Tech
OpenAI is scared of open-weight models. Should the US be?
The impressive capabilities of Chinese lab Moonshot’s Kimi K3, the biggest open-weight large language model, has kicked off a debate that conflates two things: the economic possibilities of American AI giants and the future of LLMs as a technology.
OpenAI’s head of strategic futures, Dean W. Ball, went so far as to argue that the US government should find a pretext to create regulatory fear, uncertainty, and distrust around the new models, since open-weight models must necessarily deter capital spending by the frontier labs.
People freaked out, with tech luminaries like Yann LeCun and Martin Casado arguing that open software can accelerate innovation and coexist with proprietary projects. Ball soon retracted his claims that a regulatory crackdown was the White House’s “best strategy” and that open-weight models necessarily slow down advances in the technology.
However, Axios reports that the Trump administration is considering banning K3 and other advanced Chinese models at the behest of American frontier labs. Another report from Politico said that the Department of Commerce would not take that step anytime soon.
The benefit for major AI companies is clear: Open-weight models, running on independent infrastructure or inside major enterprises, offers cheaper intelligence than Anthropic or OpenAI’s class-leading models. If users increasingly spend more outside the closed labs, that means smaller return on their massive investments in model training.
That view extends far beyond OpenAI. “Strong, frontier-caliber open source models will place a squeeze on the margins and will bring down the prices of the frontier companies,” Braden Hancock, the co-founder of Snorkel AI and a research partner at the Laude Institute, told TechCrunch. “It will not necessarily mean that the amount of AI usage goes down a little bit. You know, obviously, quite the opposite.”
That’s not a problem for people without shares in Anthropic and OpenAI. AI will still proliferate. So what’s the justification for the government to block Americans from purchasing something in our ostensibly free markets?
Concerns over Chinese models come in several flavors. One is protecting US data from the Chinese government; the US banned the import of modern Chinese EVs over concerns about their data gathering. But experts tend to think that open-weight models run on US servers are unlikely to leak data back to China, although it’s not impossible that such a thing could be done.
Another is that the models may have implicit bias toward the PRC — but it’s not clear what that might mean for, say, coding tasks.
A third common worry is that Chinese models lack the guardrails that the US government has mandated (through an opaque process), which aim to prevent leading US LLMs from being used to exploit closed computer systems or create weapons. However, those same guardrails may make US companies more vulnerable: David Sacks, the venture capitalist and Trump adviser, has been sharing cases of US companies turning to Chinese LLMs to close security gaps when US frontier models refuse to do the tasks.
But the most significant motivation for restricting the models is that fear that China will be able to outpace the US if the frontier labs slow down.
Sam Bresnick, a China-focused research fellow at Georgetown’s Center for Security and Emerging Technology, says the growing importance of AI to the US military operations gives the US a reason to support continued investment in AI at the frontier labs. But the whole question, he says, is fraught.
“Why should the weight of the U.S. government be aimed at protecting these these companies from competitors that are being locked out from the U.S. market based on their origins?” Bresnick asks.
Advocates for open AI say that the frontier companies are creating a false binary between innovation and closed models.
“The bigger impact of having these open source models come from China is less that they’re sneaking in back doors, and more that they are owning the innovation,” Hancock told TechCrunch. “You end up with, effectively, an expanded workforce on your model. PyTorch became the industry standard because it was open source, and so the whole community could contribute to it rather than just one company, and it grew and grew, and all the rest of the deep learning libraries kind of died in comparison.”
Hancock and other advocates fear that Chinese LLMs will become the locus of international research. Already, US graduate programs mainly build on open-weight Chinese models, and Hancock says that half of the papers students study are coming from Chinese institutions, with American frontier labs increasingly reticent about sharing their work widely.
“Restricting open models wouldn’t make AI safer,” said Clem Delangue, the CEO of Hugging Face, a platform for open AI collaboration. “It would simply hide the risks, concentrate power in the hands of a few and make it harder for the next generation of builders, researchers, academia, non-profits, governments to participate in making AI safer and more beneficial for all.”
Bresnick says that the real way to slow China would be to focus more on chip export controls. A better way to preserve US AI leadership would be to stop selling Nvidia H200 processors to China. “That,” he says, “could potentially keep us out of this thorny debate about banning open source technologies that huge numbers of US companies want to use.”
Part of the problem is that uncertainty around AI economics. “The open business model, the proprietary business model — neither one is figured out. AI companies are struggling to figure out how to make money on their tools, especially as training costs need to go up and up,” Bresnick points out.
The same challenges that play out in the US are also playing out in China, where AI companies are also struggling to generate revenue and access compute power, and the government is seen as encouraging open releases for policy reasons despite the challenge in capitalizing on them.
Some US companies, including Thinking Machines Lab and Nvidia, are trying to make a business around releasing open models. Hancock points out that Nvidia would do better “if there are dozens or hundreds of companies building AI than rather than two or three that are well capitalized enough to make their own chips,” which is one reason behind its investment in Nemotron, a collection of open models.
“The main point is the U.S. would be very well served to have its own very capable, much less expensive open models,” Bresnick said. “It just clashes with the approach the frontier labs have taken.”
With additional reporting from Rebecca Bellan.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Tech
Can an Apple lawsuit derail OpenAI’s hardware plans?
Apple recently filed a trade secrets lawsuit against OpenAI, accusing the AI company of a pattern of misconduct aimed at getting current and former Apple employees to share confidential information. (In response, OpenAI said it is “not aware of any evidence that this complaint has merit.”)
On the latest episode of TechCrunch’s Equity podcast, Kirsten Korosec, Sean O’Kane, and I debated whether this lawsuit will cast a shadow over OpenAI’s much-discussed plans to get into the hardware business (starting with a mobile smart speaker) and go public.
“Even setting aside whether or not the court grants any kind of injunctive relief or any kind of restraining order over what OpenAI is doing, it just naturally can lead to that sort of situation where it’s going to cause some delays in what OpenAI is working on,” Sean suggested. “Which I’m sure was probably part of the reasoning behind Apple doing this. They don’t do this stuff willy nilly.”
With all those plans on the line, will OpenAI try to settle this as quickly as possible, or did it learn from its recent courtroom victory against Elon Musk that it can endure the cost and embarrassment of a trial? Kirsten, at least, predicts the latter.
Keep reading for a preview of our conversation, edited for length and clarity.
Kirsten: Sean, how do you feel about Sam Altman listening to you with a little device maybe in your pocket?
Sean: I’m good. Maybe that’s predictable, but I’m good. No thanks.
We’ll get into it, I’m sure, but this is allegedly the first product that OpenAI has been working on in its hardware division with Jony Ive and company. They’ve been really coy ever since that weird video they put out last year of them sitting at that coffee shop or bar in San Francisco and sort of talking very vaguely about hardware and legacy devices, meaning laptops and phones. And so if this is the direction they’re headed in, all power to people who want to have somebody like that always listening to them. This is not going to be for me.
Anthony: Part of what we have to remember about those kinds of devices is also that, depending on how mobile it is, it’s not just listening to you, it’s listening to the people around you. I might be fine with it — I’m not fine with it, but let’s say I was — but then if we met up in-person at Disrupt, then suddenly it might be listening to all of us.
There are all kinds of social norms that are going to have to be renegotiated if these things become widespread. I think we should make fun of and criticize people who record other people without consent.
Kirsten: Well, I bring up the device that has been speculated about for a really long time, and we’ll see what it really ends up being once it’s officially introduced, but it’s important in the context of this lawsuit that Apple filed last Friday.
It was the biggest news of the week, certainly, and this is a trade secret lawsuit. It has some pretty wild allegations and we should very much emphasize these are allegations that have been filed in a complaint by Apple. But what it is accusing OpenAI of is a pattern of misconduct at the highest levels, specifically directed towards OpenAI employees who used to work at Apple. And in fact they’ve named the chief hardware officer Tang Tan in this lawsuit.
