Nvidia chips don't belong on the bargaining table
News that Nvidia Corp will not be able to sell its customised artificial intelligence (AI) chips in China caught both the company and markets by surprise.
The disclosure came just a day after Nvidia announced a half-trillion-dollar US investment commitment that was celebrated by President Donald Trump, and followed a media report saying that chief executive officer Jensen Huang had struck a deal safeguarding the H20 chips in China over dinner at Mar-a-Lago.
The confusion represented what is now a usual day in Washington as Trump plays out his chaotic trade games and increases pressure on China to negotiate. Beijing, meanwhile, has signalled that one of its chief concerns heading into potential talks are US policies designed to suppress its modernisation.
The scene is now set for Nvidia's H20 chips, made specifically to comply with export controls, to be turned into the ultimate tariff-related bargaining point.
An exemption would please the company, which has long argued that export controls are ineffective and give a boost to domestic players including Huawei Technologies Co. And while Beijing is not backing down, access to AI chips is one thing it desperately wants.
We have seen Trump blur the lines over tech national security concerns in the past. The various export controls on Nvidia predate his tariff blitz, and are meant to hold China back in the AI race over fears that Beijing's edge could give it military or economic advantages.
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
Sign Up
Sign Up
But the president has ignored similar worries before: he extended a deadline for TikTok to be banned in the US, turning the China-owned social media platform into a key leverage on tariffs.
The White House may yet roll back the crackdown on AI chips in China. Regardless, it is a timely reminder that this trade war is risking America's hard-won gains in the race for tech supremacy. Washington cannot fight on two fronts.
Access to chips and computing power has been at the core of Silicon Valley's lead over China in AI, but that gap is closing fast. Washington's tightening chip restrictions have been porous, but they have no doubt bought time.
One thing we have learnt is that they are ineffective without international cooperation. The US had to convince the Netherlands and Japan, key allies who oversee parts of the semiconductor supply chain, to get on board for these policies to have a fighting chance at holding Beijing back. But Trump's 'America First' rhetoric and tariff antagonism do not provide much incentive for joint action any more.
DeepSeek, the Hangzhou-based AI startup that surprised the world earlier this year, shows that China is not as far behind on AI software development as many in the West thought. America's key lead now remains in hardware and namely, advanced chips. But the likes of Huawei and Semiconductor Manufacturing International Corp are working around the clock to produce domestic alternatives to Nvidia's AI processors.
Some analysis suggests that these local players are just a couple of years away from creating viable home-grown Nvidia alternatives to power China's AI boom (while still having dramatically lower performance that Nvidia's top-of-the-line offerings). This means that the window for how long these export controls can still have any impact is closing fast.
Perhaps the biggest indication that these chip restrictions have slowed China's AI ambitions, even if they have not stopped it, has come from DeepSeek founder Liang Wenfeng.
In a rare interview, he said that the biggest barrier for his company was not money, but access to high-end chips. And these H20s are still in high demand: Chinese tech giants including ByteDance, Alibaba Group Holding and Tencent Holdings have reportedly stockpiled orders in the first three months of this year, in preparation for the long-rumoured crackdown.
Much of the recent coverage of the latest Nvidia curbs frame it as more evidence that multibillion-dollar dealmaking with Trump offers no guarantees of tariff reprieve. But the reality is the chip war predated this trade war, and blurring the lines risks losing both.
Nvidia's AI chips should not have been turned into pawns. But they are increasingly looking like the biggest concession Trump now has to get himself out of the chaos he has unleashed. BLOOMBERG
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
CNA
an hour ago
- CNA
Meta to pay nearly $15 billion for Scale AI stake, The Information reports
Meta Platforms has agreed to take a 49 per cent stake in artificial intelligence startup Scale AI for $14.8 billion, The Information reported on Tuesday, citing two people familiar with the matter. Founded in 2016, Scale AI provides vast amounts of labeled data or curated training data, which is crucial for developing sophisticated tools such as OpenAI's ChatGPT. The deal, which has not been finalized yet, appears to be beneficial for Scale AI's investors, including Accel, Index Ventures, Founders Fund and Greenoaks Capital, as well as its current and former employees, the report said. Meta, Scale AI and the startup's investors did not immediately respond to Reuters' requests for comment. As part of the deal, Scale AI CEO Alexandr Wang will take a top position inside Meta, leading a new "superintelligence" lab, according to the report. Meta CEO Mark Zuckerberg has been actively recruiting top AI researchers to boost the company's AI efforts, the report said. The company is fighting the perception that it may have fallen behind in the AI race after its initial set of Llama 4 large language models released in April fell short of performance expectations. Meta delayed the release of its flagship "Behemoth" AI model due to concerns about its capabilities, the Wall Street Journal reported last month. The company is also facing antitrust concerns related to its acquisitions of Instagram and WhatsApp. According to The Information report, the structure for the potential deal with Scale AI could be designed to avoid more regulatory scrutiny. Scale AI was valued at $13.8 billion in a funding round last spring. It generated about $870 million in revenue in 2024 and expects more than $2 billion this year, the report said. The company counts AI firms OpenAI and Cohere as well as tech giants Microsoft, Meta and Cisco Systems among its customers, according to its website.
