Tesorb Signal: Tesla, SpaceX, and the Musk Companies

SpaceX targets May 19 for the debut of Starship Version 3, carrying three times the payload of its predecessor. We break down what changed, why it took seven months, and what a two trillion dollar IPO is betting on.

The biggest rocket ever built hasn't flown in seven months. The last time Starship left the ground was October 13, 2025, when Flight 11 closed out the V2 era with a perfect mission. Since then, silence. No launches. No catches. No explosions. Just construction.

This is the Tesorb Signal and I'm Lena Ruiz. Next Monday, SpaceX is targeting the first flight of Starship Version 3. New rocket. New engines. New launch pad. And roughly three times the payload capacity of everything that came before it. Today we're breaking down what changed, why it took seven months, and what this launch actually has to prove.

Let's start with the numbers, because they tell the story. Starship V2 could lift about 35 tons to low Earth orbit. Version 3 can carry over 100. That's not a 10 or 20 percent bump. That's a 3x leap in a single generation.

To put that in context, the Space Shuttle maxed out at about 24 tons. Saturn V carried around 130. Version 3 puts Starship in Saturn V territory, but with a vehicle designed to be fully reusable.

The rocket stands 408 feet tall, about four feet more than its predecessor. But height isn't the story. The real changes are under the skin. Version 3 runs on Raptor 3 engines, the third generation of SpaceX's methane-oxygen rocket engine. At sea level, each one produces about 250 tons of thrust, up from 230.

They're lighter, and they don't need individual heat shields anymore. SpaceX redesigned the cooling so each engine protects itself during reentry. Fewer parts. Less weight. More thrust.

The booster now flies with three grid fins instead of four, cutting weight and complexity without sacrificing control. The ship carries new propellant feed connections built for orbital refueling, something Starship will need for deep space missions. And this flight will deploy 22 dummy Starlink satellites, the most payload items Starship has ever carried.

But here's the thing. Flight 12 won't attempt a booster catch. With a new vehicle, new engines, and a new launch pad all flying together for the first time, SpaceX is being cautious. Both stages will splash down. The booster in the Gulf of Mexico. The ship in the Indian Ocean.

That new pad deserves a moment. Flight 12 launches from Pad 2, a redesigned site with shorter catching arms and faster propellant loading. Pad 1 hosted every Starship flight from the concrete-cratering first test through Flight 11's farewell. It's now retired in its current form.

So why seven months of quiet? Starship flew 11 times in roughly two and a half years. SpaceX caught boosters out of the sky. Ships survived reentry at orbital speeds. The pace was relentless. Then it all stopped.

The reason is simple. V2 did its job. It proved the architecture works. Rockets can be caught mid-air. Heat shields can hold through reentry. The concept is sound. But 35 tons to orbit is a test vehicle. It's not a business.

Version 3 is the Starship that has to do real work. Deploy full-size Starlink satellites in batches. Transfer propellant between vehicles in orbit, something that's never been done at this scale. Carry astronauts to the Moon for NASA's Artemis program. And if SpaceX has its way, put computing hardware into orbit. None of that works at 35 tons. All of it becomes possible at 100.

Building that kind of leap meant redesigning nearly everything. New engines. New pad. New plumbing for orbital refueling. And SpaceX did all of it while the clock ticked on NASA contracts and ahead of what could become the largest IPO in history.

Which brings us to what's riding on Monday evening in South Texas.

SpaceX filed for a public offering this summer with a target valuation near $2 trillion. That number isn't priced on Falcon 9 launches. It's priced on what Starship can become. A rocket that launches daily, delivers crew to the Moon, and eventually puts data centers in orbit.

We covered this a few episodes back, when SpaceXAI leased its entire Colossus supercomputer to Anthropic. And just today, reports broke that Google is in talks with SpaceX to launch orbital data centers under a project called Suncatcher. Two major AI companies in two weeks, both buying into SpaceX infrastructure.

But those deals assume a rocket that can haul 100 tons to orbit cheaply and repeatedly. That rocket is Version 3.

NASA is watching closely too. The agency pushed its first crewed Moon landing to late 2027, and it needs Starship as a lander. Blue Origin's Blue Moon is now waiting as a backup. SpaceX is no longer the only path to the surface.

So what's really riding on this launch? Starship V2 proved the biggest rocket ever built can fly. Version 3 has to prove it can fly for a living. That's the gap between a prototype and a product. And right now, a $2 trillion valuation, a lunar program, and the future of orbital computing are all waiting on a 90-minute launch window next Monday.

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.

—

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.

SpaceX just filed paperwork for a $55 billion chip factory in rural Texas, with a full buildout that could reach $119 billion. That's more than the entire CHIPS Act. We break down what the filing says, why the price quintupled from the original announcement, and what it means for Terafab's chances of actually getting built.

Back in our very first episode, we told you about Terafab. The price tag that day was twenty-five billion dollars. That was March. This week, SpaceX filed paperwork with a county in rural Texas. The new number? A hundred and nineteen billion.

This is the Tesorb Signal and I'm Lena Ruiz. Today we're digging into what just changed with Terafab, why the cost nearly quintupled, and what it tells us about whether this project is actually going to happen.

Let's start with what the filing actually says. On May sixth, a public hearing notice appeared on the Grimes County, Texas website.

SpaceX, not Tesla, is requesting a property tax abatement for what the filing calls a multi-phase, next-generation, vertically integrated semiconductor manufacturing facility.

Phase one alone is estimated at fifty-five billion dollars. The full buildout could reach a hundred and nineteen billion. Let that sink in. That's not a press release. That's a number attached to a government filing.

The proposed site is a place called Gibbons Creek, about ninety miles northeast of Austin. It used to be a coal-fired power plant that shut down in 2018. The land was demolished and remediated. And now it's being proposed as the home of what could become the most advanced chip factory on American soil. A coal plant becoming a chip fab. That's a detail worth sitting with.

The county commissioners will hold a public hearing on June third to consider the tax break. Residents in the area have already pushed back. They say they weren't told much about the project before the filing appeared.

But look at what happened in seven weeks. Think about the gap between the March announcement and this filing.

In March, Musk stood on stage at a decommissioned power plant in Austin and said Terafab would cost twenty to twenty-five billion dollars. Seven weeks later, the number in the actual government paperwork is more than double that for phase one alone. And nearly five times higher for the full project.

That sounds like scope creep. But it's actually something else entirely. It's what happens when a pitch becomes a permit application.

Announcements are aspirational. Filings are contractual. When you ask a county for a tax break, you have to put real numbers on paper. And the real numbers are always bigger than the ones you say on stage.

Here's some context for just how big this is. The entire CHIPS Act, the federal law Congress passed to rebuild American chip manufacturing, authorized $52.7 billion. That was for the entire country. Every company. Every fab. Every state.

SpaceX just filed for fifty-five billion for one facility in one county. One company is proposing to outspend a federal program that took two years of congressional debate to pass.

And they're not building it alone. In April, Intel signed on to help design, fabricate, and package chips at Terafab using its 14A process. That's the most advanced manufacturing technology Intel has. It's also the first time Intel's foundry business has taken on a major outside customer.

For Intel, this is a lifeline. For Musk, it's access to decades of fabrication expertise he doesn't have in-house.

Meanwhile, Tesla is building a smaller three-billion-dollar prototype fab at GigaTexas in Austin. Musk described it on the last earnings call as capable of maybe a few thousand wafers per month. Think of it as the test kitchen. The Grimes County facility is the production line.

The division of labor is clear. Tesla iterates on chip designs at the prototype line. SpaceX builds the factory that produces them at scale. Intel provides the manufacturing process that makes the transistors small enough to compete with the best fabs in the world.

So here's what the bigger numbers actually tell you.

Most megaprojects die in the gap between the announcement and the permit. That's the graveyard. Companies hold press conferences. They put out renderings. They make big promises. And then nothing gets filed. Nothing gets built.

The price tag on Terafab didn't go up because the project is falling apart. It went up because somebody actually did the engineering. They picked a site. They measured the land. They calculated the power needs. They submitted a filing to a real county government with real dollar figures attached.

That's not a red flag. That's a checkpoint.

Does a filing guarantee this gets built? No. A hundred and nineteen billion dollars is a staggering commitment. The project has no long-term chip supplier contracts locked in yet. Analysts at Bernstein have estimated the full one-terawatt vision could cost trillions.

And all of this is happening while SpaceX prepares for a June IPO that could value the company at $1.75 trillion. That's a lot of ambition on one balance sheet.

And this is the same Elon Musk who has missed timelines on everything from the Roadster to Full Self-Driving. The June third hearing in Grimes County is the next milestone to watch. But this week, Terafab moved from a stage presentation to a government document. And in the world of megaprojects, that's exactly where the real ones separate from the vapor.

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.

—

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.

SpaceXAI just gave Anthropic access to its Colossus supercomputer — 220,000 GPUs and 300 megawatts of compute. Anthropic immediately doubled Claude Code limits. We break down why Musk is renting compute to a rival, what it means for the SpaceX IPO, and the buried line about orbital AI data centers that nobody's talking about.

This is Tesorb Signal and I'm Lena Ruiz. Today we're unpacking the deal that just turned SpaceX into an AI landlord. We'll look at what Anthropic is actually getting, why SpaceXAI would arm a competitor, and a single sentence buried in the announcement that might matter more than anything else.

The most valuable asset in the AI race isn't the smartest model. It's the biggest power bill.

Quick background if you missed it. Elon Musk recently dissolved xAI, his standalone AI company, and folded it into SpaceX. The new division, SpaceXAI, now runs all of Musk's AI work under the SpaceX umbrella, including Grok.

So here's what happened. SpaceXAI announced that Anthropic signed an agreement to use Colossus 1, the supercomputer in Memphis, Tennessee that was built in just a hundred and twenty-two days. Over two hundred twenty thousand NVIDIA chips, including H100s, H200s, and next-generation GB200 accelerators.

Anthropic is getting access to the full facility. Three hundred megawatts of compute, coming online within the month.

