Tesorb Signal: Tesla, SpaceX, and the Musk Companies

A few weeks ago, we ran an episode about Elon Musk becoming the world’s first trillionaire. This feels only fair. As of now, he isn’t one anymore. From a peak of $1.45 trillion on June 16 to under $800 billion by mid-July, roughly $600 billion evaporated in a month. His life didn’t change at all. This episode walks the dizzying rise and fall, why the scorekeepers can’t agree, and what a fortune like this really is.

A few weeks ago, we ran an episode about Elon Musk becoming the world’s first trillionaire. It was a genuine milestone, a number no human being had ever reached. So this feels only fair. As of now, he isn’t one anymore. The title arrived, and then it left. And almost nothing about his actual life changed either way.

This is Tesorb Signal and this is Lena Ruiz. Today, the un-trillionaire. We’ll walk the dizzying rise and fall of the biggest number ever attached to a person, why the scorekeepers can’t even agree on it, and what that tells you about what a fortune like this really is. Spoiler, it’s not a pile of cash.

Start with the timeline, because it’s genuinely head-spinning. In June, SpaceX went public in the largest market debut ever recorded. The value of Musk’s giant stake in it exploded, and his net worth crossed one trillion dollars for the first time. He became, on paper, the richest human in history by a wide margin.

Then it kept climbing. Here’s the peak, and it’s the number to remember. On June sixteenth, as SpaceX shares hit their all-time high, Musk’s fortune touched roughly one point four five trillion dollars. Not one trillion. Nearly one and a half. That’s the mountaintop this whole story falls from.

And fall it did. Over the following weeks, SpaceX shares slid hard from that high. A five-day losing streak. A launch attempt scrubbed. Each drop carved tens of billions of dollars off Musk’s paper worth. Within a month of the peak, his net worth had dropped back below one trillion, and at one point, by some counts, under eight hundred billion.

Do the math on that swing and it’s staggering. From the June peak to the July trough, something like six hundred billion dollars of Musk’s net worth simply evaporated. Six hundred billion. That’s more than the entire value of most of the largest companies on Earth, gone from one man’s ledger in about a month. And here’s the strange part. His life didn’t change at all.

Which brings us to the wrinkle that makes this an episode. Ask different scorekeepers what Musk is worth today, and you’ll get different answers. One tracker has him hovering near eight hundred billion. Another, closer to nine hundred. Same man, same day, roughly a hundred billion dollars apart. And the gap comes down to judgment calls, not arithmetic.

Why the gap? Because these are estimates, not bank statements. The trackers make different calls, especially about restricted stock, shares he owns but can’t freely sell yet. One major list stripped more than a hundred billion dollars of restricted Tesla stock out of its estimate, and just like that, on paper, he stopped being a trillionaire. A methodology change, not a market crash.

Let’s be fair to the number, though. It’s not meaningless. Musk really is, by every one of those measures, the richest person alive, far ahead of anyone in second place. The fortune is real, the ownership is real, and the scale of it is genuinely without precedent. This isn’t a case of the emperor having no clothes. He has a great many clothes.

And to be fair to Musk himself, he’s said this part out loud. He’s pointed out that almost none of his wealth is cash. It’s ownership stakes in his companies. Less than a tenth of one percent of it, by his own account, is money he could actually spend. On that specific point, he’s simply telling the truth.

But that truth is the whole lesson. A net worth like this isn’t a bank balance. It’s a scoreboard. It’s the market’s live guess about what Musk’s slices of SpaceX and Tesla are worth, updated every second the market is open. When the guess goes up, headlines call him a trillionaire. When it dips, they don’t. The man underneath is unchanged.

Think about what actually happened to Musk personally during that six-hundred-billion-dollar drop. Nothing. He didn’t sell anything. He didn’t lose a house, or a company, or a dollar he was spending. He can’t spend most of it anyway. The number on the scoreboard moved. His day did not.

That’s why the trillionaire headline was always a little bit of a magic trick. It measured a moment, the peak of a brand-new stock, and treated it like a permanent fact about a person. But a stock price isn’t a fact about a person. It’s a mood about a company. And moods move.

We made that same point when the title first arrived. The number was a market verdict on future bets, not a bank balance. Now the verdict has swung the other way, and the lesson only lands harder. If a headline can make you a trillionaire on Monday and un-make you by Friday, then the headline was never really about you. It was about the market.

The number moved a trillion dollars. Nothing else moved at all. And that’s the thing to carry out of this. The next time a fortune like this makes news, up or down, remember what you’re actually watching. Not a man getting richer or poorer. A scoreboard, flickering, on a company whose story is still being written.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

If you have money in a retirement account or index fund, there’s a decent chance you now own a piece of a Mars program. You didn’t choose it. On July 7, SpaceX joined the Nasdaq 100, and every fund tracking that index had to buy shares automatically. One bank estimated $4.3 billion in forced purchases. This episode explains how it happened, the rule that got bent, and why you may own Musk now without knowing it.

If you have money in a retirement account, an index fund, a workplace pension, there’s a decent chance you now own a piece of a Mars program. You didn’t choose it. You may not even like it. But this month, quietly, it landed in your portfolio anyway. And the way it got there is one of the strangest stories in finance.

This is Tesorb Signal and this is Lena Ruiz. Today, how SpaceX ended up in millions of ordinary portfolios without a single one of those investors clicking buy. We’ll walk the index that pulled it in, the rule that got bent to make it happen, and the one number that shows how much money moved on autopilot.

Start with the basics. There’s a famous stock index called the Nasdaq one hundred. It’s a list of the hundred largest companies on the Nasdaq exchange, and enormous pools of money are built to simply mirror it. When you buy one of those funds, you’re not picking stocks. You’re buying the whole list, in one click.

