The 8 things most people missed about DeepSeek

1. DeepSeek r1 is not smarter than earlier models, just trained more cheaply.
2. It is still expensive to operate (“inference”), especially if you want to make it “think” longer like OpenAI’s o3.
3. China didn’t leap ahead of the U.S. DeepSeek’s new technique is easily copied and its system’s performance is not more accurate than other techniques—but the engineering was first-rate, and it’s earned a seat at the table. The fact that it is cheaper really does change the game.
4. Nvidia’s supremacy is threatened, and a correction makes sense (indeed I called it the day before it happened). But the biggest threat isn’t to Nvidia—people will still need GPUs, albeit fewer higher-end ones—but rather to generative AI-centered companies like OpenAI and Anthropic, because price wars will undercut their hopes of making a profit. Furthermore, unlike some companies one could name, DeepSeek is actually, well, open, which may help them lure talent from more walled-off operations.
5. DeepSeek is an economic revolution and geopolitical wake-up call, but that doesn’t directly bring us any closer to artificial general intelligence (AGI).
6. AI is ultimately worth trillions, but generative AI powered by large language models may well not be. This is partly because of inherent issues with reliability and hallucinations, and partly because new innovations like DeepSeek’s keep driving down prices, almost to zero. Everyone is working from essentially the same playbook, and nobody has any obvious moat, which makes the economics of these companies, valued in the tens or hundreds of billions, really unclear. We should not be surprised if see the valuations of some of the GenAI-focused companies drop precipitously. Masayoshi Son’s reportedly investing as much as $25 billion into OpenAI on a valuation of up to $340 billion even though the company has never turned a profit will either seem like a brilliant move or a blunder on a par with his $4 billion investment in WeWork at $47 billion, an overweighting on a charismatic founder.
7. Carrying on with StarGate, a proposed $500,000 billion infrastructure play, may make little sense in the new environment. A better investment for the U.S. might be in fostering the development of newer, more reliable techniques that are more difficult to copy.

The above has been adapted from Marcus on AI, the Substack newsletter by AI expert Gary Marcus.

More Fortune coverage of DeepSeek:

• DeepSeek caused a $600 billion freakout. But China’s AI upstart may not be the danger to Nvidia and U.S. export controls many assume
• China just redefined the global AI race—with massive implications for OpenAI, Nvidia, and foreign policy
• DeepSeek’s CEO came out of nowhere to challenge Jensen Huang and Sam Altman. The overnight success is powered by Gen Z new hires
• DeepSeek caused a massive Nvidia and tech selloff. Should you buy the dip?
• After pouring billions into AI, venture capital faces a reckoning with DeepSeek breakout: ‘It’s a gut check’
• AI chipmaker Cerebras says it’s been ‘crushed with demand’ for China’s DeepSeek from business customers
• AT&T CEO warns more DeepSeek shocks will hit U.S. tech markets
• Nvidia’s Jensen Huang suffers a $20 billion blow to his net worth following DeepSeek bloodbath—it’s the biggest one-day loss in history
• Why DeepSeek is excellent news for the U.S. stock market
• How DeepSeek’s millennial cofounder went from lowkey math geek to one of tech’s hottest players
• Mark Zuckerberg: DeepSeek shows why U.S. must be AI’s ‘global open-source standard’; no reason to rethink spending
• DeepSeek used OpenAI’s model to train its competitor using ‘distillation,’ White House AI czar says
• DeepSeek just flipped the AI script in favor of open-source—and the irony for OpenAI and Anthropic is brutal 
• From TikTok to DeepSeek: America needs to out-innovate China, not ban its apps 
• The lessons and laments of DeepSeek, according to VCs

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.