The impact was immediate. Within hours, Anthropic doubled the usage limits on Claude Code for paying subscribers. Removed peak-hour restrictions entirely for Pro and Max users. And raised API rate limits for Claude Opus models by up to fifteen hundred percent on some tiers.

If you've been hitting the wall on Claude Code sessions, this deal is why that wall just moved.

Now, why would SpaceXAI do this? Grok and Claude are neck and neck. On the Chatbot Arena leaderboard, Grok trails Claude by single digits. These aren't partners. They're rivals. So renting your supercomputer to the competition sounds like handing over the playbook.

But zoom out and the logic clicks. SpaceXAI has built far more compute than any single AI lab needs right now. Colossus 1 has two hundred twenty thousand GPUs. Colossus 2, still ramping up, adds over half a million more. That's roughly eighteen billion dollars in NVIDIA hardware sitting in Memphis.

Idle GPUs don't generate revenue. They generate electricity bills.

So SpaceXAI is doing what Amazon did twenty years ago. Amazon built massive server infrastructure for its retail business, then realized it could rent spare capacity to the world. That side project became Amazon Web Services. Today AWS generates more profit than Amazon's entire retail operation.

SpaceXAI is running the same playbook. Renting Colossus capacity to Cursor for coding model training, and now to Anthropic for inference. They're not just an AI lab anymore. They're becoming a cloud provider.

And the timing matters. SpaceX is expected to go public as early as this summer at a valuation north of one and a half trillion dollars. A cloud computing revenue stream, on top of Starlink and launches, makes the IPO story a lot more compelling for investors.

And Anthropic? They need every megawatt they can find. Consider the spending spree. A hundred billion dollars committed to Amazon for five gigawatts of compute. A deal with Google and Broadcom for three and a half gigawatts of custom chip capacity starting in 2027.

Then add forty billion in new investment from Google. Thirty billion in Azure capacity through Microsoft and NVIDIA. And now three hundred megawatts from SpaceX's Colossus on top of all of it.

Anthropic's run-rate revenue jumped from nine billion at the end of 2025 to thirty billion today. Over a thousand business customers each spending more than a million a year. Demand is outrunning supply, and the bottleneck isn't talent or algorithms. It's power and chips.

This is the pattern we keep seeing. If you heard our episode on Terafab, you know the thesis. Vertical integration. Own the supply chain. If you caught our SpaceX IPO episode, you know the company needs multiple revenue streams for public investors. Starlink for internet. Colossus for compute. Launches for everything else.

Now, there's one more detail that deserves your attention. Buried at the very end of the announcement, SpaceXAI mentioned that Anthropic has expressed interest in partnering on multiple gigawatts of orbital AI compute capacity.

Think about that for a second. SpaceX filed an FCC application in January for up to one million orbital data center satellites. Solar-powered. Between five hundred and two thousand kilometers altitude. Connected to Starlink's laser mesh network.

The filing called it the first step toward a Kardashev Two civilization, one that harnesses the full power of the sun. That sounds like science fiction, and right now it mostly still is.

But the logic isn't crazy. Terrestrial data centers are already straining power grids. Some regions face multi-year waits just to get enough electricity connected. When a thirty-billion-dollar-a-year AI company signals interest in buying compute capacity in orbit, it tells you how serious the power bottleneck has become.

Here's what this deal really tells us. SpaceXAI didn't just sign a partnership. It opened a new line of business. Today it's renting a data center in Memphis. Tomorrow it could be renting server racks in orbit.

And the fact that Anthropic, backed by Amazon, Google, and Microsoft, still needed to knock on Elon Musk's door tells you everything about where this race is headed. It's not about who has the best algorithm. It's about who can keep the lights on.

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.

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 made $573 million selling products to SpaceX and xAI last year. We follow the money trail, shared factories, and converted stock to show why the Tesla-SpaceX merger isn't a future event — it's already underway.

This is the Tesorb Signal and I'm Lena Ruiz. Today we're looking at the Tesla SpaceX merger question. But not the way everyone else is covering it.

Forget the rumors. Forget the prediction markets. Look at what's already true. The money trail. The shared factories. The stock that quietly changed hands. And the obstacles that still stand in the way. The merger between Tesla and SpaceX isn't a future event. It's already underway. The only question is when they file the paperwork.

Let's start with the money, because the money doesn't lie. Tesla's amended 10-K filing, dropped just last week, revealed the full picture for the first time.

Tesla earned four hundred and thirty million dollars from xAI in 2025. Most of that was Megapack battery sales, powering xAI's massive data centers. On top of that, Tesla pulled in a hundred and forty-three million from SpaceX. Mostly Cybertruck sales.

How many Cybertrucks? SpaceX registered nearly thirteen hundred in a single quarter. That's roughly nineteen percent of all U.S. Cybertruck sales in Q4. One in five Cybertrucks sold in America went to another Musk company. Tesla's audit committee signed off on all of it as arm's-length transactions. But it's hard to call this normal.

And the money kept flowing into 2026. Tesla disclosed another seventy-eight million in xAI revenue through just the first two months of this year. That pace would put the 2026 total well above 2025.

