Tesorb Signal: Tesla, SpaceX, and the Musk Companies

LandSpace became only the third private company to land an orbital-class booster, but a post-landing fire destroyed the vehicle. With 660 successful booster landings, SpaceX’s lead is measured in reuse cadence, and China needs that cadence for its Starlink competitors.

In December of twenty fifteen, a Falcon nine booster came back from the edge of space and set itself down on a concrete pad in Florida. It took China more than ten years to do the same thing. Last month, it did. And SpaceX’s answer to that isn’t a new rocket. It’s a scoreboard that reads six hundred sixty.

This is Tesorb Signal and this is Lena Ruiz. Today, what China actually landed, why it matters more than the highlight reel suggests, and why the gap with SpaceX is measured in a number, not in years.

The company is LandSpace, a private startup out of Beijing, founded the same year SpaceX first landed a booster. Its rocket is named for the Vermilion Bird of Chinese mythology. It stands about sixty-six meters tall, roughly the height of a Falcon nine, and it’s built from stainless steel and burns methane. If that recipe sounds familiar, it should. It’s the Starship recipe.

On the morning of August nineteenth, Beijing time, the rocket lifted off from a commercial launch zone in the Gobi Desert on only its second flight ever. Two minutes in, the stages separated. The upper stage carried an internet satellite to orbit.

The booster turned around, flew back, and eight minutes after liftoff set down on its landing legs on a pad in Gansu province, nearly four hundred kilometers from where it started.

The fairness points come first, and there are two. One, this is real engineering, not a stunt. LandSpace is now only the third private company on Earth to land an orbital-class booster, after SpaceX and Blue Origin, which joined the club late last year. And it did it on the rocket’s second flight, which is faster than SpaceX managed. Falcon nine flew for five years before it stuck a landing.

Why a landing matters so much comes down to money. The booster is the biggest and most expensive part of the rocket, the majority of the cost of the whole vehicle. Throw it away and every launch starts from zero. Get it back and the price of the next flight drops to fuel, inspection, and the upper stage. That’s the entire reason SpaceX can launch twice a week and everyone else can’t.

Two, it wasn’t even China’s first landing of the summer. In July, a state-owned Long March booster came back and settled into a net on a ship at sea. So in the space of six weeks, China recovered two orbital boosters, with two different methods, from two different kinds of organizations. That’s not one team getting lucky. That’s a sector arriving.

The methods matter, too. The state rocket came down into a net on a ship. The private one came down on legs, the Falcon nine way. Legs mean a booster can land on any flat pad the company owns, which is what lets SpaceX launch from Florida and land in Florida. LandSpace picked the harder method and the one with the bigger payoff.

Now the data underneath the moment. Minutes after touchdown, video showed a fire at the base of the booster. It burned through a landing leg, and the rocket toppled over. LandSpace had said it wanted to fly that booster again. That booster won’t fly as-is. The landing was a success. The recovery wasn’t.

And there’s a detail in the design choices. Stainless steel is cheap and tolerates heat. Methane burns clean, which makes an engine easier to inspect and fly again. SpaceX chose that combination for Starship because it’s built for reuse, not for a single flight. LandSpace chose it for the same reason. The imitation is a compliment, and it’s also a strategy.

And the rocket itself is a smaller machine than the comparison suggests. In its reusable configuration it lifts roughly half of what a Falcon nine carries to low orbit. Chinese reporting rounds it up to Falcon nine class. The payload numbers don’t.

Then there’s the scoreboard. SpaceX has flown more than a hundred missions this year alone. Days after China landed its first booster, a Falcon nine booster flew for the thirty-seventh time, a record. Another one flew its thirty-fifth mission at the start of this month. The gap isn’t whether you can land a rocket.

It’s whether you can land it on a Tuesday and launch it again before the end of the month.

Back in the spring, we argued that nobody could catch SpaceX, and the reason was cadence, not ceremony. Nothing about August changes the cadence. What it changes is the question. It used to be whether China’s private rocket companies could get here at all. Now it’s how fast they climb the count.

And the climb has a customer. China is building two satellite constellations meant to compete with Starlink, and both need thousands of launches at a price only reusable rockets can hit. LandSpace doesn’t have to beat SpaceX for foreign business. It has to become the Falcon nine for a domestic market that SpaceX can’t touch. That’s a different race, and it’s one SpaceX isn’t allowed to enter.

There’s a callback worth making. Starship is built from stainless steel and burns methane, and its booster has been caught by the tower since late twenty twenty-four. But every Starship flight so far has been suborbital.

So a Chinese startup, on its second flight, recovered a stainless steel methane booster from a launch that reached orbit before SpaceX did. That’s a footnote today. It might not stay one.

So count landings, not landmarks. A first landing is the entry fee. The business is the number after it, and SpaceX’s number is six hundred sixty. China just paid the fee, with a booster that fell over. The next figure to watch isn’t whether LandSpace lands again. It’s whether the next landing comes on a booster that already flew.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Tesla opened Cybercab rides to the public in Austin, but the launch event lasted just fifteen minutes with no price, no rollout timeline, and no next city announced. The purpose-built robotaxi undercuts Tesla’s own Model Y fleet on fares, though shares fell as analysts flagged the missing details.

Tesla spent two weeks teasing it. An executive promised an exciting forty-eight hours. Elon Musk said a storm of Cybercabs was about to flood Austin. And when the moment came, the presentation ran about fifteen minutes. No livestream. No Musk on stage. No price. That was the whole show, for the product Tesla has called its future.

This is Tesorb Signal and this is Lena Ruiz. Today we’re separating the car from the show. What Tesla promised for last Thursday, what it actually delivered, and what the fare screen in the app tells us that the stage never did.