That’s the quiet power of index funds. Trillions of dollars sit in them. And they run on a simple rule. Whatever is on the list, they must hold, in the right proportion. They don’t judge the companies. They don’t read the news. If the list changes, they buy or sell to match it, automatically.

So here’s what happened this month. On July seventh, SpaceX officially joined the Nasdaq one hundred. And the instant it did, every fund built to track that index had no choice. To stay in balance, they had to go out and buy SpaceX shares. Not because a manager liked the company. Because the rule said so.

How much buying? This is the number to hold onto. One major bank estimated the forced purchases at around four point three billion dollars. Billions of dollars flowing into a single stock, in a matter of days, driven entirely by machines rebalancing to match a list. No opinion involved. Just arithmetic.

Now, normally a company waits months after going public before it can join an index like this. SpaceX went public in June and joined in July. It skipped the line. And it could, because the exchange had just rewritten its own rules earlier this year to let giant new companies enter fast. SpaceX was among the first to walk through that new door.

Let’s be fair about the logic. From the exchange’s side, this makes sense. If a company is instantly one of the largest on your market, keeping it out of your headline index looks silly. The old waiting period was built for a slower era. Updating it isn’t obviously wrong. It’s a reasonable response to a genuinely new kind of listing.

And to be fair to SpaceX, none of this is a trick the company pulled. It qualified by being big. The rules changed around it, not because of any maneuver on its part. If you’re one of the most valuable companies in the country, index inclusion is just what happens next. That part is ordinary.

But here’s the wrinkle that makes it strange. Only a tiny sliver of SpaceX shares actually trade freely. Roughly four percent. Musk and insiders hold the rest. So you have enormous, automatic demand from index funds, chasing a very small pool of available shares. A lot of forced buyers, not much to buy. That’s a recipe for a bumpy ride.

And one big scorekeeper looked at all this and said no. The people who run the most famous American index, the S and P five hundred, refused to create the same fast lane. They kept their old requirements, including a rule that a company should actually be profitable first. Two of the biggest index-makers in the world looked at the same company and reached opposite answers.

That disagreement is the whole story in miniature. One index said, it’s huge, let it in now. The other said, show us the profits first, then we’ll talk. Same company, same week, two completely different verdicts on whether it belonged. And your money followed whichever list it happened to be tracking.

We’ve talked before about how the SpaceX debut let regular people bet on Musk for the first time. This is the sequel, and it’s sneakier. Back then, buying the stock was a choice you made. This time, for millions of people, the choice was made for them, by a fund quietly rebalancing in the background. You may own Musk now and not even know it.

So this is what index inclusion really means. It’s not a stamp of approval. It’s not the market saying this company is a safe bet. It’s a mechanical rule forcing billions of dollars into a stock because its name landed on a list. The demand is real, but it’s automatic. It reflects a rule, not a judgment about the future.

You didn’t buy the story. The index bought it for you. And that’s worth remembering the next time a stock soars right after joining an index. Some of that lift isn’t investors believing. It’s just the machines, doing exactly what they were told, buying a company because the list said they had to.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

A year ago this summer, a Tesla pulled away from a curb in Austin with no one in the driver’s seat. The decade-long promise was finally real. So a year on, how much of the rest came true? This episode lines up every specific, dated robotaxi pledge against what actually happened: 500 cars became 20, half the country became two cities, and the fleet actually shrank from its spring peak.

A year ago this summer, a Tesla pulled away from a curb in Austin with no one in the driver’s seat, and a decade-long promise finally came true. Robotaxis were real. Elon Musk had said it would happen for years, and now it had. So a year on, the fair question is simple. How much of the rest of what he promised came true too?

This is Tesorb Signal and this is Lena Ruiz. Today, not a snapshot of the fleet, we did that recently. Today, the ledger. One year of specific, dated robotaxi promises, lined up against what actually happened. We’ll walk the pledges, the gate that keeps sliding, and the single number that sums up the gap.

A couple of weeks back we counted the cars on the road, and found a service map that’s growing far faster than the fleet parked under it. That’s the where. Today is the when. Because the robotaxi story isn’t really about geography. It’s about a string of deadlines, and which ones the calendar actually kept.

Start at the launch. In June of last year, Tesla put robotaxis on public roads in Austin. That part is genuine, and it deserves credit. Getting a car to drive itself on real streets, with no human inside, is a hard engineering feat that most of the industry still can’t match. The achievement is not in doubt.

What happened next is the ledger. So let’s read it, promise by promise. Because around that launch, Musk made a series of very specific predictions, with dates attached. And dates are the one thing you can check without arguing.

Promise one. On a Tesla earnings call, Musk said cars would be driving themselves in Austin with no one in them by the middle of last year. That one, roughly, he hit. The service did launch, though early rides had a safety worker sitting in the passenger seat, watching. A qualified yes.

Promise two. In the summer, he said that half the population of the United States would be covered by Tesla robotaxis by the end of last year. Reality, at year-end, was two cities. Austin, and a service in the Bay Area with human drivers behind the wheel. Not half the country. Two metro areas.

Promise three, and this is the hero number of the whole episode. Last October, Musk said Tesla would have five hundred robotaxis in the Greater Austin area by the end of the year. Five hundred. Keep that figure in your head, because it’s the promise the whole year turns on.

So what showed up? By the time the program hit its first birthday, the fleet running without a human inside was hovering around twenty cars in Austin, and by midsummer it had actually shrunk from its spring peak. Not five hundred. A fraction of a fraction, moving the wrong way. And the map kept right on growing. Just weeks ago, two more Florida cities joined it. The fleet barely budged.

Promise four is the one still hanging over us. Early this year, at a big economic forum, Musk said robotaxis would be widespread across the United States by the end of this year. We’re now past the halfway mark. Two cities have grown to a handful, mostly with tiny fleets. Widespread is doing a lot of work in that sentence.