This is all important because Apple is accusing OpenAI of essentially stealing their trade secrets, but in the context of that, this could be then used for a competing hardware product. I’m wondering if maybe we don’t get into whether this lawsuit has merits, because we haven’t gone through full discovery, but what are your initial impressions of the lawsuit aside from the fact that wow, this is going to be entertaining?
Sean: Two things. One, this is a pretty big risk potentially to whatever it is OpenAI is working on. Even setting aside whether or not the court grants any kind of injunctive relief or any kind of restraining order over what OpenAI is doing, it just naturally can lead to that sort of situation where it’s going to cause some delays in what OpenAI is working on, which I’m sure was probably part of the reasoning behind Apple doing this. They don’t do this stuff willy nilly.
The other is that we think that OpenAI is — we know that they’ve filed confidentially for an IPO. We think it might happen as early as the end of this year, or early next year, if you believe Sam Altman’s cautious language around the IPO. And this just raises a whole bunch of questions around that because, on the one hand, we think their business right now is probably overwhelmingly the software; they’re not really factoring in any hardware business into that picture at the moment.
They’re about to go to the markets and they’re going to be pitching bankers and investors on where they think their addressable market should be, and if they have a big amount of that pegged to a potential hardware division and hardware products, this could be a huge risk to that and changes a lot of the calculus of sort of how the IPO gets priced. So that’s where my head’s at.
Anthony: One [allegation] that I assume that Apple must have pretty solid numbers on is, they said more than 400 Apple employees now work at OpenAI. Granted, both of them are very large companies with many thousands or tens of thousands of employees. So as a percentage, it’s not necessarily huge. But that seems like a lot of people and a pretty serious talent drain.
And the other thing I’m wondering is related to Sean’s point. With the context of the potential IPO, how much damage did OpenAI ultimately take from a marketing and brand perspective from the trial it already went through? That it seemed to basically win, but there was a lot of not-terrible-but-kind-of-embarrassing dirty laundry that came out in the testimony. To what extent are they just like, “We do not want to go through that again”? Or did they take the lesson of, “Hey, we went through it and we survived and we’ll be okay if we have to do another trial with Apple”?
Kirsten: I fully predict the latter, by the way.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Tech
What to watch for after Jensen Huang’s Japan visit
Nvidia’s chief Jensen Huang spent two days — July 15 and 16 — in Tokyo, courting Japan’s industrial and chip-supply elite, weeks after a keynote in Taiwan, and months after a visit to South Korea. He left with deals spanning Japan’s entire tech ecosystem: a national AI factory, partnerships with the country’s leading robotics companies, and agreements with the chip-material suppliers powering Nvidia’s next generation of AI chips. His message was clear. Nvidia is targeting Japan’s factory floor, and many of the country’s biggest manufacturers are joining in. AI’s next chapter, Huang said, belongs to factory floors, robots, and machines, and he wants Japan to build it.
Thirty years ago, a $5 million Sega investment helped keep a near-bankrupt Nvidia afloat; today, Nvidia and Japan’s industrial giants need each other again — this time to build the physical-AI era, starting with these three projects:
Noetra — Japan’s sovereign-AI play. The country doesn’t want to run its factories and robots on American or Chinese AI. So, the government pulled together roughly 44 domestic firms, with SoftBank, Sony, NEC, and Honda at the core, to build its own AI for robots, vehicles, and factory floors. Tokyo is committing up to 1 trillion yen ($6.2 billion) over five years, a bet on homegrown “physical AI,” foundation models built to run machines. Japan wants to own the software brain. The hardware to build it, though, still comes from Nvidia. The U.S. chip giant is building “a Vera Rubin AI factory,” a massive data center packed with its next-generation chips, expected to launch in 2028, with 13,750 Vera CPUs and 27,500 Rubin GPUs, delivering 140 megawatts. Noetra will oversee the effort, with plans to build the data center. Noetra’s plan runs in three stages: a reasoning model heavy on Japanese-language skills starting in fiscal 2026; an omni-modal version handling text, images, video, and audio by 2028; and “Real-world Native AI” built to run robots by 2030, released to outside Noetra developers in phases.