First, the show. The event happened at a music venue in downtown Austin, invitation only. Guests had to be twenty-one, with identification checked at the door. Many were influencers who post supportive content about Musk’s companies, and they signed non-disclosure agreements and agreed to hold their posts until the presentation ended.

There was no public broadcast. Tesla had said it would stream the launch for everyone else, and in the minutes before it started, fans were posting to ask where the stream had gone. Instead, they refreshed the state of Texas’s public registry of autonomous vehicles, watching the Cybercab count tick up as a proxy for what was happening inside.

The presentation itself surfaced later on video. Tesla’s Cybercab lead told the room that in the future there will be millions of Cybercabs everywhere. Musk did not appear. Compare that with the original reveal two years ago, on a Hollywood studio lot, with Musk riding onto the stage in a Cybercab and humanoid robots serving drinks to the crowd. This was the opposite of that.

Now, the fairness points, because they matter. The thing that actually counts did happen. On Friday afternoon, Tesla opened Cybercab bookings to the general public across its Austin service area. Anyone with the robotaxi app can now hail a car with no steering wheel and be driven across town by software.

Roughly forty-five of them were registered with Texas at launch, and they cover the same map the Model Y fleet already serves. The new car didn’t change where Tesla drives, only what it drives. That is a real product carrying real strangers, and only one other company, Zoox in Las Vegas, is doing anything comparable.

And the price is the second fairness point. Riders posting screenshots showed the same trip quoted at about twelve dollars in a Model Y robotaxi and roughly a third less in a Cybercab. On day one, the two-seater undercut Tesla’s own fleet. Musk has claimed for years that a purpose-built car would be cheaper to run. That claim just showed up as a smaller number on a screen. That’s data.

And the marketing was loud right up to the door. Hours before the event, Musk amplified a chart claiming a million dollars buys eight of Waymo’s new robotaxis or roughly fifty-five Cybercabs, using an outside analyst’s guess of eighteen thousand dollars per car. Tesla Asia announced the Cybercab will tour Hong Kong, Tokyo, Beijing, and Shanghai this month.

The company was happy to talk numbers, as long as they were somebody else’s numbers.

Everything else was withheld. No consumer price. The under-thirty-thousand-dollar figure everyone cites comes from Musk’s stage remarks in twenty twenty-four, and two years later there is still no confirmed number. No production target. No rollout calendar. No next city. No way to order one. Analysts had listed exactly those items as what they were waiting for, and none of them got an answer.

There was one detail that slipped through, and it’s small but telling. The updated app terms bar riders under thirteen from a Cybercab. The Model Y robotaxis allow riders as young as eight. Tesla drew a stricter line around the car it’s proudest of. Nobody explained why.

The market kept score the way it usually does. Tesla shares rose more than five percent into the event on anticipation. They fell sharply the next day, as analysts described the launch as underwhelming and at least one bank reiterated a sell rating, pointing to the missing price and the missing rollout target. One research note published before the event had predicted exactly this.

If it was merely an unveiling with few cars committed, the stock would sell off through it.

Why so quiet? We can’t read Tesla’s mind, and we won’t pretend to. But the possibilities are worth naming. One, the federal safety audit we covered last time makes every public claim a potential exhibit, and a company under audit talks less. Two, the tightly managed guest list echoes the original robotaxi launch, when Tesla also leaned on friendly accounts to shape the first impression.

Three, maybe there was simply nothing new to say beyond what the app already showed.

So separate the car from the show. The car is real, it’s cheaper than the one it replaces, and it’s on the road. The show promised a storm and delivered forty-five vehicles, a closed room, and a video posted after the fact.

Fifteen minutes is how long Tesla was willing to spend on the questions investors most wanted answered, and the silence around price and scale is the loudest thing Tesla said all week.

Watch two things from here. Whether Cybercab rides expand beyond Austin before the end of the year, and whether a consumer price ever gets a date. The car arrived. The pitch is still in the garage.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Last Thursday, Tesla unveiled a car with no steering wheel in Austin and started offering rides. By Friday afternoon, those rides were open to anyone.

Last Thursday, Tesla unveiled a car with no steering wheel in Austin and started offering rides. By Friday afternoon, those rides were open to anyone. Within a day, the federal government opened an investigation into how that car got permission to exist. Not a recall. Not a crash probe. An audit of paperwork. And the paperwork comes down to one number Tesla wanted no part of. Two thousand five hundred.

This is Tesorb Signal and this is Lena Ruiz. Today, the story the launch coverage skipped. How Tesla certified the Cybercab, why federal regulators want to see the math behind that certification, and what happened the last time a company tried the same move.

Start with how a car gets approved in America, because the answer surprises people. It doesn’t. No federal inspector signs off on a new model before it goes on sale. The manufacturer tests it, decides it meets the federal safety standards, and puts a sticker on the door frame saying so. That’s self-certification, and it’s how every car in every parking lot got there.

So the first fairness point is simple. Tesla did what every automaker does. It certified its own vehicle. Nothing about that step is unusual, and nothing about it is illegal.

The problem is what the standards assume. They were written across six decades for cars with a person behind the wheel. They specify where the steering column goes, how the brake pedal must perform, what the mirrors must show, how an airbag deploys around a driver’s hands. The Cybercab has no steering column, no pedals, and no mirrors. It has two seats, butterfly doors, and a computer.

Tesla’s position, as reported by Reuters and the Wall Street Journal, is that the Cybercab meets all applicable standards. The load-bearing word is applicable. Tesla concluded that rules written for a driver simply don’t apply to a car that doesn’t have one. That’s not a loophole. It’s an interpretation.

And now the National Highway Traffic Safety Administration wants to see how Tesla arrived at it.

