A little-known Chinese AI startup has upended the global tech landscape, triggering a massive stock selloff and raising questions about the future of artificial intelligence development. DeepSeek, a Hangzhou-based company, introduced an AI model that claims to rival leading competitors at a fraction of the cost. The announcement sent shockwaves through financial markets, wiping nearly $1 trillion from US and European tech stocks.
Anticipation of a major AI breakthrough from China had been growing, especially during discussions at the World Economic Forum in Davos (January 20-24, 2025). DeepSeek’s R1 model was cited as a reason for nations like the US to accelerate AI advancements. Engineers on tech forums had already begun comparing its capabilities to top models from OpenAI and Microsoft. The startup’s model quietly climbed AI performance rankings, fueling curiosity about its potential impact.
Within 36 hours, DeepSeek became a major topic in the tech community. Influential figures such as investor Marc Andreessen and Meta’s chief AI scientist Yann LeCun acknowledged its significance, with Andreessen describing it as a groundbreaking development.
By the end of the weekend, DeepSeek’s AI assistant had surged to the top of Apple’s iPhone download charts and ranked among the most downloaded apps on Google Play. However, the overwhelming demand caused an hour-long outage. The company later restricted new registrations to users with mainland China phone numbers, attributing the move to what it described as large-scale cyberattacks.
Stock Market Turmoil and AI Cost Concerns
DeepSeek’s claim of developing a competitive AI model at significantly lower costs led to investor uncertainty over the spending strategies of major tech firms. This triggered a sharp decline in AI-related stocks, with Nvidia experiencing a record $589 billion market value drop on January 27. Some losses were later recovered in after-hours trading.
The emergence of a cheaper alternative has raised concerns about whether dominant AI firms may lose pricing power. Analysts, including IG Asia’s Jun Rong Yeap, suggested that the industry might need to reevaluate its aggressive investment strategies.
Despite the disruption, OpenAI CEO Sam Altman acknowledged DeepSeek’s achievement, calling the model impressive, particularly in terms of efficiency. He noted that competition was invigorating and announced plans to accelerate OpenAI’s product releases in response.
Censorship and Global Adoption Challenges
Like other AI models developed in China, DeepSeek imposes self-censorship on politically sensitive topics. Unlike ChatGPT, it avoids discussions about subjects such as Tiananmen Square, President Xi Jinping, or China-Taiwan relations. This restriction may pose challenges for global adoption, as international users may find such limitations unusual compared to Western AI models.
The success of DeepSeek’s model also challenges assumptions that AI requires ever-growing computational power. While major AI firms have relied on vast energy resources, DeepSeek claims to have achieved similar performance with significantly lower costs. The full details of its energy consumption remain undisclosed, but initial estimates suggest it required only a fraction of OpenAI’s resources.
DeepSeek’s emergence despite US restrictions on Nvidia’s advanced chips has also raised questions about the effectiveness of trade sanctions. Industry experts have pointed out differences in innovation strategies, with China demonstrating an ability to optimize technology despite limited access to high-end hardware.
Implications for the Energy Sector
Beyond the tech world, DeepSeek’s efficiency has also unsettled the energy industry. The startup has reported training its R1 model using only 2,000 Nvidia chips—far fewer than what is typically required for similar AI programs. This development could have major implications for energy demand in AI-driven data centers.
Expectations of skyrocketing energy consumption have driven significant investment in AI-related infrastructure. However, DeepSeek’s approach suggests a more energy-efficient path forward, prompting concerns among investors in energy companies. On January 27, US energy stocks declined sharply, with Constellation Energy, a major player in AI-related energy capacity expansion, losing over 20% of its market value.
Industry analysts cautioned that while AI remains a major driver of energy demand, the market may have overestimated its future growth. Travis Miller, an energy strategist at Morningstar, noted that gains in computing efficiency could challenge power generation assumptions, though long-term trends like data center expansion are still expected to support energy investments.
A Shift in AI Development?
DeepSeek’s rapid rise has reshaped industry conversations about AI efficiency, cost structures, and global competition. Its model challenges the notion that only companies with vast resources can lead in AI development.
As markets continue to react, the full impact of DeepSeek’s approach remains uncertain—but its disruption is already forcing tech and energy industries to rethink their strategies.