But the biggest number hasn't landed yet. Two billion dollars. That's what Tesla invested in xAI last year. Then in February, SpaceX acquired xAI in the largest private merger in history. And Tesla's xAI shares? They converted into SpaceX stock.

Think about what that means. If you own Tesla today, you already own a small piece of SpaceX. It's buried in a filing. But it's real.

Now let's talk about the infrastructure, because the financial ties are just the surface. Underneath, these companies are building shared systems that would be very hard to untangle.

We covered Terafab back in episode one. A twenty-five billion dollar chip factory on the Giga Texas campus. What we didn't know then is that Terafab isn't just a Tesla project. It's a joint venture. Tesla. SpaceX. xAI. One facility. One chip roadmap.

Intel signed on as the manufacturing partner. The blueprint is one roof designing chips for electric vehicles, robots, satellites, and orbital data centers. Three companies. One semiconductor strategy.

And it goes deeper than chips. xAI's Grok chatbot is now integrated into Tesla vehicles. Tesla's Optimus robots are being trained on xAI models. SpaceX's Starlink could eventually power Tesla's robotaxi network with real-time updates beamed from orbit.

Each company is becoming a building block in a single system. You can't pull one out without breaking the others.

So if they're already this connected, why hasn't the formal merger happened? One word. Mechanism. You can't do a stock-for-stock merger when one company is private. There's no liquid shares to offer. The SpaceX IPO changes that.

Once SpaceX trades on the Nasdaq, the math becomes a spreadsheet exercise. You set a share exchange ratio, shareholders vote, and the deal closes. It's the same playbook every mega-merger in history has followed. Wedbush analyst Dan Ives puts the odds of a merger at eighty to ninety percent by the first half of 2027.

Not everyone agrees. Prediction markets give it roughly twenty-two percent odds by early 2027. Gary Black at the Future Fund warned it could cut Tesla's stock by twenty to twenty-five percent because of the valuation mismatch. Tesla trades at about a hundred times earnings. SpaceX is targeting a hundred times revenue. Merging those two creates serious dilution math.

And then there's the China problem. Tesla's Shanghai Gigafactory operates under Chinese government oversight. SpaceX carries classified U.S. defense contracts, including work on missile defense. Folding both into one entity would trigger national security reviews on two continents.

But here's what I keep coming back to. Look at the pattern. xAI merged into SpaceX in February. Tesla's two billion dollar investment converted into SpaceX shares. Terafab links their chip strategies under one roof.

Five hundred and seventy-three million dollars flowed between them in a single year. And two shell companies called K2 Merger Sub were incorporated in Nevada last January with SpaceX's CFO listed as an officer. Every single move points in one direction.

Put it all together and there's only one conclusion. The debate over whether Tesla and SpaceX will merge is the wrong question. They're already converging. Shared chips. Shared revenue. Shared AI models. Shared equity.

The regulatory hurdles are real. The valuation math is complicated. But the direction is unmistakable. The formal merger, when it comes, won't be a transformation. It'll be a recognition of what's already true. One man. One vision. And increasingly, one company.

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. Corrections are posted to the episode page. See you on the next one.

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.

SpaceX just secured a $60 billion option to buy AI coding startup Cursor. We break down the deal structure, why xAI couldn't build this in-house, and what the price tag reveals about the state of Grok.

And then, two weeks ago, SpaceX announced a sixty billion dollar deal for the right to buy an AI coding startup called Cursor. Sixty billion. To acquire a tool that does what their own AI was supposed to do.

This is Tesorb Signal and I'm Lena Ruiz. Today we're looking at the SpaceX Cursor deal and what it reveals about the state of xAI. We'll break down how the deal actually works, why SpaceX needs Cursor in the first place, and what sixty billion dollars tells you about Grok that no press release ever will.

When a company that owns two hundred thousand GPUs and its own AI lab pays sixty billion dollars for someone else's coding tool, that's not a partnership. That's a confession.

Let's start with the deal itself, because the structure matters. SpaceX didn't buy Cursor. Not yet. What they secured is an option. They can acquire Cursor for sixty billion dollars later this year, or they can pay ten billion for the work the two companies are doing together. Think of it like putting down a deposit on a house you're not sure you can close on yet.

The timing is deliberate. SpaceX is planning what could be the largest IPO in history this summer, targeting a one point seven five trillion dollar valuation. They want to close the Cursor deal after the listing so they can pay with publicly traded stock. And adding a two billion dollar revenue AI business to the story doesn't hurt the IPO narrative either.

Cursor was actually in the middle of raising two billion dollars at a fifty billion dollar valuation when SpaceX stepped in and preempted the whole thing. That tells you SpaceX saw the window closing and moved fast.

Which raises the obvious question. Why does the company that built Colossus, one of the largest AI supercomputers on the planet, need to spend sixty billion dollars on a coding startup founded by four MIT students in 2022?

The answer is painfully simple. Compute isn't the bottleneck. Product is.

