Tesla opens Cybercab rides in Austin with a 45-car fleet

Share
Tesla opens Cybercab rides in Austin with a 45-car fleet

Autonomy, trade policy, and a détente in Washington all moved on the same day. Here's what landed.

Tesla said Thursday it is offering Cybercab rides in "limited areas of Austin, Texas" — the first time the two-seater with no steering wheel and no pedals has carried the public. State records show Tesla had 420 autonomous vehicles registered in Texas as of Wednesday evening, 45 of them Cybercabs, against Waymo's 988 in the state. The launch has been oddly muted: as of Thursday evening Tesla had posted no live footage of its invite-only "Cybercab launch event" and there were no public remarks from Elon Musk or other executives, while investors and creators who were flown in said they had been told not to post. Tesla did not say whether it is charging fares, or whether anyone with the app in Austin can get a ride.

The numbers matter more than the theatrics. Federal rules cap how many vehicles without steering wheels or pedals a manufacturer can sell, though it can deploy unlimited numbers for testing — and testing is not the same business as collecting fares. That is the choke point, and it lands on a company whose $1.4 trillion market value rests substantially on robotaxi ambitions while its core EV business takes heavy pressure from Chinese rivals. Musk promised in July 2025 that the network would expand at a "hyper-exponential rate" and reach half the U.S. population by the end of that year; it did not, and executives have since struck a more cautious tone about avoiding injuries. A 45-car fleet in parts of one city is a real milestone for the hardware and a very small one for the business case.


Commerce Secretary Howard Lutnick says the next round of semiconductor tariffs will be "targeted" and tied directly to whether a company builds in the U.S. "If you produce in the U.S., you pay no tariffs. But if you don't produce here, be prepared to pay tariffs to enter the world's greatest market," he said, confirming reports that duties will be linked to domestic investment. The expansion is the part that should worry AI operators: the scope reportedly widens beyond chips to finished products containing them — laptops, game consoles, and data center servers — with country-specific rates and import quotas under discussion. Rates, timelines, and exemption criteria are all still unset.

Lutnick framed the pharmaceutical tariff model as the template and claimed the policy is already working, citing a $265 billion TSMC plant in Arizona and a $250 billion Micron memory fab for more than $500 billion combined, with total U.S. semiconductor investment commitments at $1.2 trillion. Read that as a negotiating position rather than an audited figure — the Commerce Secretary is selling the policy, and the January 25% tariff on certain advanced chips already carved out data center and R&D imports. If those carve-outs disappear in phase two, the cost of an American AI build-out rises at exactly the moment capital is pouring into it. Critics have warned for weeks that this is the likely outcome.


Lutnick also said Anthropic is "back on the right side" with the Trump administration, days after a federal judge ruled the Pentagon's blacklisting of the company was unlawful. The dispute began when Defense Secretary Pete Hegseth blocked Anthropic from certain military contracts after it refused to allow Claude to be used for domestic surveillance or autonomous weapons; Anthropic sued in California and won. Neither Anthropic nor the Commerce Department commented on Lutnick's remark. We covered the ruling when it landed — Judge strikes down Pentagon's blacklist of Anthropic.

A court victory and a political rapprochement are not the same thing, and only one of them is durable. The ruling established that an administration cannot punish a lab over its use restrictions by labeling it a supply-chain risk; Lutnick's comment signals the White House has decided to stop trying, at least for now. What it does not settle is the underlying question the case raised — whether a frontier lab can hold a red line on autonomy and surveillance and still keep government business. That gets renegotiated every time a contract is up.

What to watch: whether Tesla says anything concrete about fares and fleet size in the next few weeks — and whether the phase-two chip tariff carve-outs for data centers survive the drafting.

Which of these lands hardest on the AI industry in six months: a capped robotaxi fleet, tariffs on servers, or labs drawing red lines the government keeps testing? Tell us in the comments.

Read more

AI 101 — How do AI video generators work?

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings