its name as consumer-first AI, but consumers don’t want to be productive. Step outside your own head if you’re an entrepreneur reading this. Most people on their phones are leaned back, watching YouTube, scrolling, taking a break. They are not going to pay $150 a month for a tree of agents, and they’re using AI the same way they used Google search. OpenAI’s refusal to build an ads product left it stuck. Anthropic went enterprise, where the only things that matter are efficiency and return on invested capital, and built the best harness AI has produced so far in Cowork. That enterprise demand is what lets it underwrite the data-center deals it’s signed with AWS, Google Cloud, and the Colossus buildout.

Apple sits on the consumer layer OpenAI couldn’t monetize. The difference is Apple understands services and the trust that comes with the walled garden. There’s a status-symbol shift hiding in here too. In twenty years the flex isn’t a Rolex, it’s the compute sitting in your home office. I bought a new Mac Studio precisely because I need the horsepower to keep up, and every dollar of that, plus the App Store, Apple Music, and the storefronts, accrues right back to Apple. It rhymes with 2014, when Buffett bought Apple at nine-to-eleven times earnings and made hundreds of billions as services ramped on top of the hardware.

The real edge might be jurisdictional, and it’s the part getting the least attention. Apple has kept its footing with regulators and is positioned to keep operating in the EU and in China, where most of Big Tech has made itself an enemy. New Siri AI features won’t hit those markets at launch, but Apple is one of the very few Western companies that can function there at all. So long as Apple keeps building hardware people trust and doesn’t sleep through a genuine platform shift, the diffusion of consumer AI runs straight through the iPhone. They’ve broken our trust exactly once, on AI, and I think that one comes good.

The through-line

Bonds took the wheel, a hundred billion dollars of founders are about to ring the bell, and the market is plenty big enough to pay for the party. The work isn’t deciding whether AI is a bubble. It’s owning the one or two percent that will carry the whole thing, and refusing to let your politics pick your portfolio.

See you in two weeks.

Podcast & YouTube Recommendations🎙

* Inside The Private Market Boom

* Alex Sacerdote on investing in tech cycles

* NVDA is STILL CHEAP?!

Welcome Back.

Three meetings shaped the last two weeks.

Each one looks like a different story. Each one is actually the same story.

The first was Donald Trump inside the Zhongnanhai - the inner compound of the Chinese leadership. Only three other sitting U.S. presidents have ever been invited in. The second was Mark Carney and Danielle Smith shaking hands on the West Coast pipeline framework, with an October 1 target to designate it a project of national significance and a stated goal of 8 million barrels a day of export capacity by 2030. The third didn’t happen in any one room - it happened in courthouses across Texas, Florida, Oklahoma, Ohio, and Alberta, where county after county is now ruling against the build-out of AI data centers.

That third story is the one most people are missing.

Data centers are unquestionably the most politically cancerous infrastructure that has ever been positioned in the history of North America.

73% of voters — left and right, U.S. and Canada — do not want them in their backyard. The AI capex cycle has become, almost overnight, the most hated topic in domestic politics. And that creates a problem the hyperscalers do not know how to solve.

It also creates an opportunity for the one jurisdiction on this continent that has spent forty years figuring out how to build hated-but-necessary infrastructure: Alberta.

I’m not calling a market top. I want to be clear about that. This is not a 2000 or 2007 setup. But the character of the market is changing. Long rates are breaking out in the U.S. and in Canada, Powell is out and the new Fed chair was sworn in last week, energy is leading, AI names are ripping (Micron just touched a trillion in market cap), and we are about to absorb four of the largest IPOs in human history. Cerebras already priced at $150B. The stock traded to $350 before selling off to the $270s. SpaceX is filing at $1.75T-$2.0T. That single IPO is roughly 75–80% of the entire market cap of the TSX.

When that much private money gets vacuumed into a few mega-listings, it pulls capital out of every other corner of the market to find price.

Expect volatility through the summer.

Where Mel Sharpened It

Mel pushed back on a few of my reads in ways that mattered.

On the Trump–Xi summit: my instinct was to read Trump’s deferential posture as a tactical concession. Mel reframed it through a Foreign Affairs lens - what she called the new G2 world. It is not a Cold War redux of mutually-assured destruction. It is a mutual recognition that neither the U.S. nor China is going to displace the other, and that the two powers are now finding ways to use each other inside defined domains rather than trying to win.