Let’s be fair about why. Tesla says the caution is on purpose. It’s holding the fleet small until a big rewrite of its self-driving software, version fifteen, is ready. The company argues it would rather move carefully than rush a safety-critical system onto thousands of cars. That’s a defensible position, and worth stating plainly.

And there’s a second fair point. Building a self-driving car at all is genuinely one of the hardest problems in modern engineering. Missing an aggressive deadline on something this hard is not the same as failing. Slow and working still beats fast and broken. Nobody serious wants these cars rushed.

But here’s what the ledger exposes. The problem was never a single missed date. It’s the pattern. Every time a deadline arrives, the promise doesn’t get canceled. It gets moved. Five hundred cars by year-end becomes a scale-up sometime after the software rewrite. Which becomes late next year. Which becomes, well, we’ll see.

We’ve watched this exact move before. Years ago, at a Tesla event, Musk predicted a million robotaxis on the road by twenty-twenty. That number never arrived either. The gate back then was a software update. The gate today is a software update. The date changes. The shape of the promise stays the same.

Now put Tesla next to the company actually running this at scale. We covered that race in an earlier episode. Its rival is delivering hundreds of thousands of paid, fully driverless rides every single week, across many cities, with no safety worker on board. That’s not a demo. That’s a business. And the gap, a year in, is not closing.

So this is what a year of robotaxi promises actually adds up to. One real launch, which counts. And a stack of dated pledges around it that mostly didn’t survive contact with the calendar. Five hundred cars became a couple dozen. Half the country became two cities. Not because the technology is fake, but because the timelines were always fiction.

The calendar keeps a better record than the keynotes. And the next time Musk names a date for robotaxis going widespread, the useful thing to do isn’t to cheer or to scoff. It’s to write the date down, and check back when it arrives. The ledger is doing the arguing for us now.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

Elon Musk spent three weeks in a courtroom trying to stop OpenAI from going public. He lost on a statute of limitations technicality. Three weeks later, OpenAI filed its confidential S-1. Valued at $852 billion with $2 billion in monthly revenue, it could be one of the largest offerings in history. But for every dollar OpenAI earns, it loses $1.20. This episode covers the feud, the filing, and the number everyone should sit with before they buy.

Elon Musk spent three weeks in a courtroom this spring trying to stop one company from going public. He lost. And now that same company is heading to the stock market anyway, in what could be one of the largest offerings in history. The company is OpenAI. The man who tried to block it helped start it.

This is Tesorb Signal and this is Lena Ruiz. Today, the strangest sequel in the Musk story. We’ll walk the feud that put these two men in court, the filing that just moved their fight from the witness stand to the ticker, and the one number inside it that everyone should sit with before they buy.

Start with the history, because it’s genuinely wild. Musk helped found OpenAI back in twenty-fifteen. He put in tens of millions of dollars when it was a tiny nonprofit lab. Then he left, the company turned into a commercial giant behind ChatGPT, and Musk went off and built a rival, the AI company now folded into SpaceX.

Watching his old lab become one of the most valuable startups on Earth clearly stung. So Musk sued. He argued OpenAI had betrayed its founding promise, that it was supposed to stay a nonprofit and instead became a profit machine. He asked a jury to hold them to the original deal.

In May, the jury gave its answer. They sided with OpenAI. But not on the big question of who was right about the mission. They decided Musk had simply waited too long to sue. He missed a legal deadline. The judge agreed and threw the case out.

Musk called it a technicality, and honestly, he has a point. The jury never ruled on whether OpenAI actually broke its promise. They ruled that the clock had run out. Musk says he’ll appeal. So the feud isn’t over. It just moved to a slower, higher court.

Here’s the twist that makes this an episode. Three weeks after that verdict cleared its biggest legal obstacle, OpenAI took the next step toward going public. It filed the confidential paperwork that starts the process, and announced the filing itself, expecting the news to leak anyway. The lawsuit ends, and the road to the stock market opens, in the same season.

Now, “confidential filing” sounds bigger than it is, so let me be clear. It’s not a promise to sell shares tomorrow. It’s a private draft handed to regulators, a way to start the clock quietly and keep your options open. The company hasn’t settled on timing. Reports put a debut anywhere from late this year to sometime next. But the direction is set. The company is preparing to let the public in.

And the numbers are staggering. OpenAI was last valued at around eight hundred and fifty billion dollars, and its boss reportedly won’t accept anything under a trillion. Revenue is enormous and growing fast, running at roughly two billion dollars a month. On the surface, it looks like the most valuable young company the tech world has ever produced.

Let’s give it its due. The product genuinely changed how hundreds of millions of people work and write and search. The revenue is real, and it’s climbing at a speed almost no company has matched. If you only looked at the top line, you’d call this the safest bet in technology. It’s easy to see why the valuation is what it is.

But the filing forces a harder look, and this is the number to keep. For every dollar OpenAI earns, it loses roughly a dollar and twenty cents. It spends more than it takes in, by a wide margin, because building and running these AI systems costs a fortune. The headline is explosive growth. The data underneath is explosive spending.

That’s the whole story in one line. Enormous revenue, even more enormous costs. The company is projected to burn tens of billions of dollars this year alone, most of it on the computers that run the models. Going public means, for the first time, showing all of that to the world in audited detail.

And put the scale in context. We covered the SpaceX debut a few weeks back, the biggest market offering ever recorded. OpenAI going public would land in that same weight class, a company worth close to a trillion dollars selling shares for the first time. Two of the most valuable young firms in the world, hitting the public market inside a single year.

And that’s what makes this moment matter beyond the drama. Right now, only insiders and big funds can own a piece of OpenAI. An IPO changes that. It hands the stock to regular investors and to the index funds sitting quietly in millions of retirement accounts. The bet stops being private. It becomes yours, whether you follow AI or not.