Because there was another path. Federal law lets a manufacturer ask for an exemption from specific standards if it can prove the vehicle is just as safe without them. It’s slower. It requires evidence, public comment, and a written decision. And it comes with a ceiling, a hard cap on how many vehicles an exempted design can put on the road each year.

Elon Musk has said Tesla intends to build Cybercabs by the millions. A cap of a few thousand a year is a rounding error against that ambition. So Tesla skipped the exemption and self-certified. Which means the choice was never about whether the car is safe. It was about whether Tesla would accept a limit on how many it could deploy.

The morning after the launch, the safety agency announced what it calls an audit query, covering roughly a thousand Cybercabs, to examine the process and the technical data Tesla relied on. Notice the scale. Roughly forty-five Cybercabs were registered with Texas at launch. The audit covers a thousand.

The agency is scoping what Tesla has built, not just what it has deployed, and that tells you it expects the fleet to grow while the questions are still open.

The agency was also careful about what an audit query means. It has not found a defect. It has not said the car is unsafe or that the launch was illegal. That’s the second fairness point, and it’s the agency’s own framing, not ours.

The precedent is Zoox. In two thousand twenty-two, the Amazon-owned company self-certified its own robotaxi, a boxy four-seater with no driver controls, using the same argument Tesla is using now. Within two months, regulators sent a formal demand for its reasoning. The following spring, they opened an audit query. That audit stayed open for two and a half years.

The audit closed only after Zoox gave up on the self-certification argument and applied for a limited exemption in the summer of twenty twenty-five. The full commercial exemption came a year later, in late July of this year. The standard annual cap, for two years, with regulators holding the right to tighten the terms. Even after the exemption, Zoox recalled its entire fleet, all one hundred and five vehicles, over a software failure with its smoke detection system.

Zoox began charging for rides in Las Vegas less than two weeks later.

So put the two timelines side by side. Zoox self-certified in the summer of twenty twenty-two and got to charge a fare in the summer of twenty twenty-six. Four years. Tesla self-certified, launched, and charged its first fare in the same week. And the audit landed the same week too.

The Tesla situation isn’t identical. The agency has spent this year loosening the exemption rules and talking about balancing innovation with oversight, and Zoox’s audit began under a different administration. Tesla also has far more road data from its Model Y robotaxis than Zoox had in twenty twenty-two. It may get a faster hearing.

But the two possible endings are very different. If regulators accept Tesla’s reading, then any company can strip out the wheel and pedals, declare the old rules inapplicable, and deploy without a cap. Tesla’s paperwork becomes the industry’s template. If regulators reject it, Tesla gets pushed onto the same road Zoox took, and that road comes with a number on it.

Put the launch and the audit on the same page. Tesla didn’t bypass the regulators. It bet that the rules don’t apply to a car without a driver, and it made that bet in public with paying passengers in the back seat. Two thousand five hundred is the ceiling Tesla refused to accept, and the ceiling is now the whole question.

We covered the launch itself in a recent episode, and the gap between the hype and the fare screen gets its own turn next time. This week, the document to watch isn’t a product page. It’s a docket.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Two hundred forty-seven percent. That’s how fast the newest business inside SpaceX grew over the past year.

Two hundred forty-seven percent. That’s how fast the newest business inside SpaceX grew over the past year. Not Starlink. Not launch. Artificial intelligence. And Elon Musk just told his own employees that this month, that business will out-earn the rockets, the satellites, and everything else the company does. Combined.

This is Tesorb Signal and this is Lena Ruiz. Today, the moment a rocket company stops being a rocket company. We’ll walk through what Musk actually said, what the second quarter numbers show, and the doubling that has to happen for the prediction to come true on schedule.

In August, Musk sat for a half-hour all-hands that was later reported widely. Buried in it was a line Wall Street is still chewing on. Our AI revenue, he said, not probably, definitely, will exceed all other SpaceX revenue, probably in September. Then he went further. It will significantly exceed everything else by the fourth quarter.

A quick refresher on what that segment even is. When SpaceX absorbed xAI, it inherited Grok, the chatbot, and Colossus, one of the largest computing clusters on the planet. Since then the company has been renting that computing out to outsiders, from coding tools to rival AI labs. We flagged that landlord business on this show back when the first big rental deal was signed.

The second quarter numbers explain his confidence. Total revenue nearly doubled from a year earlier. The AI segment brought in about two and a half billion dollars of it, growing faster than every other line of business. And the company pointed to roughly fourteen billion dollars in contracted cloud deals, agreements already signed and waiting to turn into revenue.

One question worth holding onto is what sits inside that AI line. Chatbot subscriptions, computing rentals, and infrastructure deals all get counted together. Renting chips to a rival lab is steady utility money. Subscription growth can evaporate. The headline number is clean. The stuff inside it is not all the same quality.

Give the bull case its due, because it’s earned. The growth is real, not accounting. Outside companies are paying real money for Colossus computing time. Goldman Sachs models the AI line at more than fifteen billion dollars for the full year. And SpaceX has a habit, one this show has documented, of building physical infrastructure faster than anyone thought possible.

The strategic logic holds up too. Launch revenue is lumpy, and Starlink, while growing fast, is a subscription business that eventually meets the limits of its market. Computing demand right now looks bottomless. Musk told employees AI will be ninety-nine percent of the company’s value within five years. He’s describing a different company than the one investors bought.

He also named the destination. Ten gigawatts of AI computing by the end of next year, which by his own math would generate three hundred to five hundred billion dollars a year. Notice what that math assumes. Every watt built is a watt sold, at premium prices, indefinitely. Capacity isn’t revenue. Ask any airline with empty seats.