Cursor is an AI-powered code editor built by a startup called Anysphere. It was founded in 2022 by four MIT students, and it's become the fastest-growing software product in recent memory. Over two million users, more than a million paying customers, and two billion dollars in annual revenue that doubled in just three months. Half the Fortune 500 uses it.

Here's the part that should sting if you're xAI. Cursor lets developers choose between Claude, ChatGPT, Gemini, and yes, even Grok for their AI coding. It's model agnostic. And most developers aren't choosing Grok.

Meanwhile, xAI has been struggling. If you listened to our episode on the co-founder exodus, you know that all eleven of xAI's original co-founders left after the SpaceX merger. More than eighty engineers followed. Musk himself said xAI was, quote, not built right the first time around, and is being rebuilt from the foundations up.

TechCrunch put it bluntly. Unlike Google's purchase of Windsurf, which was structured as a talent acquisition, SpaceX would keep the entire Cursor team intact because SpaceX, quote, lacks a meaningful AI workforce.

So what does this deal actually tell us? It tells us that in the AI race, having the biggest computer isn't the same as having the best product. SpaceX has Colossus. Two hundred thousand GPUs. Enough raw power to train frontier models. But raw power without the right team building the right product is just an expensive electricity bill.

And the competition isn't standing still. Anthropic has Claude Code. OpenAI has Codex. Both are growing fast, and both are eating into the AI coding market that SpaceX is now spending sixty billion dollars to enter. That's the race Cursor was already running before SpaceX showed up.

The Cursor deal is Musk acknowledging that gap with his checkbook. And the price tag tells you how wide the gap is. Sixty billion dollars is more than Ford is worth. It's more than General Motors. That's what it costs to buy your way back into a race you were supposed to be leading.

And this connects to something even bigger. Just last week, Musk was on the stand in federal court and was asked to rank the world's leading AI companies. He put Anthropic first, then OpenAI, then Google, then Chinese open source models. He described xAI as a much smaller company with just a few hundred employees. Under oath, the founder of xAI ranked his own company last.

So strip it all away, and here's what's left. The SpaceX Cursor deal isn't really about Cursor. It's about what xAI couldn't become on its own. Two hundred thousand GPUs. Billions in funding. The world's richest person behind it. And it still wasn't enough.

Because in AI, compute is necessary. But it's not sufficient. The product has to work. The team has to stay. And right now, SpaceX is writing a sixty billion dollar check to solve for both.

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.

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 June 2012, the first Tesla Model S rolled off the line at Fremont. Fourteen years later, the last one has too. We break down why Tesla is retiring the cars that built its reputation, what's replacing them on the factory floor, and what it tells us about the company Tesla is becoming.

This is the Tesorb Signal and I'm Lena Ruiz. Today, we're talking about the end of the Model S and Model X. Why Tesla is retiring the cars that built its reputation, what's taking their place, and what it tells us about the company Tesla is becoming.

Let's start with what the Model S actually accomplished. Because it's easy to forget just how radical this car was when it launched. In 2013, Motor Trend named it Car of the Year. That was the first time an electric vehicle had ever won the award.

By 2015, the Model S was the best-selling plug-in electric car on the planet. Not the best-selling EV from a startup. The best-selling plug-in, period. It outsold everything.

And it wasn't just a sales success. The Model S forced every major automaker to take electrification seriously. Billions of dollars in competitor investment trace directly back to a sedan that proved you didn't have to sacrifice range, performance, or luxury to drive electric.

The Model X followed in 2015 with those falcon-wing doors and a design that pushed the boundaries of what an electric SUV could be. Together, these two vehicles built Tesla's brand as a premium automaker.

But then came the Model 3. When the more affordable sedan arrived in 2017, it upended everything. Suddenly Tesla was selling hundreds of thousands of cars a year, and almost all of them were Model 3s and later Model Ys.

The flagship models didn't just take a backseat. They were quietly forgotten. Tesla stopped breaking out individual Model S and X sales numbers in 2023. They lumped them into an "other models" category alongside the Cybertruck and the Semi.

By 2025, Tesla was selling roughly 30,000 Model S and X units a year on a production line built to handle 100,000. That's a 70% utilization gap. The cars that built Tesla had become an afterthought.

So on January 28th of this year, Elon Musk made it official. On Tesla's Q4 2025 earnings call, he said it was time to give the Model S and X an "honorable discharge." He told anyone interested to order one while they still could.

Production wound down through the first quarter. On April 1st, Musk confirmed that custom orders were closed and production was finished. He posted a throwback photo from the original 2012 launch and wrote, "I love those cars."

As of right now, there are roughly 600 Model S and X vehicles left in global inventory. When they're gone, they're gone.

But Tesla isn't letting them go quietly. The company announced a Signature Series, a final production run of 350 vehicles. 250 Model S units and 100 Model X units. All Plaid models, invite-only.

Each one is finished in an exclusive Garnet Red paint with gold accents throughout, and individually numbered. The price? $159,420. That's more than $30,000 above the standard Plaid.

On May 12th, Tesla is holding a sunset delivery event at the Fremont factory. The same factory where it all started. Invite-only. 350 buyers picking up the last Model S and Model X vehicles that will ever be built.

