A popular app from China sent the tech markets plummeting Monday, fueled by fear that the United States may be losing its primacy in artificial intelligence.
Nvidia shares dropped 16.9%, contributing to a 1.5% slide in the S&P 500, while tech titans contributed to the Nasdaq Composite Index falling 3.1%, its worst decline in over a month.
The setbacks are being attributed to an announcement by China-based DeepSeek that it has developed an AI model that can compete with the likes of ChatGPT, Claude, and Gemini at a fraction of the cost and the rise over the weekend of the company’s free app to the top of the charts in Apple’s App Store in the U.S.
“DeepSeek has gained popularity because it has shown simply how good it is at the tasks consumers care about,” said Hodan Omaar, a senior AI policy analyst at the Center for Data Innovation, a think tank studying the intersection of data, technology, and public policy, in Washington, D.C.
Hodan called out DeepSeek as a rising star in a report she wrote in June. One of the key points of that report was that the gap between the leading models from U.S. industry leaders and those developed by China’s foremost tech giants and start-ups was quickly closing.
“So while some may be surprised that DeepSeek’s new model is rivaling on English language tasks, the signs that it was heading in that direction with the strength of its earlier models were there,” she told TechNewsWorld.
An AI Bargain
Mark N. Vena, president and principal analyst at SmartTech Research in Las Vegas, noted that the DeepSeek app has recently surged in popularity due to its introduction of advanced AI features, such as real-time contextual insights and personalized search recommendations, which make information retrieval faster and more precise.
“Its new integrations with widely used productivity tools have also streamlined workflows, attracting professionals across industries,” he told TechNewsWorld.
“Furthermore,” he continued, “recent updates emphasizing enhanced data security and privacy protections have resonated with users seeking trustworthy digital solutions.”
“It has the reputation of being one of the best generative AI engines but is substantially less expensive to use than its competitors,” added Rob Enderle, president and principal analyst with the Enderle Group, an advisory services firm in Bend, Ore.
“People like a bargain,” he told TechNewsWorld.
Willy Leichter, CMO of AppSOC, an application security and vulnerability management provider in San Jose, Calif., agreed. “Any breakthrough that reduces the massive computing resources and costs of large-scale AI will be popular,” he told TechNewsWorld. “Lowering the costs to entry will benefit innovators, criminals, and anyone else using AI on a budget.”
Curiosity may also be contributing to the app’s popularity. “Its ability to produce results comparable to Western AI giants using non-premium chips has drawn international interest,” explained Ted Miracco, CEO of Approov Mobile Security, a global mobile application security company.
“It’s also possible that interest in Chinese apps has been piqued by the TikTok ban, RedNote, and other Chinese apps in the news,” he told TechNewsWorld.
Questions About Performance
DeepSeek raised some eyebrows last week when it released its DeepSeek R-1 model, which it stated beats OpenAI’s latest o1 model on several benchmark tests.
“Here’s the rub and my biggest concern about DeepSeek,” Vena said. “Verifying DeepSeek’s performance claims can be challenging due to limited transparency and potential restrictions on independent testing or auditing, especially since it operates within China’s regulatory environment.”
“Concerns around access to proprietary data and algorithms, as well as potential bias in publicized benchmarks, further complicate the validation process,” he added.
“Additionally,” Vena continued, “geopolitical tensions and mistrust between nations may make third-party verification more difficult, raising questions about the credibility of its claims.”
Leichter noted that verifying DeepSeek’s claims and doing apples-to-apples comparisons will be difficult. “The shortcuts these Chinese developers found may have hidden flaws, massive security gaps, or other issues we can’t anticipate yet,” he said.
“For people building AI on the cheap, this may not matter, but large businesses building AI applications will need much more assurance that China is likely to provide,” he noted.
Miracco explained that verifying DeepSeek’s claims is challenging due to both language barriers and limited transparency. “Chinese tech companies operate under strict government oversight, which may prevent the release of unbiased benchmarks or third-party audits,” he said.
“We believe the Chinese government has a vested interest in ‘proving’ that sanctions in semiconductors and AI will be ineffective,” he added. “The release of DeepSeek performance claims will no doubt support that narrative.”
Export Controls’ Downside
Miracco argued that DeepSeek is trying to demonstrate that innovation can bypass hardware limitations through optimization and ingenuity.
“This is intended to prove that export controls alone may not be sufficient to curb technological progress in adversarial nations,” he said. “The U.S. may need to ‘double down’ on sanctions by focusing more on software-based regulations — such as open-source restrictions — partnerships, and fostering innovation to maintain its edge, rather than relying solely on restricting hardware exports.”
Enderle countered, however, that restrictions on chip exports implemented by the Biden Administration may have had some unintended consequences. “DeepSeek showcases those blocks are forcing China to advance its own technology more quickly and China appears to be even better than Japan was at reverse engineering or even stealing competitive technology,” he observed.
“The blocks made duplicating and exceeding U.S. tech a top national priority for China, and DeepSeek is only one of the troubling results,” he said.
Vena asserted that DeepSeek’s ability to achieve results comparable to leading U.S. AI firms using less advanced chips challenges the rationale behind export controls.
“This demonstrates that cutting-edge AI capabilities are not solely dependent on the most advanced hardware,” he explained. “Consequently, restricting access to top-tier Nvidia products may not fully prevent the development of competitive AI systems, potentially undermining the effectiveness of such controls.”
“However,” he added, “export controls may still serve a broader purpose by retarding technological advancements in rival nations and preserving U.S. leadership in the most sensitive and high-performance AI applications.”
Democratizing Cutting-Edge Tech
Andrew Bolster, senior R&D manager at Black Duck Software, an applications security company in Burlington, Mass., maintained that the release of DeepSeek undeniably showcases the immense potential of open-source AI.
“By making such a powerful model available under an MIT license, it not only democratizes access to cutting-edge technology but also fosters innovation and collaboration across the global AI community,” he told TechNewsWorld.
“However,” he continued, “DeepSeek’s rumored use of OpenAI Chain of Thought data for its initial training highlights the importance of transparency and shared resources in advancing AI. In the context of ‘Open Source AI,’ it’s crucial that the underlying training and evaluation data are open, as well as the initial architecture and the resultant model weights.”
“DeepSeek’s achievement in AI efficiency — leveraging a clever reinforcement learning-based multi-stage training approach, rather than the current trend of using larger datasets for bigger models — signals a future where AI is accessible beyond the billionaire classes,” he predicted.
“Open-source AI, with its transparency and collective development, often outpaces closed-source alternatives in terms of adaptability and trust,” he said. “As more organizations recognize these benefits, we could indeed see a significant shift towards open-source AI, driving a new era of technological advancement.”
Nevertheless, the markets may be selling the existing AI ecosystem short. “I think the markets are overreacting to the source of the innovation, as opposed to the innovation itself,” observed Chirag Dekate, vice president analyst at Gartner, a research and advisory company based in Stamford, Conn.
“Because of that, it is amplifying jingoistic perceptions,” he told TechNewsWorld.
“It is also amplifying perceptions that you don’t need data centers, [and] you don’t need GPUs when the underlying facts are completely the opposite,” he said. “In order to deliver AI at scale, in order to create AI-native societies of the future, you still need underlying infrastructure.”
Read the full article here
