Tesla's first production Cybercab drives itself in Austin

Kind: captions Language: en Tesla posted footage today of its first production cyber cab, the purpose-built robo-taxi with no steering wheel and no pedals, driving itself through Austin traffic. And the head of its self-driving software confirmed the car is, in his words, driving by itself. We've also got Tesla's first senior hire for its own chip factory, the first Optimus production line landing in Fremont, a striking endorsement from Nvidia's chief executive, and a look at the delivery number due Thursday. This is the Tesla brief after the close on June 30th. Welcome to the show. A quick look at the stock. Tesla closed up 2.8% at $420.60. The Nasdaq rose 1 and 1/2%. And the S&P 500 gained 8/10 of a percent. It was a strong day across the market, but Tesla outran it by a wide margin. And with production cyber cab footage and that Nvidia endorsement both landing today, there was real company news behind the move, not just the rising tide. Let's start with the cyber cab. The cyber cab is Tesla's purpose-built robo-taxi, a two-seater with no steering wheel and no pedals, designed from scratch to have no human driver at all. It's meant to be the vehicle at the center of Tesla's robo-taxi network. For months, the story has been the same. We've seen pre-production units filmed on closed courses and quiet streets, but no clear sign that a real, factory-built car was out driving itself in traffic. That changed today. Tesla's own account posted footage of a production cyber cab running engineering tests on public roads in Austin. And Ashok Elluswamy, Tesla's head of autopilot software, confirmed on X that the car is definitely driving by itself. Photos from the street show it. No wheel, no pedals, moving through Austin traffic under its own control. This is Tesla, on its own channel, saying the missing piece is now real. The context came from a long-form interview released the same day with Lars Moravy, Tesla's VP of vehicle engineering. He said cyber cab production is right where Tesla expected it to be and that the automation is ahead of schedule actually. He put the current build count at roughly 150 cars at the Texas factory. Said the assembly lines are more than 90% automated. And confirmed most of that early fleet has gone into collecting training data for the self-driving software. That last detail matters and it's worth slowing down on. The Cybercab software is not a simple copy of what runs on a Model Y today. The car itself is different. Different shape, different cameras, different everything. So the neural network has to be retrained on real-world data from this specific vehicle before it's ready to scale. That's why you build 150 cars and send them out to drive. And now the software needs the miles to catch up. And Moravi flagged a specific date. A major announcement about scaling is coming a week from Tuesday on July 7th. There's a smaller related sighting worth mentioning. Tesla was spotted in Austin with a Cybercab wrapped in a red, white, and blue livery reading "Made in Texas, land of the free, home of the brave." Complete with cartoon eyes and a mouth painted on the front. Tesla has never wrapped a Cybercab like this before and it lands just ahead of the 250th US Independence Day. Put it together with the July 7th announcement and it looks like Tesla is staging a holiday weekend moment. Maybe public rides, maybe a live stream, maybe new city coverage. One honest word of caution. Tesla has telegraphed events like this before that turned out to be more show than substance. So weigh it accordingly until the real thing lands. But step back and the significance is clear. For a year, the fair pushback on the robotaxi story was that we kept seeing prototypes and promises. Today there's a factory-built car with no steering wheel driving itself on a public road on Tesla's own account with a scaling date on the calendar. That's the concrete evidence the near-term robotaxi catalyst was missing. And it landed alongside an unusually strong outside endorsement. Elon amplified a clip of Nvidia's executive, Jensen Huang, saying on camera that Tesla's stack is the most advanced autonomous vehicle stack in the world, and that he's fairly certain they were already using end-to-end AI. Two things make that worth more than a typical compliment. First, Nvidia doesn't sell Tesla the chips inside its cars. Tesla designs its own. So, this isn't a supplier flattering a customer. It's the most watched figure in AI infrastructure handing the top ranking to a company that competes with his other self-driving customers, like Waymo, who do buy Nvidia's hardware. Second, that phrase end-to-end AI is the whole technical argument. It means one single neural network that goes straight from what the cameras see to how the car steers, instead of a stack of separate handwritten rules bolted together. That approach is exactly what Tesla says sets it apart, and having Huang endorse the framing carries real weight in the ongoing debate about whether Tesla's autonomy lead is genuine or just narrative. Away from the road, Tesla made its first known senior hire for Terafab, the in-house chip factory Elon unveiled back in March. The hire is Gary Jain, who spent nearly 18 years at Intel, and most recently managed the transfer of Intel's cutting-edge manufacturing process, the one it calls 18A. His new title, on his own LinkedIn, is director of Terafab at Tesla. Here's what Terafab is trying to do. Tesla wants to bring chip design, chip making, memory, packaging, and testing all under one roof. The point is to collapse the loop between we want to change something and we have the new chip in hand from months down to days. When you design your own chips and make them in the same building, you can iterate on your self-driving computers, your Optimus robots, and your cybercabs far faster than if you're waiting in line at an outside foundry. A regulatory