Now the data check. In the second quarter, the AI segment averaged under nine hundred million dollars a month. Everything else averaged about one point seven five billion. For AI to pass the rest of the company in September, its monthly revenue has to more than double from that pace. In roughly ninety days.

Big contracts switching on could do it. The signed cloud deals don’t ramp gradually, they land in steps, and a single step can move a monthly number a long way. But that cuts both ways. A forecast built on a few giant contracts is a forecast that lives or dies on a few signatures and start dates. It’s a very specific promise on a very short clock.

And this show keeps a ledger for a reason. Probably in September comes from the same forecaster who promised robotaxis every year since twenty nineteen, and a Roadster that took nine years. Sometimes the promises land. The rockets really do come back and land on ships. But the timelines run long often enough that the word probably is doing real work in that sentence.

The market’s reaction is its own story. Through the summer, the stock spent stretches trading below its debut price, weighed down by expiring lockups and a capital spending plan that would make a utility blush. Investors are still pricing a launch company with an AI experiment attached. Musk is describing an AI company with a launch division. One of those descriptions is wrong.

So what settles it? Not the all-hands. The next earnings report, which will show September’s revenue mix in black and white. If AI crossed over, it shows up in the segment table. If it didn’t, watch whether the language shifts from September to fourth quarter to next year. On this show, that slide has a name. It’s how promises age.

Growth of two hundred forty-seven percent is a fact. The crossover is still a forecast. When a company redefines what it is, watch the revenue mix, not the announcement. The mix doesn’t have a hype cycle.

So what does a rocket company become when rockets are the side business? We may find out within the month. If the crossover lands, it rewrites how SpaceX gets valued, and it makes the biggest IPO in history look like it was priced off the wrong business. If it slips, it joins the ledger. Either way, we’ll cover the numbers when they print.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

AI data centers have pushed 29.4 billion dollars in power costs onto the electric bills of 65 million people. Elon Musk’s companies sit on both sides of the meter, with xAI driving demand and Tesla selling the grid-scale batteries to manage it.

Twenty-nine point four billion dollars. That’s how much of the capacity bill in America’s largest power market traces back to data centers, according to the market’s own independent monitor. Four annual auctions. Nearly half the total. And that bill doesn’t get mailed to the tech companies. It gets folded into the electric bills of sixty-five million people.

This is Tesorb Signal and this is Lena Ruiz. Today, the power bill episode. The AI boom has started showing up in the one monthly payment nobody can opt out of. We’ll look at what the grid data actually shows, how Elon Musk’s companies ended up on both sides of your meter, and who pays for the machines from here.

Start with the shape of the problem. For most of the past two decades, American electricity demand barely moved. Utilities planned around flat lines. Then AI arrived, and every big model needs a building full of chips running around the clock. A single large data center can draw as much power as a mid-sized city. Hundreds more are waiting in the connection queue.

Researchers at the Federal Reserve Bank of Dallas put numbers on it this year. By their estimate, the data centers already built have pushed wholesale electricity prices up a few percent nationwide. Where the buildings cluster, the effect is far bigger. And if proposed construction keeps coming, they project generation costs climbing another twenty percent or more by twenty twenty-eight.

The clearest paper trail runs through PJM, the grid operator for thirteen states and Washington DC. Every year it runs an auction that pays power plants to promise they’ll be available at the grid’s peak hour. Think of it as a retainer for electricity. And that retainer gets passed straight through to customer bills.

Because your electric bill is really three bills stapled together. The energy itself. The wires and poles that carry it. And that reliability retainer. Data center demand pushes on all three at once. That’s why the increases feel like they come from everywhere and nowhere, and why they’re so hard to argue with on a line-item basis.

The retainers have exploded. In Virginia, home to the world’s densest data center corridor, residential electricity prices climbed more than thirteen percent in a single year. Polls show most voters there blame the data centers. Whether they’re entirely right or not, the political mood around these buildings has turned.

Now put Elon Musk’s companies on the map. xAI’s Colossus cluster around Memphis is one of the hungriest new loads in the country. It crossed roughly a gigawatt of computing this spring, with a target near two. That’s the demand side of the ledger. We’ve covered the fights it sparked, from gas turbines to a federal courtroom, in earlier episodes.

But the same ecosystem is also selling the fix. Tesla’s energy division makes the Megapack, the grid-scale battery built to smooth exactly the kind of violent, spiky load an AI cluster creates. xAI alone has bought more than a billion dollars’ worth. Utilities are lining up behind it. One Musk company drives the demand shock. Another sells the shock absorber.

Time for the other side of the ledger, because it exists. The Electric Power Research Institute, a utility industry research group, found that through at least twenty twenty-four, data centers actually pushed retail electricity costs down. More demand spread the grid’s fixed costs across more buyers. The problem isn’t data centers as a category. It’s the speed of this particular wave.

And the industry isn’t pretending the problem away. Battery-buffered sites like Colossus can back off the grid at peak hours, which is part of how it got connected so fast. Some AI companies have gone further. Anthropic, the maker of Claude, publicly committed this year to covering the electricity price increases its data centers cause for consumers.

So where does the fairness run out? In the gap between when the costs land and when the benefits do. The bills are rising now. The promised payoff, cheaper medicine, smarter software, abundant intelligence, arrives on someone’s roadmap. Ratepayers are fronting a loan to the AI industry whether they signed up for it or not.

And the rules for splitting the check are still being written. Grid operators are debating whether data centers should bring their own power plants, sign long-term supply contracts, or keep buying from the shared pool. State regulators are weighing special rate classes for big computing loads. How those fights land will decide what your bill looks like in twenty thirty.

For the Musk ecosystem, the stakes cut both ways. Every dollar of public anger at data center power bills is a regulatory risk for xAI’s expansion. And every utility scrambling to firm up its grid is a customer for Tesla’s batteries. The same headline that threatens one company’s buildout is a sales pitch for the other’s product line.

