Halfway through this week, the entire tech world and well beyond remains hyper-focused on DeepSeek.
The Chinese artificial intelligence (AI) startup has managed to send shockwaves through the entire industry with the release of its open-source large language model R1.
💰💸 Don’t miss the move: SIGN UP for TheStreet’s FREE Daily newsletter 💰💸
Between its alleged use of less advanced Nvidia (NVDA) AI chips and its ability to build the model for only $5.6 million, the one-year-old company has had no trouble fascinating investors and consumers.
Conversation around DeepSeek has surged over the past few days, with experts speculating on just how much it may change the world of AI. Leading chip stocks such as Nvidia and Advanced Micro Devices (AMD) , which ultimately fell on the news, have since rebounded.
But one of the tech sector’s most prominent voices, someone well versed in the global AI arms race, predicts that the U.S. is approaching a pivotal turning point.
The U.S. and China could be nearing an AI smackdown
For years, both the U.S. and China have been working hard to stay ahead of each other in the race to global AI domination. In 2022, the launch of OpenAI’s ChatGPT gave the U.S. a much-needed boost as the Silicon Valley organization showed the world what AI models could do.
But now, China seeks to have its own ChatGPT in the form of the DeepSeek R1, a similar AI model that seemingly requires less money to train and less advanced AI chips. Some experts believe it could be the historic phenomenon that puts China ahead of the U.S. in the AI arms race.
Related: Analysts offer blunt take on DeepSeek AI panic
Eric Schmidt, former CEO and chairman of Google (GOOGL) and co-founder of Schmidt Sciences, recently laid out his take on what DeekSeek means for the global AI market in a Washington Post editorial, co-authored by Dhaval Adjodah, co-founder of MakerMaker.AI.
Almost a year ago, Schmidt claimed that China couldn’t match the U.S.'s AI progress, highlighting the AI chip shortage and a lack of language material available in China for model training as likely reasons why.
Now, however, he seems to have adopted a different mindset, acknowledging that even those “on the cutting edge of AI research” have been shocked by DeepSeek’s progress. As Schmidt sees it, this early-stage Chinese startup is demonstrating AI innovation at an astounding speed.
“What’s even more remarkable is that DeepSeek’s entire collection of models is open-source — which in this case means they have open weights that anyone can reproduce and build on top of,” the editorial states.
The authors note that the trend of companies creating open-source AI models that all developers can access is much more common in China than in the U.S. With the exception of Meta Platforms (META) , which has stressed its commitment to “openly accessible AI,” many U.S. tech leaders have opted to keep their technology to themselves.
Schmidt praises not just the R1 model but its new Janus Pro, a multimodal AI model that can generate images from text and, according to DeepSeek, can outperform Dall⋅E 3, a similar model from OpenAI.
More tech news:
- Microsoft announcement lays out urgent advice for tech companies
- Tesla shareholders raise concerns about Elon Musk pre-earnings
- New grassroots group wants to save social media from billionaires
“Until now, closed-source models such as OpenAI’s o3 and Anthropic’s Claude 3 Opus were considered the industry standards with the most advanced capabilities — and they were built in the United States,” Schmidt and Adjodah state. “But DeepSeek’s R1 and Janus Pro show just how quickly the tides of technological supremacy can turn.”
Can the U.S. compete with China in a post-R1 market?
If DeepSeek’s recent progress is as significant for China’s AI dominance as Schmidt and Adjodah suggest, it raises some key questions: Can the U.S. pull back ahead of China? And if it can, what would such a feat entail?
Until the R1 rollout, the U.S. did seem to maintain the lead that Schmidt predicted in 2024, as tech leaders such as Google, Microsoft (MSFT) and Meta, as well as startups like Anthropic, demonstrated progress in producing AI models of their own.
Related: Experts sound the alarm on controversial company’s new AI model
Schmidt and Adjodah also offer a potential solution, stating that “the United States already has the best closed models in the world. To remain competitive, we must also support the development of a vibrant open-source ecosystem.”
They highlight, though, that they see clear pressure on U.S. tech companies that stems from DeepSeek’s ability to produce an AI model in a more cost-effective way.
Many such names have plenty of capital to devote to producing their new AI models but now they face a new challenge: having to compete with a smaller company using free resources.
Related: Veteran fund manager issues dire S&P 500 warning for 2025