Now, here's where this story becomes about more than nostalgia. Because the question isn't just what Tesla is retiring. It's what's taking the place of these cars on the factory floor.

If you listened to our episode on Optimus, you know Tesla has massive ambitions for humanoid robots. Well, now we know exactly where they plan to build them.

Musk said the Fremont production line is being converted into an Optimus factory with a long-term target of one million robots per year.

It's like watching a legendary theater get demolished to build a warehouse. Except in this case, Tesla thinks the warehouse is worth more than the theater ever was.

Tesla isn't replacing the Model S with a next-generation luxury sedan. It's not building an answer to the Mercedes EQS or the Porsche Taycan. It's building robots. The same assembly line where workers once bolted together the car that won Motor Trend's Car of the Year will soon be producing humanoid machines.

That's not a product decision. That's an identity statement. Tesla is telling you, in the clearest possible terms, that it sees its future in autonomy and robotics, not in selling premium cars. When the Signature Edition buyers drive off the lot on May 12th, Tesla's consumer lineup drops to three. The Model 3. The Model Y. And the Cybertruck.

No flagship sedan. No luxury SUV. The company that proved electric cars could be aspirational is walking away from the segment it created.

Whether that's visionary or reckless depends on whether you believe Optimus and the robotaxi program will deliver the revenue that the Model S and X no longer can. What's not debatable is that Tesla just made the biggest strategic bet in its history. And it made it by saying goodbye to the cars that made everything else possible.

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.

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.

SpaceX's S-1 filing reveals Starlink generated $11.4 billion in revenue last year, making it the financial engine behind everything Elon Musk builds in space. But one uncomfortable number shows where all that money is going.

This is the Tesorb Signal and I'm Lena Ruiz. Today we're looking at what SpaceX's IPO filing just revealed about Starlink, why it's become the financial engine behind everything Elon Musk builds in space, and what one uncomfortable number in the S-1 tells us about where the money's actually going.

On April first, SpaceX confidentially filed its S-1 with the SEC. That's the document every company has to submit before going public. And when details started leaking out, one number jumped off the page.

Starlink generated eleven point four billion dollars in revenue in 2025. And four point four billion of that was operating profit. To put that in context, that's more annual profit than companies like Spotify or Snap generate in total revenue.

This wasn't a side project propping up the rocket business. It was the other way around. Starlink accounted for more than sixty percent of SpaceX's total revenue last year. The rockets, the NASA contracts, the defense work, all of that combined didn't match what the satellite internet business brought in.

And the growth behind those numbers is unlike anything in telecom history. Starlink had one million subscribers in late 2022. As of February this year, it crossed ten million. That's a tenfold increase in just over three years.

That's the kind of growth curve you'd expect from a social media app. Not a broadband provider that asks you to bolt a dish to your roof.

But Starlink isn't just dishes anymore. The company rebranded its phone-based satellite service as Starlink Mobile. It works with your existing phone. No special hardware. T-Mobile launched it commercially last July, and more than twelve million people already rely on it for connectivity where no cell tower reaches.

Now, here's where the scale gets hard to picture. Over ten thousand active Starlink satellites orbit Earth right now. That's roughly two out of every three operational satellites in space. SpaceX has already launched fifty missions this year, and forty-two of them were Starlink flights.

The aviation side is moving just as fast. British Airways started flying with Starlink in March. Lufthansa Group signed a deal to equip all eight hundred and fifty of its aircraft. A half dozen other major airlines have followed. They're replacing their old wifi systems and offering Starlink for free as a competitive edge.

And managing that infrastructure means operating at a scale nobody else has attempted. Earlier this year, SpaceX announced it's lowering forty-four hundred satellites from five hundred and fifty kilometers down to four hundred and eighty kilometers.

That's half the constellation, repositioned to reduce collision risk and speed up deorbit times for any satellite that fails. It's the largest orbital reconfiguration ever attempted, and they're doing it while launching new satellites every few days.

But here's the part that doesn't add up. If Starlink is this profitable, why did SpaceX actually lose money last year?

In February, SpaceX completed its acquisition of xAI, Musk's artificial intelligence company. And the S-1 reveals what that merger actually cost.

xAI posted a six point four billion dollar operating loss in 2025. That's four times what it lost the year before. Its capital spending alone hit twelve point seven billion dollars. That's more than SpaceX spent on rockets and satellites combined.

Starlink's four point four billion in profit didn't just get absorbed. It got erased. SpaceX ended the year with an overall loss of four point nine billion dollars.

Think of it like this. Starlink is the reliable engine generating cash every month. And right now, nearly all of that cash is being fed into an AI bet that hasn't proven itself commercially.

The S-1 claims a total addressable market of twenty-eight point five trillion dollars, with over ninety percent of that coming from AI. That's an enormous promise built on a business that's currently burning about a billion dollars a month.

There's one more tension worth watching. As Starlink's subscribers quadrupled, its average revenue per user dropped eighteen percent, to about eighty-one dollars a month. International pricing as low as forty-five dollars a month is fueling growth.