So step back and look at the shape of this. Musk sued to keep his old lab from becoming exactly what it’s now becoming, a for-profit company selling shares to the public. He lost on a calendar technicality. And the company he tried to freeze in place is about to do the most public thing a company can do.

The courtroom was the sideshow. The S-1 is the verdict. Because a jury argued about the past, but a stock market prices the future. And when OpenAI finally opens its books, we’ll all learn something the trial never settled. Not whether Musk was wronged, but whether the most famous company in AI is actually worth what everyone says.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

There’s a new bank in America. It has no branches, lives inside a social media app, and over the past few weeks it quietly opened its doors to millions. X Money launched with 6% APY, 3% cashback, a metal Visa debit card, and FDIC insurance up to $10 million. This episode unpacks the pitch, the perks, and the reason a social network wants to hold your paycheck.

There’s a new bank in America. It has no branches. It doesn’t want your signature. It lives inside a social media app, right next to the arguments and the memes. And over the past few weeks, it quietly opened its doors to millions of people. It’s called X Money.

This is Tesorb Signal and this is Lena Ruiz. Today, the payments product Musk has promised since he bought Twitter, what it actually launched with, and the one eye-catching number that tells you what it’s really trying to do. We’ll cover the pitch, the perks, and the reason a social network wants to hold your paycheck.

Start with what happened. In late June, the app formerly known as Twitter began inviting its paying subscribers in the United States into X Money. Then, on July 27, it threw the doors open to all of them. It’s a digital wallet built right into the feed. You can send money to another person, hold a balance, and spend it with a metal debit card that carries your username on the front.

This has been a long time coming. Musk first floated a payments business back in 2022, right after buying the platform. The idea was always the same. Turn a place where people talk into a place where people also pay. He calls it the everything app, a single place for messages, shopping, and money.

For years it was mostly talk. Licenses took time. Regulators asked questions. The target date was April, then April came and went. Then, in late June, the wallet finally started showing up in real accounts. After four years of the pitch, the product is real.

So what did it launch with? A lot, actually. Three percent cash back on purchases. No foreign transaction fees. Deposits that are federally insured well past the usual quarter-million-dollar limit, up to ten million dollars, by spreading your balance across a network of partner banks. And the money in your wallet earns interest at six percent a year.

Sit with that last number for a second. Six percent. That’s the hero fact of this whole episode, and it’s the one you should keep in your pocket. Because a typical big bank pays you almost nothing on a checking balance. Six percent is many, many times the going rate. So the obvious question is, how can they possibly afford that?

They can afford it because the six percent isn’t really a savings rate. It’s a customer-acquisition cost. In plain terms, it’s a bribe to get you to move your money in. Every dollar sitting in X Money is a dollar that used to sit at your old bank. The interest is the price of prying it loose.

And they’re not subtle about it. On the cheaper subscription tier, you only get the six percent at all if you link your direct deposit. The rate is the bait. The paycheck is the catch.

Let’s be fair here. On paper, this is a genuinely good deal for the user, at least for now. A high yield, cash back, and no foreign fees, all in an app you already have open. If you were going to keep cash somewhere anyway, the terms beat most traditional accounts. That part is real.

And it solves a real problem for a lot of people. Millions of Americans are underbanked, stuck with fees and minimums and branches that closed years ago. A wallet inside an app they already use, with federally insured deposits, is a legitimately useful thing. It’s not just hype.

But look at who can actually get it. This isn’t open to everyone. It’s gated behind a paid subscription, it’s United States only, and it still isn’t legal in every state. X doesn’t hold the money transmitter licenses it needs in New York or Massachusetts. Two of the biggest states in the country, and nobody there can touch it.

The everything app, at launch, is an app for a specific slice of paying, verified users in approved places. And there’s a bigger question hanging over all of it. When a company starts holding people’s savings, the rules get serious fast.

Senator Elizabeth Warren wrote to Musk back in April, before any of this went live. Her concern wasn’t the yield. It was the bank behind it. X Money runs on Cross River Bank, and the FDIC has cited Cross River twice for unsafe or unsound practices. Once in 2018, and again in 2023.

X would point you to the safeguards. Visa’s fraud protections, passkeys, spending limits, and that ten million dollar insurance backstop. Those are real. But Warren’s second point is harder to wave away. The agency built to police consumer products exactly like this one, the Consumer Financial Protection Bureau, was gutted last year, in a campaign Warren notes Musk himself helped drive.

So the product arrives just as the referee walks off the field. And think about what’s being combined here. The same account that pays your rent sits one tap away from a feed built to provoke, to move fast, and to reward outrage. Your money and your worst impulses, in the same app. That isn’t a flaw in the design. That is the design.

So why does Musk want this so badly? Follow the deposits. A social network makes money from attention, which comes and goes. A payment network makes money from being the place your money lives, which is far stickier. Once your paycheck lands there, you don’t leave. You can’t, without moving your whole financial life.

That’s the real prize. Not the debit card, not the cash back. The goal is to become the account, the default place where the money sits. He said it plainly himself. He wants X to be the central source of all monetary transactions. Every feature is a step toward that one outcome.

So this is what the six percent is really telling you. It’s not a gift. It’s the opening move in a much bigger play, one where a company that already knows what you think would also like to know, and hold, what you’re worth. The generosity is temporary. The ambition is not.

The perks are the pitch. The deposits are the point. And the moment to judge X Money isn’t now, while the rate is high and the terms are sweet. It’s a year from now, when the company has your balance and decides what that six percent is really worth to keep.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

Elon Musk called what Neuralink is doing “Jesus-level miracles.” Letting the blind see. Underneath that promise sits a real device: a chip placed directly on the visual cortex with 3,000 electrodes, standing in for an optic nerve with a million fibers. Musk’s own description of the early result: “like Atari graphics.” This episode covers what Blindsight actually does, why more electrodes may not mean more sight, and what the first human implant will really show.