And the next giant load is already under construction. The chip factory rising on the Giga Texas campus, the one that kicked off this whole show, will draw industrial amounts of power before it ships a single wafer. The ecosystem isn’t done asking the grid for more. It’s just getting started.

So bring this back to the number on your own bill. Twenty-nine point four billion dollars in one power market, traced to one customer class, recovered from everyone. The AI boom’s first invoice didn’t go to investors. It went to ratepayers. The race now is whether the benefits show up before the patience runs out.

Watch three things from here. Whether regulators force big computing loads to pay their own way. Whether battery prices fall fast enough to blunt the peaks. And whether more AI companies follow the pledge to cover what they cost their neighbors. Any one of those bends this story. All three together end 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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

In October twenty twenty-four, Elon Musk stood on a movie lot in Los Angeles and said the Cybercab would cost under thirty thousand dollars. This month in Austin, invited guests finally rode in one on a public street.

In October twenty twenty-four, Elon Musk stood on a movie lot in Los Angeles and said the Cybercab would cost under thirty thousand dollars. This month in Austin, invited guests finally rode in one on a public street. Nearly two years later, the car is real. The price tag still isn’t.

This is Tesorb Signal, and I’m Lena Ruiz. Three things today. What actually happened at the Cybercab launch. What Tesla built to get here. And the three numbers that still aren’t on the table.

Let’s set the scene. Tesla ran a sweepstakes in the second half of August. Every robotaxi ride in Austin earned an entry, and five winners got a seat at an invitation-only event on September third. Guests had to be twenty-one. IDs were checked. Plus-ones were allowed, but not if they were content creators. This was a controlled room, and Tesla wanted it that way.

The vehicle itself is the one you’ve seen in renders since twenty twenty-four. Two seats. No steering wheel. No pedals. Butterfly doors. Wireless charging. The production version adds a Starlink antenna, which makes it the first Tesla to lean on SpaceX hardware for connectivity. Tesla has told analysts it’s the first of several purpose-built robotaxi shapes, with a van to follow.

Two fairness points. First, this was not a loop around a closed lot. The twenty twenty-four reveal was exactly that, and critics said so. Tesla spent the summer doing the work a public launch needs. Employee rides on private roads, then employee rides on public roads. Emergency-response training with Austin first responders.

A permit filing for a charging and car-wash hub in South Austin with forty-eight charging posts. That’s infrastructure, not theater.

Second, the staged approach is the responsible one. A vehicle with no manual controls has no human fallback in the cabin. Tesla says a remote operator layer runs alongside the car’s Starlink and cellular links. Building driving data on a new platform before opening it to the public is what Musk said the company would do on the second quarter earnings call, and that’s what it did.

Now the ledger. Three numbers were promised or implied in twenty twenty-four, and each one is still open.

Number one, the price. The reveal put a specific figure on it. Tesla hasn’t sold a Cybercab to anyone, hasn’t said when it will, and hasn’t said whether individuals will ever be able to buy one. A price nobody can pay is a number with no test. Until a sale happens, the most famous figure from the reveal stays a slide.

Number two, volume. Cybercab production started at Giga Texas this year, and roughly two hundred fifty units had been spotted in the factory lots by late July. But Tesla has said meaningful volume is a twenty twenty-seven story, and that battery packs, not the assembly line, are the near-term limit. Two hundred fifty cars in a parking lot is not a fleet. It’s inventory waiting for software.

Number three, the fleet. Texas now has about four hundred twenty vehicles registered to the robotaxi service across Austin, Dallas, and Houston, most of them Model Ys. But the number active at any moment is far smaller. Tesla has said it’s deliberately holding back more Model Y conversions to make room for Cybercab. So the question for September isn’t whether a Cybercab can carry a passenger.

It’s how many are in the app on a Tuesday afternoon, and whether that’s more than the Model Ys it replaced.

Think of it like a new restaurant that opens with a tasting dinner for five contest winners. The food is real. The kitchen works. But the restaurant isn’t open until a stranger can walk in and get a table. For Cybercab, the stranger is anyone in Austin who opens the robotaxi app and gets matched to a car with no steering wheel. Tesla has said public access follows the event.

Watch for the first day that actually happens.

And there’s a Vegas wrinkle. Nevada just cleared Tesla for up to five thousand robotaxis in Clark County, and Tesla told the regulators it will start that service with Model Ys, with Cybercab joining later. So the car that launched in Austin this month is already second in line in the state that gave Tesla its biggest permit. Two cities, two fleets, one vehicle that hasn’t scaled yet.

One more thing the event can’t show you. Production Cybercabs reportedly carry a more powerful driving computer than today’s Model Ys, built for the next software version, which Tesla says has roughly ten times the parameters of the current build. Musk has said the robotaxi network won’t expand in a big way until that software ships.

So the car launched before its brain did. That’s not unusual for Tesla. It does mean the launch is a hardware milestone, and the scale-up is still a software date.

So forget the event. Watch the app.

The Cybercab launch answered the question everyone stopped asking, which is whether the car exists. It does, and it drives. The questions it didn’t answer are the ones from the reveal. Under thirty thousand dollars. Volume production. A fleet that grows on its own. Nearly two years in, the car made it to the street. The numbers are still on the stage.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

The Boring Company is tunneling in three states at once. It signed its first international contract, in Dubai.

The Boring Company is tunneling in three states at once. It signed its first international contract, in Dubai. It gave free tunnels to Dallas and New Orleans, and it’s talking to Universal in Orlando. And the number of Loop systems anywhere on Earth actually carrying paying passengers is still one.

This is Tesorb Signal, and I’m Lena Ruiz. Three things today. What’s actually being dug right now, city by city. What slipped, because one big promise already has. And the difference between a hole in the ground and a transit system, which is the whole business.

Start in Nashville, because that’s where the machines are. The Music City Loop is a privately funded tunnel network connecting downtown to the airport, the first segment of a plan that stretches to twenty-five miles. The first boring machine has been underground since early this year. A second machine finished commissioning in June. A third was scheduled to ship in August.

Three machines under one city is more digging than this company has ever done at once.

Washington noticed. The U.S. Transportation Secretary went down into the Nashville tunnel in late June for a tour. The company has locked in most of its permits, including state approval for twenty-five miles of tunnel and an agreement for a station at the airport itself. It’s also bought land on the west side of town for a second line aimed at locals rather than tourists.

Dubai is the other headline. In February, the company signed a definitive contract with Dubai’s transport authority. Phase one is about four miles of tunnel and four stations connecting the financial district to the Dubai Mall, at a cost around a hundred fifty million dollars, with a full network planned at around five hundred forty-five million.

Two fairness points. First, the digging is real, and it’s getting fast. In March, one of the company’s machines completed a two point three mile tunnel in Las Vegas, its longest ever. The machines are iterating like rockets, each new unit built with data from the last one. And the company is now boring through three different geologies at once.

Clay in Vegas, limestone in Nashville, sandy clay in Texas. That’s an operating company, not a science project.

Second, the one system that exists actually works. The original convention center loop was built in about a year for a fixed price of forty-seven million dollars, a fraction of what comparable transit costs. The Vegas network has carried more than four million passengers and moved eighty-two thousand people during a single trade show.

Whatever the limits of Teslas in tunnels, the product does what the brochure says at the scale it currently runs.

Now the slip, because our ledgers don’t skip those. When the Dubai project was announced, a UAE minister said the first loop would be operational by the second quarter of this year. The second quarter came and went. Surface construction is now underway, but tunneling hasn’t started. Operational became under construction, and the timeline stretched by at least a year.

That’s not a scandal. It is a familiar shape.

And Nashville comes with a fight attached. The project routes under state highways, which means it never needed approval from the city’s own council. The mayor has raised questions about surface integrity along the route. Residents packed meetings. And tunneling professionals told Bloomberg they’re baffled by the company’s cost claims, especially in middle Tennessee’s sinkhole-prone limestone.

The company says the project is safe, permitted, and fully self-funded. Both of those things can be true at once, and that tension is the story to watch underground.

Here’s the part that matters more than any single city. This company spent years as a punchline with one short tunnel. That era is over. It has a pipeline, a factory for boring machines, a global operations center in Texas, and governments courting it on two continents. The open question moved. It’s no longer whether the Boring Company can dig. It’s whether it can open.

Because opening is the hard part. A transit system needs stations with fire safety sign-offs, vehicles with operators or approved autonomy, insurance, fares, and a city willing to certify all of it. Vegas took years to grow from three stations to ten. Every announced city is at the beginning of that process, not the end. The machines can outrun the paperwork, and so far they always have.

Think of it like a restaurant group that’s signed leases in five cities and has construction crews in three of them. The leases are real. The crews are real. But a restaurant group with one open location is still a one-restaurant company at dinner time. The second opening is the one that proves it’s a chain.

So don’t watch the drills. Watch the doors.

One operating system, three cities of digging, five cities of promises. Nashville’s first segment is the nearest test, and Dubai is the proof point for everything international. Whichever opens first becomes system number two, and system number two is the moment this stops being a Vegas curiosity and becomes an infrastructure company.

Until a second set of doors opens, the count that matters is stuck at one.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Last November, Tesla shareholders approved the largest pay deal in corporate history. Up to a trillion dollars in stock for Elon Musk, if Tesla grows into an eight point five trillion dollar company and hits a dozen operational targets along the way.

Last November, Tesla shareholders approved the largest pay deal in corporate history. Up to a trillion dollars in stock for Elon Musk, if Tesla grows into an eight point five trillion dollar company and hits a dozen operational targets along the way. Nine months in, the number of tranches he’s unlocked is zero. Today we read the package the way it was actually written. As a scoreboard.

This is Tesorb Signal, and I’m Lena Ruiz. Three things today. How the trillion dollar package actually works, because the design is stranger than the headline. Where every milestone stands as of this month. And why a package like this paid off once before, which is the strongest argument that it could again.

Start with the design. The package is twelve equal slices of Tesla stock, about thirty-five million shares each. To unlock a slice, two things have to happen at once. Tesla’s market value has to climb to the next rung on a ladder that starts at two trillion dollars and tops out at eight point five trillion. And the company has to check off one operational goal from a list of twelve.

The operational list is the telling part, because it’s not stock-price games. Deliver twenty million vehicles, cumulative. Put a million robotaxis into commercial service. Deliver a million humanoid robots. Sign up ten million Full Self-Driving subscribers. And grow annual adjusted earnings through eight tiers, starting at fifty billion dollars and ending at four hundred billion.

Musk gets no salary. He has ten years, and most shares can’t be cashed until the next decade.

Now the scoreboard, milestone by milestone, using the most recent numbers.

Market value first. The ladder starts at two trillion. Tesla today is worth roughly one point four trillion. The first rung, the easiest one on the whole ladder, still requires almost doubling. The last rung is about six times today’s value.

Vehicles. Twenty million cumulative is the target, and this is the healthiest line on the board. Tesla crossed ten million total vehicles built this summer. Halfway there, with a real factory system behind it. At recent production rates the second ten million takes five to six years. Tight against a ten year clock, but on the board.