But that growth is diluting the value of each customer. The question is whether sheer volume can outrun that math.

SpaceX is betting it can. Analysts at Quilty Space project Starlink could reach nearly seventeen million subscribers by the end of this year. We talked about the SpaceX IPO back in episode four, when it was still speculation. Now it's real.

The roadshow is expected the week of June eighth, with a valuation target between one point seven five and two trillion dollars. If that holds, it would be the largest IPO in history.

So when you cut through the S-1, one number tells the whole story. Starlink isn't just SpaceX's best product. It's the financial foundation holding up a trillion-dollar ambition.

The rockets get the headlines. The AI gets the hype. But the satellite internet service beaming broadband from ten thousand spacecraft? That's what pays for everything.

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.

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.

It costs fifteen cents to move a fully loaded Tesla Semi one mile. A diesel truck? About fifty cents. We break down the math behind the Semi's cost advantage, real-world fleet data, and what it means for the trucking industry.

This is Tesorb Signal and I'm Lena Ruiz. Yesterday, Tesla rolled the first Semi off its new high-volume production line in Nevada. After nine years of delays, the truck that nobody believed would ship is finally here. But today I don't want to talk about the truck itself. I want to talk about the math behind it. Because the Semi's real weapon isn't its electric motor. It's a spreadsheet.

Let's start with the number that's turning heads on Wall Street. This spring, investment bank Bernstein published an analysis comparing the Tesla Semi to the Freightliner Cascadia, which is the most popular diesel truck in America. Their conclusion? The Semi now holds a three percent total cost of ownership advantage.

Three percent might sound small. It's not. Trucking is a business where full-year profit margins typically run between two and five percent. A three percent structural cost edge, compounding across an entire fleet over hundreds of thousands of miles, isn't a nice-to-have. It's a competitive moat.

Here's where the math comes from. The Semi runs at about fifteen to twenty-five cents per mile on energy. Diesel rigs run fifty to seventy cents. In California, diesel just crossed seven dollars a gallon, which makes that gap even wider. On a truck logging two hundred thousand miles a year, the fuel savings alone approach seventy thousand dollars.

And then there's the stuff fleet managers don't usually talk about in public. No oil changes. No diesel particulate filter rebuilds. No turbocharger failures. No exhaust fluid top-offs. The Semi doesn't just cost less to fuel. It eliminates entire categories of maintenance that diesel operators budget for every quarter.

Now let's talk about the sticker price, because that's the first objection everyone raises. The Semi lists at around two hundred and ninety thousand dollars. A comparable diesel truck costs about a hundred and fifty thousand. So yes, it's roughly double upfront.

But in California, fleet operators can qualify for vouchers up to three hundred and fifty thousand dollars per truck. That can bring the out-of-pocket cost below two hundred thousand, narrowing the gap against a new diesel to about twenty-five thousand dollars before a single mile is driven.

Now, spreadsheet projections are one thing. Real-world data is another. So let's look at what the trucks on the road are actually doing.

A few hundred Tesla Semis have been running freight routes across the country. They've collectively logged over thirteen million miles. And the numbers are holding up. Mone Transport, a Texas-based carrier, tested the Semi on active freight lanes and recorded energy consumption of 1.64 kilowatt hours per mile. That actually beats Tesla's own target of 1.7.

DHL ran a fully loaded Semi on a 390-mile route and got 1.72. ArcBest pushed one through a 7,200-foot climb over Donner Pass in the Sierra Nevada and still hit 1.55. These aren't cherry-picked demo runs. These are real trucks hauling real freight for companies that measure everything in pennies per mile.

And here's the reliability number that might matter most. Tesla's Semi fleet is reporting ninety-five percent uptime. When breakdowns do happen, seventy-five to eighty percent are resolved within twenty-four hours. Nearly half are fixed in under an hour. For fleet operators, uptime is revenue. Every day a truck sits in a shop is a day it's not earning.

If you listened to our very first episode on Tesla's Terafab chip factory, the pattern here might feel familiar. Tesla didn't just build a truck. They built the factory right next to the battery plant. The 4680 cells that power the Semi are manufactured in the same complex in Sparks, Nevada. That vertical integration is the same playbook.

But let's be clear-eyed about what's still missing. The Megacharger network is in its infancy. Tesla has one public station open in Ontario, California. They've mapped sixty-six locations across fifteen states, and they've signed a deal with Pilot Flying J, the nation's largest truck stop operator. But mapping a location and opening a charger are two very different things.

And there's the ramp itself. The factory is designed for fifty thousand trucks a year. Analysts think deliveries in 2026 will land somewhere between five and fifteen thousand. Tesla has a long history of ambitious production targets that take longer than promised. Just ask anyone still waiting for the Roadster we covered a few episodes back.

The demand signal, though, is hard to ignore. In California's Clean Truck voucher program, the Tesla Semi accounted for 965 out of 1,067 applications. Daimler, PACCAR, and Volvo combined? Fewer than a hundred. The fleets have done the math. Now it's Tesla's job to deliver.