Elon Musk recently described what his brain-implant company is doing with a phrase that’s hard to beat. Jesus-level miracles, he called it. Letting the blind see. It’s a staggering promise. And underneath it sits a real device, real surgery, and a physics problem that no slogan can solve.

This is Tesorb Signal and this is Lena Ruiz. Today, Neuralink’s Blindsight, the implant meant to restore vision. What it actually does. The distance between Atari graphics and a miracle. And why, this time, more electrodes may not mean more sight.

Start with the idea, because it’s a real feat of engineering. Most blindness comes from damage to the eye or to the optic nerve, the cable that carries signals from eye to brain. Blindsight skips that cable completely. It’s a chip placed directly on the visual cortex, the part of the brain that handles sight, stimulating it with tiny electrical pulses.

Do it right, and the person perceives a single point of light, floating in their field of view. Stimulate many points in a coordinated pattern, and in theory you can sketch rough shapes. It means someone with no working eyes at all, even someone blind from birth, could perceive something. That is not marketing. Direct brain stimulation like this is real, and it’s hard-won science.

The full system works like a relay. A small camera, mounted on a pair of glasses, captures the scene in front of the person. A phone processes that image into a simplified pattern. And that pattern is beamed wirelessly to the chip in the visual cortex, which fires the matching electrodes. Eyes replaced by a camera. The optic nerve replaced by a phone and a radio link.

Neuralink’s edge is scale. Its Blindsight chip carries about three thousand electrodes, far more than earlier vision implants, which topped out in the dozens or low hundreds. More electrodes, the logic goes, means more dots of light, means a sharper picture.

But hold that three thousand against what it’s replacing. A healthy optic nerve carries roughly one million nerve fibers. So this implant is standing in for a cable with a million lines, using about three thousand. Musk’s own description of the early result is, to his credit, blunt. Low resolution, he’s said. Like Atari graphics.

The device does carry an FDA breakthrough label, and that phrase gets misread a lot. It isn’t approval. It’s a fast-track designation that speeds up the agency’s review. A signal of promise, not a verdict that the thing works.

And Neuralink didn’t invent this idea. Cortical vision implants have been attempted, in experimental form, for decades. What Neuralink brings is scale and precision, far more electrodes, placed by a surgical robot with a steadiness no human hand can match. The ambition is real. So is the fact that the hardest part was never the wiring.

And this is the claim that outruns the device. Musk has also said resolution could one day exceed normal human vision. Researchers who model exactly this push back hard. A team at the University of Washington simulated an ideal cortical implant and found that even with tens of thousands of electrodes, the picture doesn’t sharpen the way a screen would.

Their point is fundamental. Neurons in the brain are not pixels on a monitor. Stimulating one spot doesn’t produce one clean dot. It smears, and it overlaps with its neighbors.

Adding more electrodes doesn’t simply add clarity, because the brain was never wired to be written to like a display. Better-than-natural sight isn’t a wiring problem you can out-engineer. It runs into how vision actually works.

None of which makes the device pointless. Far from it. For someone living in complete darkness, even a handful of controllable points of light could be life-changing. Sensing a doorway. Crossing a room without help. Reading the shape of a face. That would be a genuine gift. The trouble isn’t the machine. It’s the size of the promise wrapped around it.

We’ve come back to Neuralink twice lately. The patients already using its implants to move a cursor and type with their thoughts, and the company’s push to scale up surgeries. Vision is the next frontier, and a far harder one. Reading signals out of the brain is one challenge. Writing a picture into it is a different, deeper one.

The company is now recruiting its first vision patients across the United States, Canada, and the United Kingdom. If the timeline holds, someone who has lived in darkness, possibly since birth, could perceive computer-generated light for the first time this year. That’s a genuine milestone, whatever the resolution turns out to be.

And there’s a quieter reason to watch the language closely. When a company frames early, low-resolution results as miracles, it sets an expectation the device may not meet for years. Patients and families deserve the honest version. Real progress, arriving slowly, not sight switched back on like a light.

So set the miracle aside and weigh the machine.

Weighed honestly, Blindsight is two things at once. A serious attempt to give some form of sight to people who have none, which deserves to be taken seriously and cheered on. And a promise, better than natural vision, that gets ahead of the biology by a very wide margin.

The first human implant is expected this year. When it happens, watch what the patient can actually do. Not the word miracle. The dots on the black. That’s where the truth of this will show up first.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

Elon Musk introduced Grok 4.5 with a four-word pitch: “Opus-class, but cheaper.” Independent testing firm Artificial Analysis published scores within a day. Grok 4.5 ranked fourth, not first. But it posted the biggest single-generation jump on the leaderboard, at 60% below frontier pricing. The real story isn’t the ranking. It’s that near-frontier intelligence just got dramatically cheaper, with one serious asterisk: the hallucination rate more than doubled.

On the eighth of July, Elon Musk introduced SpaceX’s newest AI model with a four-word pitch. Opus-class, but cheaper. That kind of claim used to be impossible to check, because nobody outside the company had the numbers. This time, independent numbers showed up within a day. And they tell a sharper story than the slogan.

This is Tesorb Signal and this is Lena Ruiz. Today, Grok 4.5, and what independent testing actually found. The claim Musk made. Where the model really landed against the competition. And the one number in the fine print the launch didn’t mention.

Quick context, because the corporate map is strange. Grok is the AI built by xAI, which is now part of SpaceX after the two companies merged earlier this year. So, yes, a rocket company ships a chatbot. And it built this particular model alongside Cursor, the coding startup SpaceX bought right after it went public.

That Cursor detail isn’t trivia. Cursor is one of the most popular tools programmers use to write code with AI. Training a model on real sessions from inside that tool is like teaching a chef in a working kitchen instead of handing them a cookbook. It’s why the model’s strongest results, by the independent numbers, show up in coding and agent work, not in raw general smarts.

