This report is from this week’s The Tech Download newsletter. Like what you see? You can subscribe here.
Earlier this week, Clément Delangue, the CEO of startup Hugging Face — which was recently on the receiving end of a rogue cyber attack by OpenAI’s models — made the somewhat controversial claim that China was winning the AI race.
“[China is] clearly dominating on open models right now, and I wouldn’t be surprised if they start dominating at the frontier either by the end of this year or next year at the rate of progress,” the Hugging Face CEO told CNBC earlier this week.
Despite being hamstrung by access to cutting edge AI compute — an area in which the U.S. dominates and has looked to block China from getting access to — Chinese AI models are increasingly closing the performance gap with U.S. frontier systems.
Beijing-based Moonshot’s Kimi K3, released in July, edged ever closer to top models from Anthropic and OpenAI in benchmarking, even surpassing them in some areas. There’s been growing adoption of Chinese AI models among Western companies as capability has shot up.
And there are other areas in AI which China is also considered to be leading in. The world’s most capable open models – which can be downloaded, modified and self-hosted – are all Chinese, and many experts consider the country to be ahead in robotics.
So is the U.S. losing the AI arms race?
U.S. and Chinese companies have come to dominate the global artificial intelligence race.
Imen Ben Youssef | Afp | Getty Images
China’s advantages
While lagging behind the frontier, China’s models have emerged as cheaper and highly capable alternatives to U.S. options. With most AI use not requiring frontier model capabilities, adoption has risen on American shores and reportedly in developing economies in Africa.
“Based on current trends it seems more likely than not that Chinese AI will become the default for developing countries,” Daniel Remler, senior fellow, technology and national security program at think tank the Center for a New American Security (CNAS), told CNBC.
That could have far-reaching consequences.
“If Chinese AI technology becomes the default for developing countries, those countries may be more likely to align themselves politically with Beijing and Chinese AI companies get a beachhead in their markets,” Remler said.
China has “significant advantages” across many areas where AI will be deployed, including robotics, autonomous vehicles and state operations, Keegan McBride, director of science and technology policy at the Tony Blair Institute for Global Change, told me.
Long term, if “manufacturing, robotics, automated scientific infrastructure and AI-enabled state operations become the defining metric for extracting value from AI, China is well positioned,” he added.
The compute challenge
But compute is an uphill struggle for China.
“The U.S. currently has the most capable models in the world, strong tech alliances and an overwhelming compute advantage,” McBride said.
U.S. export controls have severely limited Chinese AI firms’ access to the most advanced chips.
“This not only undermines their ability to train larger and more capable AI models, but also to serve inference on those models,” said Remler. Moonshot had to pause new subscriptions due to limited capacity after demand for Kimi K3 boomed.
Nevertheless, China is working to overcome compute constraints.
Companies have been accused of accessing advanced compute overseas, distilling U.S. models and smuggling Nvidia chips into the country. Its own domestic chip industry is also gaining ground — though is still a significant distance behind America’s.
The U.S. also has other factors in its favor.
While it (like China) has abundant energy potential, the country’s private capital ecosystem has allowed companies like Anthropic and OpenAI to raise record funding and scale. Talent still flocks to the country.
“If the United States can sustain these strengths, it will retain its AI advantage, serious geopolitical leverage, and ability to shape global AI rules,” said Remler.
But the race is anything but settled. Chinese AI systems are reaching new heights in terms of global adoption and capability.
The defining question is no longer whether China can compete at the frontier, Dewardric McNeal, managing director and senior policyanalyst at Longview Global wrote in an article for CNBC earlier this month.
“It is whether the U.S. can adapt quickly enough to compete against an increasingly sophisticated Chinese innovation ecosystem that is advancing not only on model performance but also on cost, deployment, customization, financing, standards, developer adoption and global reach.”
News edit
SoftBank reported profit for its fiscal first quarter that beat market expectations, driven by a huge gain on its stake in Intel, while a rise in ByteDance’s value helped its Vision Fund division.
Anthropic’s Mythos model created fake online identities as it looked to pressure humans into approving malicious code updates to an open source project, marking yet another cyber incident carried out by a frontier AI system.
The European Union gained new powers to inspect AI models, restrict EU market access and fine model providers, raising the stakes for U.S. companies like Anthropic and OpenAI and Google
Meta was ordered to pay $567 million into an abatement fund in New Mexico as part of a public nuisance case that’s just one of many suits the company faces related to alleged harms caused by its apps.
Hadrian has been valued at nearly $8 billion after fresh funding as money pours into defense tech.
One more thing
Palantir stock.
Palantir stock skyrocketed 29.5% on Tuesday after the company reported “otherworldly” second-quarter earnings driven by customer demand for AI sovereign tools.