Strip away the slogans, and here's what's left. The Tesla Semi's pitch isn't about saving the planet. It's about saving money. Fifteen cents a mile versus fifty. A three percent cost advantage in an industry that survives on two to five. That's not a green energy talking point. That's a profit margin.

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.

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.

Every original co-founder of Elon Musk's AI company xAI has left. Now SpaceX operators are running the rebuild, with a $2 trillion IPO on the line. We break down what happened and what it means.

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 an Oakland courtroom, the world's richest man took the witness stand and told a jury that artificial intelligence could kill us all. We break down the Musk versus OpenAI trial that just kicked off in federal court — why OpenAI was built as a nonprofit, why that model couldn't survive contact with reality, and what this trial could actually change about who controls AI.

This is Tesorb Signal and I'm Lena Ruiz. Today we're breaking down the Musk versus OpenAI trial that just kicked off in federal court.

Here's our roadmap. First, why OpenAI was built as a nonprofit in the first place. Second, why that model couldn't survive contact with reality. And third, what this trial could actually change about who controls AI. Here's the one thing I want you to hold onto. This case is the first real legal test of whether artificial intelligence can be developed outside of corporate control.

To understand this trial, you have to go back to 2015. Google had everything. The money. The talent. The computing power. After acquiring DeepMind, Google was pulling ahead in the AI race with almost no competition.

That worried Musk. He testified today that the whole idea for OpenAI started with an argument with Google co-founder Larry Page. Musk said Page accused him of caring more about human survival than the future of AI.

It sounds dramatic, but it shook him. Musk told the jury he started talking to everyone he knew about the dangers of AI. His brother eventually told him to stop bringing it up at parties because it was, quote, a buzzkill.

That's what led him to team up with Sam Altman. Together they put in about thirty-eight million dollars to launch OpenAI as a nonprofit. The structure was the point. No single company. No single billionaire. The technology would be open and developed for the public good.

Musk was clear on the stand today. He said he came up with the idea, the name, recruited the key people, and provided all the initial funding. And he chose to make it a nonprofit specifically to benefit humanity.

That was the promise. And for about three years, it held.

Now here's where the story turns. Building frontier AI turned out to be astronomically expensive. The computing costs alone run into billions of dollars per year. Training a single large model can cost hundreds of millions. Musk's thirty-eight million was a start, but it wasn't nearly enough to keep up with Google's budget.

Musk left OpenAI's board in 2018 after a power struggle over who would run the company. He says the nonprofit mission was still intact when he walked away. What happened next is what this trial is really about.

In 2019, OpenAI created a for-profit subsidiary to raise capital. Microsoft came in with billions in investment. By 2025, the company had restructured again into a public benefit corporation. The nonprofit foundation kept a controlling twenty-six percent stake. Microsoft held twenty-seven percent.

Today, OpenAI is valued at eight hundred and fifty-two billion dollars. It's planning a trillion-dollar IPO later this year. The charity Musk seeded with thirty-eight million now sits on a stake worth over a hundred and thirty billion.

Musk's argument is simple. They stole a charity. His attorney told the jury today that the for-profit side looted the nonprofit. He compared it to a museum gift shop stealing the Picassos.

OpenAI's response is just as direct. Their lead attorney told the jury that Musk quit, they succeeded without him, and now he's jealous. They point out he went on to launch his own for-profit AI company, xAI, and argue this lawsuit is really about hobbling a competitor.

Now let's talk about what's actually at stake. Of the original twenty-six claims Musk filed, only two survived to trial. Breach of charitable trust and unjust enrichment. Punitive damages were thrown out back in March.

Musk wants three things. Revert OpenAI back to a nonprofit. Remove Altman and president Greg Brockman from the board. And return roughly a hundred and thirty billion dollars in what he calls ill-gotten gains to the nonprofit foundation.

The jury here is advisory. Judge Yvonne Gonzalez Rogers makes the final call. And there are really three ways this could play out.

First, a settlement. Both sides have reasons to deal. Musk gets symbolic governance reforms. OpenAI limits the damage to its IPO timeline. Most legal observers think this is the likeliest outcome.

Second, a partial Musk win. The jury finds unjust enrichment, the judge orders structural changes. Maybe a bigger nonprofit stake or new governance rules. The IPO gets delayed but not killed.

Third, a full OpenAI win. The jury agrees Musk left voluntarily, that the statute of limitations has passed, and that his real motivation is competitive. Nothing blocks a trillion-dollar public offering.

This trial is expected to run about four weeks, with testimony from Altman, Brockman, and Microsoft CEO Satya Nadella still to come. Deliberations could begin by mid-May.

But here's the question that goes beyond this courtroom. If a nonprofit can't fund frontier AI because the costs are too high, and a for-profit won't prioritize the public good because shareholders come first, then who gets to decide what this technology becomes?

That's the question this trial is forcing into the open. And no jury verdict is going to settle it.

Here's what to take away from today. This trial isn't just two billionaires arguing over a company. It's a test case for whether AI development can exist outside corporate control. The answer so far is no. And that's worth paying attention to, no matter who wins in Oakland.

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.

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.