Straits Times
an hour ago
- Straits Times
EU proposes lowering Russia oil price cap in new sanctions
An apartment building hit by a Russian drone strike, in Odesa, Ukraine, on June 10. PHOTO: REUTERS BRUSSELS - The European Union on June 10 proposed slashing a price cap on Russia's global oil exports, as part of a new package of sanctions over Moscow's war in Ukraine. The move comes ahead of a Group of 7 (G-7) summit in Canada next week where allies will push US President Donald Trump to be more aggressive in punishing the Kremlin. 'We are ramping up pressure on Russia, because strength is the only language that Russia will understand,' European Commission president Ursula von der Leyen said. 'Our message is very clear, this war must end. We need a real ceasefire, and Russia has to come to the negotiating table with a serious proposal.' The European Commission, the EU's executive, suggested cutting the current oil price cap from US$60 to US$45 as Moscow drags its feet on a ceasefire in Ukraine. The cap is a G-7 initiative aimed at limiting the amount of money Russia makes by exporting oil to countries across the world. Set at US$60 by the G-7 in 2022, it is designed to limit the price Moscow can sell oil around the world by banning shipping firms and insurance companies dealing with Russia to export above that amount. To have most impact, the EU and other G-7 partners need to get the United States to follow suit and agree to the cut in level. But Mr Trump so far has frustrated Western allies by refusing to impose sanctions on Russia, despite President Vladimir Putin's failure to agree to a Ukraine ceasefire. 'My assumption is that we do that together as G-7,' Dr von der Leyen said. 'We have started that as G-7, it was successful as a measure from the G-7, and I want to continue this measure as G-7.' 'Massive' sanctions threatened Mr Trump last week said he had a deadline to sanction Russia 'in my brain', but warned that he may also target Kyiv if no advances are made in his peace push. European leaders in May threatened Moscow with 'massive' sanctions if it did not agree a truce. 'Russia lies about its desire for peace. Putin is taking the world for a ride. Together with the United States, we can really force Putin to negotiate seriously,' EU foreign policy chief Kaja Kallas said. As part of its 18th round of sanctions since Russia's 2022 invasion, the EU also proposed measures to stop the defunct Baltic Sea gas pipelines Nord Stream 1 and 2 from being brought back online. Officials said they would also look to target some 70 more vessels in the 'shadow fleet' of ageing tankers used by Russia to circumvent oil export curbs. The EU in addition is looking to sever ties with a further 22 Russian banks and add more companies, including in China, to a blacklist of those helping Moscow's military. One EU diplomat described the latest proposals as 'one of the most substantive and significant packages we've discussed recently'. 'It will hurt Russia's ability to finance its war machine. Now let's see how the discussions evolve.' The sanctions will need to be agreed by all 27 EU countries, and could face opposition from Moscow-friendly countries Hungary and Slovakia. AFP Join ST's Telegram channel and get the latest breaking news delivered to you.
CNA
2 hours ago
- CNA
AI startup Glean valuation surges to $7.2 billion in latest fundraise
Glean said on Tuesday it was valued at $7.2 billion in its latest funding round - the third capital raise for the U.S. enterprise AI search startup in less than two years. The round marks a nearly 57 per cent jump in valuation for Glean, underscoring strong investor appetite for startups leveraging AI use cases. In the previous funding in September, the company's valuation had more than doubled in just over six months. Businesses and governments worldwide are rushing to adopt artificial intelligence, with applications ranging from enterprise productivity and drug discovery to infrastructure and beyond. Palo Alto, California-based Glean raised $150 million in a Series F funding round led by asset manager Wellington Management and other new investors, including Khosla Ventures. The company was founded in 2019 by a team of former Google search engineers. It uses AI assistants and large language models to generate personalized answers to queries. Earlier this year, the company rolled out its Glean Agents offering, which allows businesses to use AI to automate operations. It is on track to support 1 billion agent actions by the end of the year, the company said. Industry leaders have hailed AI-based agents as a transformative-use case of the technology. Microsoft CEO Satya Nadella has also suggested that the agents will disrupt how we use software-as-a-service, a business model that has long been the staple of software startups. The latest investment will help Glean to enhance its product offerings and bolster international presence, the company said. Glean, which surpassed $100 million in annual recurring revenue in its last fiscal year, said it was opening a new San Francisco office to support its growth.