That distinction matters for Canadian foreign policy. If the U.S. is comfortable with a working G2, the question of how Carney’s posture on China (EVs, canola, capital flows) will be received in Washington is suddenly more open than it looked six months ago.

She also pushed me on a question I underweight: where is the line between Trump the person and Trump the president? With most leaders that distinction is invisible. With this one it is unusually visible — and getting that read right is what separates analysts who take him literally but not seriously from analysts who do the opposite.

The most useful frame Mel offered was on the data-center backlash itself. She introduced Maslow’s hierarchy as the right lens. We make the mistake of treating policy choices as binary… yes/no on the ballot, when they are actually continuums of trade-offs. Canadians have had the luxury of not having to make those trade-offs ourselves, which has produced a generation of luxury beliefs about energy that don’t survive contact with the actual physics of running a modern economy.

Caring more about the environment than economics is a position you can only afford when someone else is producing the energy you’re using.

The Macro Read: AI’s New Constraint Is Permission, Not Compute

For two years the binding constraint on the AI cycle has been compute and power. That is changing fast.

Gavin Baker had the cleanest line on it: TSMC could produce $2 trillion in chips this year if they didn’t care about the longevity of their business. There is that much demand. Instead they will produce about 20% of that — roughly $400B — because they are pacing themselves. The chip side of the constraint is being managed by one disciplined Taiwanese foundry.

The newer constraint is permission. Data centers need cool climate, cheap power, abundant land, and a community that will actually let them build. The hyperscalers spent two decades building wherever they wanted because latency didn’t matter much and the physical footprint was small.

That world is gone.

Latency now matters. Bandwidth now matters. Megawatts per square foot has gone vertical. So the build-out is colliding with NIMBY politics in every county in North America right as the cycle needs to accelerate.

We would have a bubble if we could actually get the infrastructure built. We can’t. We have a governor on this runaway wagon, and it is the most healthy thing that could possibly happen to this cycle.

The CEOs running the leading AI labs are not equipped for this fight. Dario Amodei keeps telling everyone the world is going to end and that only he can save it. Sam Altman is, charitably, not lovable. Elon Musk is the most polarizing human being on earth outside of Trump. Alex Karp can’t sit still in a chair. These are not people who win school-board fights in Lubbock or Strathcona County.

The AI CEOs of the leading labs have all failed at the political game. It’s over. Game over. They need to find a new path forward.

That path forward is going to be checks in mailboxes. A municipality is not going to be moved by a one-time $50M payment to a county budget. It will be moved by a $10,000 annual check to 14,000 households for 20 years. That is the math the hyperscalers are about to learn.

Which brings us to Alberta.

The Policy Read: Alberta Is the Most Investable Place on the Continent

Alberta is uniquely positioned for this opportunity for reasons that compound on each other. Northern, cool climate. Cheap energy. Stranded molecules that can be moved behind-the-meter into power generation rather than flared. The only deregulated electricity market in Canada, which means private generators can build and sign power purchase agreements directly with data-center operators - something you cannot do in a system where the government owns the generation. We have engineering talent that knows how to build large, complex, hated infrastructure on time and on budget. We have been doing this for forty years.

The data-center boom is the same political problem as the pipeline - and we are the only province that has spent a generation getting good at solving it.

On the MOU itself, Mel did the heavy lifting again:

* The Carney–Smith agreement on the industrial carbon price gave industry clarity but not competitiveness. Those are two different things. Clarity is necessary. It is not sufficient.

* Brownfield projects will be fine. The economics of greenfield investment - new pipelines, new facilities, the marginal barrel - are still not competitive with comparable jurisdictions elsewhere in the world.

* The pipeline is contingent on Pathways. Pathways is contingent on the pipeline. Someone has to move first.

* Alberta will submit its proposal to the Major Projects Office on July 1. Designation as a project of national significance is targeted for October 1. Separation referendum is October 19.

If you’re the federal government, and you are about to designate a project of national significance two and a half weeks before a separation referendum, you are going to ensure that project is still inside our nation when the dust settles.

Mel was unambiguous on the referendum. She is a federalist. She is voting to stay. And her argument — which I agree with — is that separation does not solve the infrastructure problem. It almost certainly makes it worse. If your primary reason for wanting to separate is that you can’t get a pipeline built, leaving Canada does not get the pipeline built. The path forward is to use the leverage we have inside the federation.