Grok 4.5 launched on July eighth, aimed squarely at coding and agent tasks, the kind of work where an AI runs on its own for a stretch. Musk’s pitch, word for word: an Opus-class model, but faster, more token-efficient, and lower cost. Opus being the frontier model from Anthropic that many developers treat as the bar to beat.

So did it beat the bar? Within a day, the independent testing firm Artificial Analysis published its scores. And the headline result is a ranking. On their overall intelligence index, Grok 4.5 came in fourth. Not first. Fourth.

And it’s worth being clear about who’s scoring. Artificial Analysis is an independent shop, not SpaceX and not a rival. It runs the same battery of tests across every model and publishes the results. That’s exactly why its numbers carry more weight than any launch-day slide, from anyone.

The three models sitting above Grok come from Anthropic and OpenAI, and that includes the current Opus, the exact model Musk name-checked. So on the specific comparison he invoked, Grok lands a notch below it. Now, fourth out of every model on the board is genuinely strong. It’s just not the picture the words Opus-class tend to paint.

And this is the part the skeptics skip. Grok 4.5 is the biggest single-generation jump anyone has posted on that leaderboard, up sixteen points from the previous Grok in a matter of months. It tops at least one demanding agent test outright. And it does all of it at a price roughly sixty percent below Opus or the comparable OpenAI model.

And cheaper frontier-grade AI is not a footnote. When a near-top model shows up at a fraction of the going rate, it pressures everyone else to cut prices too. Some analysts are already calling this the opening shot of an AI price war. That’s the part of this launch that actually reshapes the market, more than any single leaderboard slot.

Where does it genuinely shine? On coding-agent tests, the independent numbers put Grok 4.5 roughly level with OpenAI’s strongest coding model, at a small fraction of the cost per task. For a developer paying the bill, matching performance at lower cost isn’t a tie. It’s a win.

Musk also quietly walked the slogan back a notch, and that matters. He clarified that Grok 4.5 is roughly comparable to the previous Opus, not the current one. And that version of the claim, the independent data broadly supports. The billboard oversold it. The footnote was basically right.

But there’s a number in the fine print worth sitting with. That same testing firm found Grok 4.5’s hallucination rate, how often it states something false with full confidence, more than doubled versus the previous version. From about one in four answers, to more than half.

Read that twice. A real jump in capability, and a real jump in confident wrongness, shipped in the same model. And xAI released it without a public model card, the standard document a lab puts out to describe what its model can’t do and where it fails. Skipping that, on a release with a known accuracy problem, only buys a company so much benefit of the doubt.

A few episodes back, we covered how Grok 5 keeps slipping its release date. This is the model that shipped instead. Not the leap that was promised. A sharp, cheap, in-between step, with an asterisk attached.

So the slogan is the wrong thing to argue about.

Whether it’s Opus-class misses the real signal. The story isn’t that Grok matched the best model, because it didn’t, quite. The story is that near-frontier intelligence just got dramatically cheaper. A model this capable, at a fraction of the price, changes who can afford to run one at all. That’s the shift worth watching. Not the bragging rights. The price tag.

And the open question sitting underneath it. Whether a model that now makes things up more than half the time can be trusted with exactly the unsupervised work it’s being sold to do.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

Tesla tore out the Model S and Model X line at Fremont and replaced it, in about 46 days, with the first production line ever built for a humanoid robot. The design capacity: one million Optimus units per year. The actual count: a few hundred ever built, zero sold. This episode covers what Tesla genuinely pulled off this summer, why the gap between the line and the number matters, and what to listen for on the earnings call.

Elon Musk walked a factory floor in Fremont on the first of July. Arms crossed, hard hats all around him. Behind him, the first production line ever built for a humanoid robot. In front of him, a claim that this same line will one day make a million robots a year. The line is real. The million is not. Not yet.

This is Tesorb Signal and this is Lena Ruiz. Today, Tesla’s Optimus robot, and the gap between a factory line and a factory. What Tesla actually built this summer. Why the number one million keeps getting attached to it. And what the real production count looks like today.

Start with what Tesla genuinely pulled off, because it isn’t nothing. This spring, the last Model S and Model X rolled off the line in Fremont, closing a run of more than a decade for two cars that helped make the company. Then Tesla tore that line out.

In its place, by some accounts in about forty-six days, the company installed a production line built specifically for Optimus, its humanoid robot. Think about that swap. The space that used to build luxury sedans now builds a machine with arms, hands, and a torso, one that’s supposed to walk.

A Tesla executive described the new line as modular. Designed to be rebuilt as the robot’s hardware keeps changing, so the company doesn’t rip everything out with each new version. For a product this early, that’s a smart way to build. Fast, flexible, and honest about how much is still in flux.

The version being built is called Gen three, the third major redesign, shown off earlier this year. New hands. Better balance. A real attempt at a machine you could put to work. That part is genuinely exciting. The design is maturing. The open question is whether the manufacturing can catch up to it.

And a humanoid is hard to build in a way a car never was. A car is a box on wheels Tesla has made millions of. A humanoid needs hands that grip without crushing, motors small enough to fit a finger joint, and software that keeps two legs upright. Each of those is its own frontier. Stacking them into one machine, at volume, is close to unprecedented.

So where does one million come from? That’s the design capacity Tesla has stated for the Fremont line. One million Optimus units a year, once it’s fully ramped. It’s a real target, named by the company. It’s also, right now, a spec on a slide. Not a number coming off the floor.

Because the gap is stark. Tesla publishes no Optimus production figures at all. The best independent reporting, from the tech outlet The Information, put the total number of robots ever built at a few hundred as of last year. And not one of them has been sold to an outside customer. The handful that exist work inside Tesla’s own buildings.

