DeepSeek, a relatively unknown Chinese artificial intelligence (AI) startup that we wrote about last month, has upended the tech world and Wall Street with its innovative and cost-effective AI model. The startup’s achievements have sparked conversations about the future of AI development, raised questions about Silicon Valley’s dominance, and sent shockwaves through global markets.
Founded just a year ago in 2023 by Liang Wenfeng, a co-founder of the Chinese AI quantitative hedge fund High-Flyer, DeepSeek develops open-source AI models. Open-source technology, which allows developers worldwide to access and contribute to the software, is at the heart of its strategy.
Last week, the company unveiled its groundbreaking AI models, DeepSeek-R1-Zero and DeepSeek-R1, demonstrating capabilities comparable to established leaders like OpenAI and Meta. What sets DeepSeek apart, however, is its ability to achieve these results at a fraction of the cost.
READ ALSO:Forget GPT-4: DeepSeek V3 is the Open AI revolution you didn’t see coming
According to the company, training the R1 model cost just $5.6 million—staggeringly low compared to the billions of dollars U.S. tech giants like OpenAI, Meta, and Google spend on similar models. This cost efficiency was achieved using less advanced hardware, a direct counterpoint to the prevailing belief that cutting-edge AI requires massive investment in high-performance chips, such as those produced by Nvidia.
DeepSeek’s success challenges several assumptions in the AI industry. For years, U.S. firms have been spending heavily on advanced chips, infrastructure, and data centers, with Nvidia emerging as the go-to supplier.
This ecosystem has thrived on the premise that such investments are necessary to maintain a technological edge. DeepSeek’s achievement turns that narrative on its head, showing that cost-effective solutions can deliver comparable results.
Marc Andreessen, a prominent tech investor, called DeepSeek’s work “one of the most amazing and impressive breakthroughs I’ve ever seen.” Such praise, coming from a veteran of Silicon Valley, underscores the potential of this Chinese upstart to redefine industry norms.
Market reactions
The announcement sent ripples through financial markets, particularly among tech stocks. On Monday, the S&P 500 fell by 1.4%, and the tech-heavy Nasdaq plunged 2.3%. Nvidia, a cornerstone of the AI hardware market, saw its stock tumble by 12%. Other chipmakers, including Marvell, Broadcom, and Micron, also suffered significant losses.
Meta, which has committed to spending upwards of $65 billion on AI development in 2025, and Google, whose parent company Alphabet also invests heavily in AI, were not spared. The broader market impact was stark, given that tech stocks make up nearly 45% of the S&P 500.
DeepSeek’s rise has implications beyond market volatility. It highlights a growing shift in the global AI landscape, where innovation is no longer confined to Silicon Valley. Despite U.S. restrictions on the export of advanced AI chips to China, DeepSeek’s engineers achieved their results using less powerful GPUs, signaling a potential workaround to these constraints.
“DeepSeek’s model rollout is leading investors to question the lead that U.S. companies have, how much is being spent, and whether that spending will lead to profits—or overspending,” said Keith Lerner, an analyst at Truist quoted by news agencies.
The disruption extended to industries reliant on AI growth. Energy companies, which have seen rising demand for electricity to power AI data centers, experienced steep losses. Constellation Energy, the company behind the planned revival of the Three Mile Island nuclear plant for AI-related power needs, fell nearly 20%. Natural gas and oil futures also dropped, reflecting a broader recalibration of expectations.
Cryptocurrencies weren’t immune either. Bitcoin and other digital currencies tumbled, as speculative investments faced renewed scrutiny.
Despite the excitement, some experts remain cautious. Michael Block, a market strategist at Third Seven Capital, noted that while DeepSeek’s achievement is remarkable, it remains to be seen whether the company can sustain its success. “The race is on as to what technology works and how the big Western players will respond and evolve,” he said.
DeepSeek’s claims about cost savings have yet to be independently verified, and its focus on consumer-facing large language models may limit its ability to tackle industrial AI applications that require extensive infrastructure. Giuseppe Sette, president of AI market research firm Reflexivity, pointed out that the U.S. still has a strong talent and capital base, which could keep it at the forefront of AI innovation.
As DeepSeek’s models gain traction, they are also sparking renewed interest in undervalued Chinese AI companies. ByteDance, Tencent, Baidu, and Alibaba, among others, are now reevaluating their pricing strategies to compete with DeepSeek’s aggressive cost model.
Meanwhile, U.S. tech giants are gearing up for a response. Meta is already investing in the next iteration of its Llama model, aiming to reclaim its competitive edge. Nvidia, facing pressure from cost-efficient alternatives, may need to rethink its pricing and product offerings to stay relevant.