Her closing line was the cleanest framing of the whole episode, and it’s the one I keep coming back to:

Alberta is not in a parent–child relationship with Ottawa. If we stomp our feet and don’t get what we want, we become the actors the rest of the country accuses us of being. This is good for Alberta. This is good for Canada. It should be done on merit. - Mel

What I’m Watching

* The IPO calendar. SpaceX filing at $1.75T is the largest IPO in history. Expect a capital vacuum that pulls money out of mid-caps and small-caps across the index. Use the volatility to add, don’t chase.

* Long rates. The 30-year is breaking out in both the U.S. and Canada. New Fed chair just sworn in. The bond market is the variable that resolves the next regime.

* Alberta data-center build. Watch for the first behind-the-meter announcement from a hyperscaler in Grande Prairie, Edmonton, or Calgary. That is the leading indicator of the regional capex cycle turning on.

* MOU sequencing. July 1 (Alberta submits to MPO) → October 1 (designation target) → October 19 (separation referendum). These three dates compress an enormous amount of political capital into a 16-week window.

* AI infrastructure earnings. Stay long the picks-and-shovels - memory, power semis, optical/photonics, transformers. Sell-side estimates still drag the real demand curve.

Podcast & YouTube Recommendations🎙

* Dan Loeb - ILTB

* Meb Faber and Tom Lee Talk Macro

* A Beautiful Tribecca Apartment

Every twenty or thirty years a new general-purpose technology shows up and quietly rewrites the entire economic order. Steam. Electricity. The PC. The internet. Mobile.

Every single time, the same pattern plays out. One group adopts early and compounds. Another group waits, calls it a fad, tells themselves their experience makes them immune.

By the time the second group looks up from their chosen ignorance, the gap is already too wide to close.

We are sitting inside that window right now. Most people can’t see it.

The line I keep coming back to, and the title of today’s show, is this:

“I don’t use AI” is the new “I don’t use computers.”

In every previous shift, you had a decade or four to catch up. With AI, the gap between early and late adopters is closer to eighteen months. The moat you spent twenty years building gets crossed in a year and a half if you stop defending it.

Mel pushed me on this on the show.

Her question: “is it really that people don’t want to adopt, or is it that they don’t feel they can?” is the right one.

The honest answer is: it’s intimidating, it’s expensive at the leading edge ($30–$250/month per seat, and tens of millions a year if you’re plugging models into an enterprise), and most people don’t yet have a good mental model for what these tools actually do.

But the cost of not engaging is now structural. AI is a motorcycle for the mind in the same way the personal computer was a bicycle. Output expectations are going up everywhere, quality, speed, polish, and the people sitting on the sidelines aren’t holding still. They’re falling behind a curve that is accelerating.

If you take the stance “AI isn’t for me” and frame it as a values statement - that’s actually a career strategy. And it’s a bad one.

The Bull Market Nobody Believes In

The S&P 500, the Nasdaq, the Dow, the TSX, even Chinese equities (the CSI 300 just printed a fresh four-year high) are all hitting new highs at the same time.

Most people don’t realize that. Most people are still telling you it’s a bubble.

Real bubbles are built on euphoria. Right now, AAII bearishness has run hot for ten of the last eleven weeks. Consumer sentiment is near all-time lows. Half of American and Canadian adults think now is a bad time to invest.

That’s not mania. That sounds like disbelief to me...

George Soros’s framing is the one I keep returning to: when he saw a bubble forming, he ran in to buy it. Reflexivity.

Bubbles are powerful trends, and being short them is one of the most expensive seats in the market. Even if this were a bubble (it isn’t), the data says you’d want to own more, not less.

The two truths that can be true at the same time are these:

A technology can be revolutionary. The equities tied to it can still go to zero. Both. Together. Always both.

Cisco at the March 2000 peak: 201× P/E, $555B market cap, briefly the largest company in the world. Revenue kept growing for years afterward. The stock did not.

Today’s setup is genuinely different. Top 10 names are ~40% of the S&P (vs. ~27% at the dot-com peak), but the multiple premium is narrower — 31× for the top 10 vs. 21× for the rest, vs. 43× / 21× in 2000. Multiples less extreme. Concentration more extreme. Different mix, same family of risk.

The other thing that’s different: the data centers are lit up. In 1999 the fiber was dark and the rail cars were empty. Today the hyperscalers cannot buy enough capacity. The picks-and-shovels names are sold out into 2027. Power semis broke a five-year base. Memory pricing is up ~90% in a single quarter. Payback periods on these builds are now under three years.

This is not a bubble looking for a pin. This is a real capital cycle, and we’re closer to the middle of it than the end.