So sit with the contrast. A line designed for a million robots a year. A few hundred built so far. Zero sold. That’s the distance between the announcement and the assembly floor.

For now, the plan is deliberately internal. Tesla has said it wants thousands of Optimus units working in its own plants before it sells a single one outside. That’s a reasonable way to shake out the problems. It also means the million-a-year headline and the paying-customer reality are separated by years, not months.

To his credit, Musk hasn’t hidden this. When people online claimed production had quietly begun, he pushed back. Walking that same line on the first of July, he said output will be, in his words, extremely slow at first, as everything is new. That’s not the voice of a man promising a million by Christmas.

And there’s a real reason for the caution. Optimus has around ten thousand unique parts. A car line leans on decades of suppliers who already know how to make each piece. For a humanoid robot, a lot of that supply chain simply doesn’t exist yet. Somebody has to invent it, part by part, before scale is even possible.

There’s a bigger version of this planned too. A second Optimus factory at Giga Texas, aimed at ten million units a year, with production penciled in for the summer of twenty twenty-seven. That’s the long vision. Cars taught Tesla how to build hard things at massive scale. The whole bet is that robots are the next thing it scales.

We talked a while back about how the clearest signal on Optimus wasn’t the robot itself. It was the real estate. Tesla clearing a flagship car line to make the room. This is that same story, one step further along. The room is cleared. The line is installed. What’s still missing is the thing the line is supposed to make.

And the timing makes the silence louder. Tesla just posted a strong quarter of car deliveries, and the whole investor story is tilting away from cars, toward robots and autonomy. A real Optimus number would push that story forward. Its absence says something too.

So don’t watch the render. Watch the floor.

A production line is not production. Tesla has built the place where a million robots could someday be made. Whether that place becomes a real factory, or stays a very expensive room with a very new line in it, comes down to one number the company still won’t print. An actual count.

The next honest checkpoint is the earnings call on July twenty-second. Everything else is a photo on a factory floor. Listen for whether anyone, finally, says how many Optimus robots Tesla has actually built.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

Every time someone asks Grok a question, electricity flows from somewhere. For xAI’s Colossus 2 supercomputer, that somewhere is dozens of trailer-mounted gas turbines in Mississippi. A federal lawsuit says not one of them ever got an air permit. Then the Department of Justice stepped in, calling the turbines a national security matter. This episode covers the power behind Colossus, the lawsuit, and the $2.8 billion turbine plan buried in SpaceX’s own IPO paperwork.

Every time someone asks Grok a question, and the company says that happens hundreds of millions of times a day, electricity flows from somewhere. For xAI’s newest supercomputer, a lot of that somewhere is a site just across the Mississippi state line, where dozens of gas turbines hum around the clock. A federal lawsuit says not one of them ever got an air permit.

This is Tesorb Signal and this is Lena Ruiz. Today, the power behind Colossus 2. The lawsuit over how xAI keeps its supercomputer running, the national security defense the government just mounted on its behalf, and the number buried in SpaceX’s own IPO paperwork that suggests this fight is only beginning.

Start with the facts on the ground. Colossus 2 is xAI’s second giant data center, in Southaven, Mississippi, just south of Memphis. To power it, the company installed a fleet of trailer-mounted gas turbines. Industrial machines, cousins of jet engines, each one bolted to a flatbed. Effectively, a power plant on wheels.

In April, the NAACP sued xAI, represented by two environmental law groups, arguing the turbines violate the Clean Air Act because they run without air permits.

The plaintiffs’ filings say the turbines could emit more than seventeen hundred tons of smog-forming pollution a year, in a region that already fails federal air standards. They also note the site sits about half a mile from homes and a mile from an elementary school.

xAI’s position, and Mississippi regulators agreed when the turbines went in, is that these are mobile units. And mobile equipment is treated differently under air law than a permanent power plant. The plaintiffs answer that a turbine that never actually moves is a power plant, whatever the paperwork calls it. That definitional fight, mobile versus stationary, is the whole case.

There’s history here, too. At its first Memphis data center, the original Colossus, xAI ran dozens of unpermitted turbines starting in 2024, according to the same groups now suing. After a legal notice, the company removed most of them and got permits for the rest.

The plaintiffs say xAI then repeated the playbook across the state line. The company, they note, had described its own plan for powering the second site as copy and paste.

Then in June, the case took a turn almost nobody saw coming. The Department of Justice asked the court to step in on xAI’s side, arguing the turbines are a matter of national security.

A Pentagon official filed a declaration saying Grok is one of only a handful of frontier AI systems supporting American defense work, and that losing Colossus 2 would degrade capabilities the military now relies on.

Sit with that for a second. An air permit dispute in a Mississippi suburb is now, officially, a national security question. Whatever you make of the merits, that is a genuinely new kind of argument in American environmental law. The judge hasn’t ruled. The turbines are still running.

And the stakes reach beyond Grok. Colossus doesn’t just train xAI’s own models anymore. We’ve covered how SpaceX rents that supercomputer out to rival AI labs. The turbines in this lawsuit power a business, not just a chatbot.

The near-term thing to watch is the injunction. In May, the plaintiffs asked the court to shut the turbines down while the case proceeds. That single ruling decides whether Colossus 2 keeps its power supply through the trial, and it’s the reason the government’s national security filing arrived when it did.

And this is where the show’s favorite document comes back. The S-1, the IPO filing we’ve mined before for everything SpaceX had to disclose on its way to going public. The filing acknowledges the lawsuit. It also discloses the plan.

SpaceX intends to buy roughly two point eight billion dollars worth of additional gas turbines over the next three years, with at least two billion of that earmarked for mobile units. The exact category this lawsuit is about.

That’s the number to hold onto today. Two point eight billion. The turbines aren’t a stopgap while something cleaner gets built. In the company’s own capital plan, they are the roadmap.