Robotaxis. One million in commercial operation is the target. The unsupervised fleet today is a couple dozen vehicles. Whatever you believe about the technology, the current number and the target number differ by a factor of tens of thousands. Same story for the robots. A million Optimus units delivered is the target, and deliveries to outside customers today are effectively zero.

Earnings. The first tier is fifty billion dollars in adjusted annual earnings. Tesla’s most recent full year came in under twelve billion, and earnings this year have been falling quarter to quarter. The eighth tier, four hundred billion, would be more than double what Apple earns today.

So the honest reading is that nothing has been unlocked, and nothing is imminent. Two fairness points, and the first one is the whole counterargument. This exact structure looked absurd once before and paid in full. The twenty eighteen package required Tesla to grow from about sixty billion to six hundred fifty billion dollars in market value. Analysts called it laughable.

Tesla blew through every rung within five years, and that package is now worth north of a hundred billion dollars. The board built this one the same way on purpose.

Second, the alignment case is real. Musk earns nothing unless the company grows several fold, which means existing shareholders would gain trillions before he collects the headline number. Seventy-five percent of shareholders voted yes with their eyes open. Whatever you think of the size, the structure only pays on delivered results. In a world of guaranteed CEO bonuses, that’s not nothing.

But the ledger discipline cuts both ways, so two things belong on the board next to the fairness. Proxy advisers who opposed the package flagged that the fine print gives the board discretion to hand over slices even if targets are missed. And Electrek has reported that a merger with SpaceX could trigger vesting provisions automatically, without any milestones at all.

Side doors matter on a scoreboard, because they change what the score means.

Think of it like a gym membership where the trainer only gets paid when you hit posted numbers. Lift this much, weigh this much, run this fast. The chart is on the wall and everyone can read it. The catch is that the gym’s owner is the trainer’s biggest fan, and there’s a clause that says the owner can pay him anyway. The chart is honest. Whether the chart is binding is a different question.

The reason to keep this scoreboard running is that it turns a decade of Tesla promises into something you can check quarterly. Every earnings report now updates five lines. Deliveries, robotaxi count, robot count, subscriptions, earnings. The package converts the vision into arithmetic, and arithmetic doesn’t do vibes.

So read the pay package as a to-do list.

An eight point five trillion dollar finish line, twelve boxes, zero checked so far, and a man who cleared an equally ridiculous list once before. Both of those facts are true at the same time, and the next few years get scored in public. When the first slice unlocks, whenever that happens, check which box got ticked and how. The box tells you what Tesla became.

The side doors tell you whether the scoreboard ever mattered.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Try to name one big, dated Elon Musk promise that arrived on the date. The Roadster didn’t.

Try to name one big, dated Elon Musk promise that arrived on the date. The Roadster didn’t. The Semi didn’t. The million robotaxis didn’t. Grok 5 hasn’t. Over the past two years on this show, we’ve built promise ledgers for ten of the biggest deadlines across the Musk companies. The score, as of this summer, is zero for ten on time. And yet most of them still mattered.

This is Tesorb Signal, and I’m Lena Ruiz. This is our one hundredth episode, so we’re doing something different. Three things today. What the ledgers say when you stack them on top of each other. Why the misses keep happening, in a pattern you can actually use. And the exchange rate that turns a Musk date into a real one.

Start with the stack. The Roadster was unveiled in twenty seventeen for a twenty twenty launch. It still hasn’t shipped. The Semi was unveiled the same year for twenty nineteen. Volume production started this April, seven years late. The Cybertruck was promised for twenty twenty-one and arrived at the end of twenty twenty-three. A million robotaxis were promised for twenty twenty.

Six years later the unsupervised fleet is a couple dozen cars.

The pattern holds off the road too. Full Self-Driving was supposed to demonstrate a coast to coast drive in 2017. The Las Vegas airport tunnel connector was supposed to be underground by early this year. It opened late, and partly on surface streets. Dubai’s tunnel system was supposed to be operational by the second quarter of this year. Tunneling hasn’t started.

And Grok 5 has now missed three announced windows in nine months.

So that’s the score. Now, two fairness points, and they’re load-bearing today. First, late is not the same as never. Of those ten promises, most eventually became real things. The Cybertruck exists and sold in volume. The Semi just took its largest order ever, five hundred trucks from Einride. Reusable rockets were mocked as fantasy and have now landed more than six hundred times.

The dates were wrong. The direction, most of the time, was not.

Second, the deadlines aren’t really forecasts, and Musk has said as much. He’s described his timelines as aspirational, as targets designed to compress effort. Inside the companies, an impossible date functions as a filter and a fuel. People who study Musk’s operations keep landing on the same observation. The absurd deadline is the management tool. The miss is the cost of the tool.

You don’t get the six hundred landings without the culture that also produces the Roadster.

But here’s what the stacked ledgers show that no single one does. The size of the slip isn’t random. It scales with how novel the promise is. Things Tesla and SpaceX had built before ran months late. Things nobody had ever built ran years late. And things that required other people’s behavior to change, customers, regulators, whole cities, ran longest of all, or haven’t landed yet.

That gives you a rough exchange rate. A Musk date on familiar hardware, add about half the stated time. A Musk date on first-of-its-kind hardware, double it, maybe triple it. A Musk date on software that has to work everywhere, or on a business that needs a million customers to show up, treat the date as a direction, not an appointment. Not one entry in our ledgers has beaten that conversion.

Several have needed more.

Think of it like a friend who is always late, but reliably late. If she says seven, you know it’s eight thirty. You don’t stop inviting her. You stop showing up at seven. The people who get burned by Musk time aren’t the ones who believe in the projects. They’re the ones who set their watches to the announcement.