The robotics coalition — Japan’s industrial giants line up behind Cosmos. Nvidia is targeting Japan’s factory floor, and many of the country’s top robotics and manufacturing players are signing on. Fanuc, Yaskawa, Kawasaki Heavy, Fujitsu, Hitachi, NEC, Sony, SoftBank, Kubota, and robotics group AIRoA say they plan to build on Nvidia’s Cosmos models, an open-model effort Nvidia started in May with a handful of global AI labs. In Tokyo, Nvidia gave them a reason to commit, unveiling Cosmos 3 Edge, a version of the model that runs on its Jetson Thor chips inside the machines themselves. Some are already testing a shared control system; others, like Honda R&D and Omron, are building on the tools now. “The next frontier of AI is in the physical world, and this is a once-in-a-generation opportunity for Japan,” Huang said in the company’s statement. “Japan invented modern manufacturing. Now, it has the opportunity to reinvent it for the age of intelligent industries.”
Toyota — cars and physical AI. Toyota uses Nvidia chips across much of its stack. It committed its next-generation vehicles to Nvidia’s Drive platform at CES in January 2025; the newer work extends Nvidia into its manufacturing, where simulations are used to design production lines, into the software that runs its vehicles, and into systems that read road traffic. Toyota’s cars will run advanced driver assistance, which steers and brakes but still requires a driver, a more conservative approach than Waymo and Tesla, which are developing systems that rely less on a human driver.
Huang’s visit put physical AI at the center of Japan’s industrial strategy, and Tokyo is spending to back it. Facing a shrinking workforce, Japan wants 10 million AI-equipped robots across 18 sectors by 2040, backed by $65 billion in public and private physical-AI investment.
The longer game is bigger. Japan’s AI Robotics Strategy, released in March, aims to capture more than 30% of the global AI robotics market by 2040, a market Tokyo values at roughly ¥20 trillion, or about $133 billion. METI is funding a domestic foundation model to run the machines, and Noetra’s Nvidia-powered factory is where models of that scale, into the trillions of parameters, would be trained.
Underneath the industrial case is a sovereign one. As the U.S. and China pull ahead in large-scale AI, Tokyo wants its own data, its own compute, and less dependence on infrastructure it doesn’t control. Huang appeared on July 16 alongside trade minister Ryosei Akazawa at the government’s physical-AI launch, with Prime Minister Sanae Takaichi joining by video. The Takaichi administration has made AI and semiconductors the centerpiece of a growth plan chasing ¥370 trillion ($2.3 trillion) in public and private investment by 2040. Noetra’s factory — which Nvidia bills as “the world’s first national AI infrastructure” — is the clearest bet yet. Japan’s push for independence, at least for now, rests on American chips.
In two days, Huang sat across from nearly every name that matters in Japanese tech — the CEOs of Toyota, Fanuc, Yaskawa, Fujitsu, and Kawasaki over lunch, and dozens of supply-chain chiefs over skewers and whisky in a Kanda izakaya.
It’s the same playbook he ran weeks earlier — a homecoming keynote in Taiwan, fried chicken, and a 50,000-GPU deal in Seoul last fall. This time, it was Tokyo’s turn, with the robots, the supply chain, and the chips underneath.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Tech
Netflix paid $587M for Ben Affleck’s AI filmmaking startup
In a new regulatory filing, Netflix revealed that it paid $587 million in cash for InterPositive, a startup co-founded by actor and director Ben Affleck.
The streaming company announced the acquisition in March, with a statement from Affleck saying he wanted to “protect the power of human creativity.” According to Affleck, InterPublic’s AI tools help filmmakers improve their footage in post-production, particularly when it comes to making up for “real-world production challenges such as missing shots, background replacements or incorrect lighting.”
At the time, Netflix announced that the entire InterPositive team would be joining the company, with Affleck joining as a senior advisor, but it didn’t disclose the financial terms of the deal. A subsequent report in Bloomberg suggested that the deal could be worth up to $600 million.
In its most recent earnings report, Netflix said that around 300 of its titles have already used generative AI.
Blue Jays look to take advantage of skidding Rays, improve position in standings
Phillies have room for improvement going into series against Dodgers