Two fairness notes, because both matter. First, xAI is not alone here. The entire AI industry is slamming into the limits of the power grid, and waiting years for a utility connection is a real constraint every data center builder faces. Gas turbines are the industry’s shortcut, not just Musk’s.

Second, xAI has responded to pressure before. The turbine removals and permits at the first Colossus happened after the public pushback, not before. Slow, but not immovable.

Regular listeners might remember our episode on how the AI boom runs on batteries, with Megapacks buffering power at these same data centers. Both stories are true at once. The batteries are the polished half of the power story. The turbines are the half that ended up in federal court.

The turbines were never hidden. The permits were the question. Colossus 2 exists because xAI moved faster than the grid and faster than the permitting process, and whether that reads as bold or reckless is now up to a federal judge.

But the S-1 already told you the ending the company expects. Billions more in turbines, on trailers, wherever the next data center rises. The headlines will cover the verdict. The capital plan already made the decision.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.

Tesla already told the world it delivered 480,000 cars last quarter. A blowout. The stock fell anyway. The earnings report on July 22 is where the real question gets settled: did those cars make money? This episode previews what the income statement needs to show, the three lines to watch on the call, and why $25 billion in capex makes the margin question sharper than ever.

Tesla already told the world how many cars it delivered last quarter. Roughly four hundred eighty thousand. A blowout. The stock fell anyway. So what’s left for the earnings report on July twenty-second? Only the question that actually decides Tesla’s year.

This is Tesorb Signal and this is Lena Ruiz. Today we’re previewing Tesla’s second quarter earnings. Why the delivery number is already old news, the lines on the income statement that aren’t, and what the market is really asking Tesla to prove.

First, the part that’s already public, because Tesla reports its delivery count three weeks before it reports its money. Four hundred eighty thousand one hundred twenty-six vehicles delivered. Up twenty-five percent from a year ago. The best second quarter in company history, and the first real growth after two straight down years.

We covered that delivery beat in a recent episode, along with the strange selloff that followed it. The market’s verdict was blunt. Selling more cars is not the same thing as earning more money. The earnings report is where that distinction gets settled.

The quarter had other footnotes worth carrying in. Energy storage kept growing at its usual breakneck pace. And the books quietly closed on the Model S and Model X, which we gave a proper sendoff earlier this year. Tesla now sells exactly three vehicles. The lineup has never been simpler, and the company has never depended more on the Model 3 and Model Y.

One more piece of context for the day itself. The report lands after the market closes on July twenty-second, with the call that afternoon. And it arrives with the stock already lagging the broader market this year, even after the delivery news. Wall Street is bracing for a wide range of outcomes, in both directions. Whatever gets said, expect the reaction to be loud.

Now the part that’s genuinely unknown. Margins. Selling seventy-four thousand more cars than Wall Street expected is only good news if those cars made money.

Think of a restaurant that fills every table by discounting every meal. The dining room looks triumphant. The books tell a different story. The delivery report showed us Tesla’s dining room. July twenty-second opens the books.

There’s a reason to ask. Gas prices spiked this spring during the conflict in the Middle East, and expensive gas pushes buyers toward EVs. That tailwind is already fading with the ceasefire. So the question inside the question is this. Did Tesla buy this quarter with price cuts, or earn it with demand? And does whatever powered it survive the summer?

Then there’s the spending, and this is the number to hold onto today. Tesla raised its capital spending plan for the year to more than twenty-five billion dollars. Twenty-five billion. Chips, data centers, robot production lines, robotaxi infrastructure. And the company has signaled that free cash flow may run negative for the rest of the year while it builds.

That sets up the real trial on July twenty-second. Not whether Tesla grew. We know it grew. Whether the growth and the spending can coexist without the margins giving way underneath them.

Three things to listen for on the call. One, automotive gross margin. The single line that says whether the delivery beat was profitable growth or bought growth.

Two, any hard numbers on the robotaxi service, which just expanded to Miami, its third state. Riders, vehicles, revenue. Anything beyond a map. We’ve covered the gap between that map and the actual fleet before, and every earnings call is a fresh chance for Tesla to close it with data.

Three, the language around cheaper models, because this quarter’s beat leaned almost entirely on two vehicles, and the entry-level strategy decides whether the growth continues once fuel prices normalize.

And keep one eye on the energy business while everyone stares at the cars. We’ve covered before how Tesla’s batteries earn roughly twice the gross margin of its vehicles. Storage deployments surged again this quarter. Quarter by quarter, that quieter business does more of the profit lifting, and July twenty-second will show exactly how much.

In fairness to the bulls, heavy spending isn’t a flaw in their story. It is the story. If you believe Tesla is an AI company that happens to make cars, then twenty-five billion in capex is the price of admission to the future, not a leak in the ship. The bear case and the bull case will read the exact same numbers that afternoon and walk away disagreeing.

One more thread to watch. The gap with BYD, the Chinese giant we covered when it took the global EV crown. That gap narrowed sharply this quarter. BYD’s electric sales fell while Tesla’s jumped. If the trend holds, the crown becomes contested again by year end. The earnings call is where Tesla gets to say whether it’s chasing that crown or done playing that game entirely.

So the deliveries were the opening argument. July twenty-second is the cross-examination. One number settles it. Automotive gross margin.

If it holds firm while deliveries surge, the growth is real and the machine is finally working again. If it sags, then the market’s summer selloff was simply early, not wrong. Either way, don’t let anyone tell you the delivery beat answered the question. It only asked it.

That’s the signal. I’m Lena Ruiz with the Tesorb Signal podcast. For more news about Tesla, SpaceX, and Elon Musk’s companies, visit our website at tesorb.com.

Got a tip or feedback? Send it to [email protected], we’d love to hear from you. This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. See you on the next one.