The risk is not that the music stops. The risk is that you capitulate, either out of the trade because you’re tired of being right, or into the trade at the top because the FOMO finally gets you. That’s the Druckenmiller story in 2000: shorted internet, took the loss, covered with discipline, then bought $6B of tech hours from the absolute top because his juniors were printing 3% a day and the social pressure of underperforming his own kids was unbearable. Six weeks later he was down $3B. His own line, years later: “I didn’t learn anything. I already knew I wasn’t supposed to do that.”

The most dangerous moment in a bubble is when staying disciplined feels stupid.

The Alberta Trade — A Once-in-a-Generation Setup

The other big thread from our podcast on monday was Alberta. Mel did the heavy lifting walking through what’s actually happening with Carney, Premier Smith, the November 2025 MOU, the Major Projects Office, and the Pathways CCUS condition.

Here’s the punchline you should walk away with:

The math has changed. U.S. shale plateaued in November 2025. The UAE has fractured away from OPEC+. There is a structural global inventory deficit of ~1.5 billion barrels. Energy demand isn’t plateauing the way the IEA and the EIA modeled…

Why?

Data centers have rewritten the demand curve. We’re heading higher: 150, 160, possibly 200 million barrels a day on the long-run forecast.

Canada has, in spades, exactly what the world is going to need: hydrocarbons, LNG, hydro, uranium, critical minerals, and the geographic position to deliver them. The Western Hemisphere is structurally long this regime.

The thing standing in the way is policy clarity. As Mel walked through, you have three concentric circles that all need to overlap:

* federal government,

* provincial government, and

* industry.

Industry will pull the trigger on egress west the moment they can underwrite a 25% return on invested capital. They have no obligation to drill more for national security reasons — that is not their job, and it doesn’t happen anywhere else in the world.

Where I disagree slightly with Mel, and where she pushed back on me, is on who has more political risk in this negotiation. My read was Smith. Mel’s read was Carney. She’s right. Smith wins politically either way: if Carney blocks the pipeline, that’s a generational gift to a Conservative provincial government heading into the polls. Carney is the one who has to spend political capital and depart from prevailing Laurentian orthodoxy. He has the upside and the downside.

If Carney is serious about economic independence from the United States, which is the entire framing of the elbows-up project, there is a literal silver-bullet sitting in front of him. We just have to build.

Energy companies pay twice — corporate taxes and royalties. Bitumen royalties alone were ~$17B on a $60–70B Alberta budget in ‘24–’25. That’s more than RBC, CIBC, Scotiabank, and Manulife pay in taxes combined. That’s the money that pays for hospitals, social services, education, housing. If we want better health care, more teachers, and more infrastructure, we need to sell more of what we do best. End of story.

Podcast & YouTube Recommendations🎙

* Eric Nuttal talks about 200$ Oil

* The economics and math behind operating and building LLMs

* Buffett Breaks His Silence

By 2027, Bernie Sanders and his cabal of pessimists are going to be in the streets protesting AI. Blaming it for everything…

That's the political tape we're walking into.

The reason isn't ideology. It's economics.

We are watching the most consequential technology transition since the industrial revolution play out in real time, and the early returns look exactly like the early returns of the last one: capital is winning, productivity is exploding, and the wages of the people doing the displaced work are flat to down.

The week the catalysts all printed in one direction

Before we get to the politics, let’s grade the tape, because the tape is what makes the rest of the argument unavoidable.

TSMC’s Q1 2026 call was the cleanest piece of corporate communication I’ve read this quarter. They raised full-year 2026 revenue guidance to better than 30% year-over-year. Gross margin printed a new high of 66.2% — well above the 56%+ long-term guide. Capex is now tracking the high end of the $52–56 billion range, and management said explicitly that revenue growth will outpace capex growth. That’s a margin-expansion story sitting on top of a unit-volume story. They almost never travel together. They are travelling together right now.

ASML revised 2026 revenue up to €36–40 billion (from €34–39 billion) and flagged an EUV supply shortage in 2027. Anthropic’s annualized revenue went from over $9 billion at the end of 2025 to over $30 billion by April 2026. That’s a tripling of ARR in three months. I have spent enough time around early-stage capex cycles to know what they look like, and that is what they look like.

The single most important sentence said on a corporate call this quarter came from TSMC CEO C.C. Wei. I’m going to put it here in full because if you only remember one thing from this newsletter, it should be this:

“The shift from generative AI and the query mode to agentic AI and command and action mode is leading to another step-up in the amount of tokens being consumed.”