filing in May put the initial commitment at around $55 billion with total potential investment as high as $119 billion, all driven by Tesla's own hunger for chips. Why it matters today? For months, Terafab was a Musk announcement with no one named to run it. Putting an actual Intel manufacturing veteran with hands-on experience on Intel's most advanced process in the director's chair is the first concrete sign Tesla is staffing this as a real factory rather than a slide. And it ties directly to the AI and compute pillar of the whole Tesla story. Sticking with manufacturing, Murati shared one more thing in that same interview. Tesla's first Optimus production line has arrived at the Fremont factory and installation has started. He described it as a modular line, meaning Tesla can swap out pieces of it as the robot's design keeps changing and said there will be dozens of lines over time. Two things make this worth surfacing. First, until now the Optimus production timeline was just a quote, Elon saying on the last earnings call that production would begin late July or August. A Tesla executive confirming the physical line is on the floor and being installed puts a real object behind that schedule. Second, that word modular is a deliberate choice. It's exactly how Tesla ran the Model 3 and Model Y lines, letting them make rolling changes to the product without shutting the line down. It signals Tesla still expects the Optimus design to evolve after production starts, not before, which is a realistic way to approach a first-of-its-kind product. Now to the number everyone's actually watching this week. Gary Black, who runs the Future Fund, put his model for Tesla's second quarter deliveries at roughly 410,000 vehicles, about 7% growth from the same quarter last year and ahead of Wall Street's consensus of 406,000. He credits the beat to gasoline prices surging during the quarter, which tends to push buyers back toward electric cars. If the number lands there, it would be Tesla's second straight quarter of year-on-year growth above 5% after two straight years of declines. But here's the tension worth naming out loud. The same analyst modeling a delivery beat this week is telling investors he doesn't own the stock. He points to Tesla trading at around 200 times its expected earnings, very expensive by any normal measure. And he cites rising competition in driverless rides from Google, Baidu, WeRide, Pony, and Amazon's Zoox, which he says are collectively running about a million paid rides a week now. Both things can be true at once. A strong delivery quarter and a cautious view further out. And both will shape how Thursday's number gets read. One more from the analyst desk, and it's about the other Musk company. Dan Ives at Wedbush went on CNBC and reframed SpaceX for the market, calling it much more of an AI play, well-positioned to become major hyperscaler. A hyperscaler is one of the handful of companies that runs the world's biggest AI data centers. Think Microsoft, Amazon, or Google. That framing came with a fresh $190 12-month target on SpaceX. Why should a Tesla investor care? Ives is the same analyst who's been putting odds on a Tesla-SpaceX merger. If he's now telling big institutional clients that SpaceX belongs in the AI infrastructure group, rather than the aerospace group, that changes how a merger, or even just how Musk splits his attention and capital between the two companies, should be valued. It also feeds a debate we keep coming back to. Is Tesla still the AI company in Musk's portfolio, or has SpaceX quietly become the main AI vehicle? Ives has now planted his flag firmly on one side. A quicker one before we look ahead. Tesla's AI account promoted an analysis showing that the newest self-driving software, FSD version 14 light, is now running on HW3, the self-driving computer Tesla shipped 7 years ago. The gap being closed is real. That older computer has roughly 1/8 the memory bandwidth of the current one, which is three to five times faster on the work that matters for self-driving. Getting the newest software to run acceptably on the older, weaker hardware is a genuine engineering feat. And it matters to the business in two ways. It protects the resale value of the millions of older Teslas already on the road. And it answers the loudest complaint from those owners who'd been told they were left behind. The fact itself surfaced yesterday. The news today is Tesla's own account choosing to stand behind it publicly. Which brings us to Thursday and it's shaping up to be one of the biggest sessions of the quarter for the stock. Two things land the same morning. First, Tesla reports its second quarter delivery and production numbers. The question that print resolves is simple but important. Was the growth Tesla posted in the first quarter the start of a real recovery from two years of decline or a one-quarter bounce? A number above 410,000 hands the bulls a clean two-quarter recovery story going into the late July earnings call. A number below 400,000 puts the demand fatigue argument right back at the top of the pile. And on the very same morning at 8:30 Eastern, the US releases its June jobs report. That one matters to Tesla for a specific reason. The new Fed chair, Kevin Warsh, has already signaled he's leaning toward raising rates rather than cutting. And Tesla's price leans heavily on businesses that pay off years from now. Robotaxi, Optimus, self-driving. So, higher rates weigh on it more than on most stocks. A hot jobs number reinforces that path toward keeping interest rates higher. A soft one reopens the conversation about a cut this fall. Two of the biggest swing factors for the stock colliding in the same hour. One scheduling note. US markets are closed Friday for the observed Independence Day holiday, so there's no episode Friday. That's the Tesla brief for today. If it's been useful, please subscribe. Thanks for joining us and we'll see you in the next episode.