And that’s who this actually matters for. When a date lands in a headline, it moves money on that day. We’ve watched a Nevada permit add five percent to a stock in an afternoon, priced on a fleet that doesn’t exist yet. The market keeps buying seven o’clock. The record says eight thirty. The gap between those two is where both the risk and, honestly, some of the opportunity lives.

One more thing the stack shows. The promises that eventually paid off shared a trait. Each had a factory, a launch pad, or a production line you could watch in the meantime. The ones still floating, the million robotaxis, the full Grok 5, are the ones where the visible progress and the promised scale are furthest apart.

When the gap between what you can see and what’s being promised gets wide enough, that’s the tell.

So set your watch to it.

Zero for ten on time. Most of them real anyway. That’s not a contradiction, it’s a system, and after a hundred episodes of ledgers it’s the most consistent finding this show has produced. When the next date drops, and one will drop this quarter, don’t ask whether it will slip. Ask what you can watch while it does. The factories tell the truth on a delay. The dates never did.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.

Before anything else today, thank you. This is episode one hundred of Tesorb Signal.

Before anything else today, thank you. This is episode one hundred of Tesorb Signal. When the first one went out in April, the plan was one episode a week. You kept showing up, so we kept going. A hundred episodes in a little over four months. Everyone who listened. Everyone who shared. Everyone who wrote in. You made this happen. So today we’re doing something a little different.

This is Tesorb Signal and this is Lena Ruiz. No single headline today. Instead, the three stories that defined our first hundred episodes, and where each one is headed next. SpaceX becoming a public company. Tesla turning from a car maker into a robot maker. And Starlink, the business that quietly outgrew every headline written about it. Then a promise about the next hundred.

One number to carry with you. Twelve million. That’s how many Starlink subscribers SpaceX reported at the end of June. A year earlier, it was six. We’ll come back to it.

Start with the biggest story we’ve ever covered. When this show launched, SpaceX was the most valuable private company on earth, and the IPO was a rumor with a price tag. By mid May, the filing was public. By mid June, the stock was trading on the Nasdaq under the ticker SPCX at a valuation of one point seven seven trillion dollars. The largest IPO in history, and it wasn’t close.

We called that filing one company and three bets. Rockets. Satellites. And AI. Going public didn’t change that. What it changed is that every claim SpaceX makes now arrives with a quarterly report attached. For a show built on headlines versus data, that’s the best thing that happened to us all year.

And the stock has done what new stocks do. It spiked to a high within days of listing. It fell by more than half by early August as the first insider shares came free. Then it climbed back. As of this week, the company is worth about one point nine trillion dollars. Above the IPO price. Through all of that, the analyst consensus never moved off buy.

The next chapter is already being written in public. This week, SpaceX announced a hundred billion dollar spaceport in Louisiana. Musk put a date on his three and a half trillion dollar revenue target, somewhere around twenty thirty-three. Those used to be posts. Now they’re guidance. And guidance is something we can hold a company to.

Second story. In January, Musk gave the Model S and Model X what he called an honorable discharge. In May, the last ones rolled out of Fremont. Then Tesla tore out the line in forty-six days to make room for Optimus. We said at the time that the clearest signal wasn’t the robot. It was the real estate. A car company doesn’t clear its flagship floor for a side project.

Since then, Tesla has kept moving. Cybercab, the car with no steering wheel, entered production in Texas this summer and started carrying employees around the campus. The robotaxi service now runs unsupervised in six cities across Texas and Florida, seven days a week. FSD subscriptions are approaching a million and a half. That’s a company that has picked its future and is acting on it.

But the ledger still has open lines. Optimus production was promised for late July or August. Tesla’s July letter moved that to later this year, and the company has still never published a robot count. The unsupervised robotaxi fleet, by independent estimates, is under a hundred vehicles. Waymo runs about thirty-five hundred. The direction is right. The distance is real.

Watch for two moments. The first paying ride in a Cybercab. And the day Tesla prints its first Optimus number. That’s when the robot story stops being a floor plan and becomes a factory. We’ll be there for both.

Third story, and the one that keeps embarrassing the skeptics. Back in the spring, we said Starlink was SpaceX’s real financial engine. Then the filing came out and proved it. Starlink was the only one of the three segments making money. It’s the reason the IPO was possible at all.

The growth hasn’t slowed since. Connectivity revenue in the second quarter was about four point three billion dollars, up two thirds from a year earlier. Amazon’s rival network has a few hundred satellites in orbit. Starlink has more than eight thousand. In most of our episodes, the data trails the headline. Starlink is the one where it runs ahead.

The next act for Starlink isn’t dishes. It’s phones. Direct to cell service already reaches tens of millions of people through carrier deals, with no hardware to buy. A carrier joint venture. A possible standalone listing. Data centers in orbit. All of it is on the table for a business that was a footnote three years ago.

So step back from the week and look at the four months. SpaceX became a public company. Tesla cleared its oldest factory floor to build robots. Starlink doubled to twelve million subscribers. These stories aren’t separate anymore. Each one leans on the others, and the question for the next hundred episodes is whether they become one company or stay three bets.

That’s what makes this the most exciting moment we’ve had to cover. Not because everything promised will arrive on time. Most of it won’t. But because for the first time, the announcements and the numbers are landing in the same place, on the same schedule, and the gap between them is closing where everyone can see it.

So this is our promise for the next hundred. Tesla. SpaceX. xAI. Neuralink. Whatever gets built next. Every headline gets a fair hearing. Every number gets the last word. And when we get something wrong, we’ll say so on the next episode.

And one small ask. Tell us which episode stuck with you, or which story you want us to dig into next. Send it to the address in the closing. We read every one.

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.

This podcast was developed with the help of using AI assistants, including the voice, and undergoes a detailed review during production. Corrections are posted to the episode page.