Read it twice. Generative AI was a query layer — a person typing a prompt and reading an answer. Agentic AI is a labor layer — a system doing the work end-to-end. The token math compounds, it doesn’t add. And the entity that consumes those tokens isn’t a curious knowledge worker on a free tier. It’s the workflow that used to belong to that worker.

The mix shift inside TSMC tells the same story without the press release language. In the fourth quarter of 2019, high-performance computing was 29% of TSMC’s revenue and smartphones were 53%. Today HPC is 61% and smartphones are 26%. The positions have completely flipped in five years. Advanced nodes — 7nm or below, where TSMC is a de facto monopoly — are now roughly three-quarters of revenue.

I keep coming back to this: 2026 revenue estimates for TSMC are up roughly 25% since September 2025, and the stock is up roughly 60% over that window. That’s a multiple holding while estimates run up. That is the signature of the market pricing more upside, not less. The bubble signal would be one of MSFT, GOOG, META, or AMZN guiding capex down on this week’s prints.

UPDATE; All 4 hyperscalers beat and guided up last night.

The Canadian canary nobody is talking about

If you want a concrete data point that this is no longer a Silicon Valley story, look at Rogers. Last week Rogers Communications cut its 2026 capex plan by roughly 30%, from up-to-$3.5 billion down to $2.5–2.7 billion, and offered voluntary departure packages to about 10,000 employees. That is not a hyperscaler. That is a regulated Canadian incumbent. If a slow-growth telecom is willing to take a 30% capex axe and put 10,000 buyout offers on the table, the cultural permission to use AI to compress headcount is no longer a coastal phenomenon. It’s a TSX-60 phenomenon.

The signal isn’t Rogers specifically. The signal is the speed of cultural adoption. Six months ago, a Canadian incumbent doing this would be a one-off. Today it reads as a template. I am watching BCE and Telus closely for the next two prints, and I am watching the Big Six banks even more closely. The first big bank to publicly acknowledge it is the one to own as a stock and to fear as an employee.

What I think happens next

I want to be clear that I am not a doomer. I am see justification in the AI capex thesis, and can already see the productivity gains that come with it. The world ends up richer on the other side of this. That has been true of every previous technology transition, and there is no reason to think it won’t be true here.

But the transition is the part that’s going to be politically and culturally disorienting. Productivity gains and wage compression can — and during Engels’ Pause, did — coexist for decades. The political response to “the country is much richer in aggregate but my paycheck is flat and my kid can’t get an entry-level analyst job” is not going to be measured. It is going to be Bernie Sanders in the streets, and it is going to be sooner than people think.

The 2027 timing isn’t a guess pulled out of the air. It’s the lag I’d expect between visible displacement and the political reaction to it.

Productivity is already ticking up.

Layoff announcements are already climbing. E

ntry-level wage growth has already stalled. That is the smoke. The fire is the next two years.

The picks-and-shovels remains the story. The foundation models are accelerating and monetizing. The AI-power complements are becoming more cemented, because the agentic layer is, at the end of the day, an electricity load. And budget for political volatility - because in a world where capital is winning this hard, this fast, the politics catch up.

They always do.

Podcast & YouTube Recommendations🎙

* Dwarkesh and Reiner Pope Teach Us Ai:

* Kill Them With Kindness:

Best Links of The Week🔮

* Labour productivity in non-conventional oil extraction is nearly $580 per hour; approximately ten times the rest of the economy Trevor Tombe, University of Calgary

* According to a December 2025 C.D. Howe Institute study, just ONE natural resources project has been approved under the new federal Impact Assessment Act

* saw this new interview with Ben Uyeda on About Art. Ben is an architect-turned multidisciplinary maker/artist/hotelier. He created Reset Hotel, where we stayed in Joshua Tree recently, and I got to spend a little time with him while we were there. A quote really stood out to me:

“The difference between creativity and taste. The culture we’re in right now is mistaking having great taste for being creative. People are mistaking consumption with production, and I think creativity is always about production.“

It’s second nature for me to curate lots of great design ideas and products at this point. It’s what I constantly do online.

It’s much, much harder, and probably much more valuable, to build great things.

* Rogers Communications Inc. is offering voluntary departure packages to about 10,000 employees in its workforce as the telecom giant moves to cut costs amid slowing industry growth. - BNN

* Engles Pause and Employment - Fabricated Knowledge