For years, the AI arms race seemed like an exclusive game for the biggest players—OpenAI, Google, Meta, and Anthropic—who collectively raised billions to develop cutting-edge AI models. The assumption? That more money meant better models. But then, DeepSeek happened.
DeepSeek, a Chinese AI startup, shattered expectations by developing a powerful AI model with just $6 million—a fraction of the billions poured into AI giants. The company’s success sent shockwaves through Silicon Valley, raising uncomfortable questions: Have investors overestimated the costs of AI breakthroughs? Have the industry’s biggest players miscalculated what’s truly needed to lead in AI?
Jordan Jacobs, investor at Radical Ventures, found himself fielding calls from panicked investors. “Let’s focus on companies actually building real businesses, not chasing science fiction,” he reassured them. But the debate had already begun.
Over the past two years, venture capitalists have pumped over $155 billion into AI startups, with OpenAI alone securing $24 billion and reaching a valuation of $157 billion—higher than Citigroup or Pfizer. Anthropic followed closely, raising $16 billion. Yet, if DeepSeek could rival these firms with a shoestring budget, was all that capital truly necessary?
Matt Turck of FirstMark Capital summed up the skepticism: “It’s not a good look for AI companies claiming they need ever-larger scale to win.”
Investors who passed on funding foundation models now feel vindicated. Marc Andreessen had warned that investing in large AI models was like “selling rice”—a commodity race to the bottom. Benchmark’s Eric Vishria called foundation models “the fastest depreciating asset in human history.”
Meanwhile, AI investors scrambled to justify their bets. Gavin Baker of Atreides Management pointed to data access as the real moat, arguing that X.ai, backed by Elon Musk, had a unique advantage with exclusive data from X (formerly Twitter).
Beyond Wall Street, DeepSeek’s rise reignited global AI tensions. Some blamed U.S. regulations for stifling competition, while others saw an urgent need to accelerate domestic AI development. “That’s capitalism,” quipped General Catalyst’s Niko Bonatsos, noting that hundreds of startups had already copied DeepSeek’s model within days.
The AI landscape is shifting faster than anyone predicted. And as new players enter the arena, one thing is clear: the billion-dollar AI playbook just got rewritten.
