DeepSeek brings out the war hawks and Cold War rhetoric; at Davos, AI leaders clash over safety and $100bn Stargate project; AI is still top of mind for leaders in 2025
India races to build its own AI industry; has Europe missed its big AI moment?; an AI-powered robot is helping scientists identify deep sea species
When Chinese AI research lab DeepSeek released R1, an open weights AI model that was on par with OpenAI’s o1 reasoning model, the response from many in the American media and industry circles was swift and alarmist. The achievement was quickly framed as a "Sputnik moment"—a term evoking the Soviet Union’s launch of the first artificial satellite in 1957 and the subsequent Cold War space race. This analogy is misleading: R1 is not a disruptive step change with military and geopolitical implications, it’s just a great example of dogged engineering that produced a capable model.
In a normal world, DeepSeek-R1 should’ve seen the same reception as 01.ai’s Yi models which were released around the same time a year ago: both companies showed how open research accelerates engineering and scientific efforts, helping smaller teams to innovate faster and build upon previous efforts. Both Yi and R1 matched or outperformed similar models from American AI labs, were quickly and warmly embraced by the AI community, climbed to the top of the Hugging Face scoreboard, and introduced some interesting and innovative approaches. And both also relied heavily on other open models such as Meta’s Llama.
But we’re not in a normal world anymore. Instead, what we’ve seen over the last week is a dangerous regression into a Cold War-era mindset that reduced scientific progress to a zero-sum game of geopolitical dominance and triggered a tech selloff where hundreds of billions of dollars in value disappeared in a few hours.
For the past two years, the discourse around artificial intelligence has been shaped by American companies and policymakers with a vested interest in framing AI as an arms race. By positioning China as an existential technological threat, Silicon Valley has successfully lobbied for more favorable government contracts, regulatory carve-outs, and investments in AI infrastructure under the guise of national security.
This rhetoric has not gone unnoticed in Beijing. As American voices drumed up fears of a looming technological showdown, China has responded in kind and portrayed AI development as a strategic battleground. Chinese government officials, already engaged in a war-like rhetoric over Taiwan, Tibet and other regional disputes, have also escalated tensions with the United States in a cycle of mutual suspicion.
Caught between these geopolitical disputes, many Chinese tech companies, including some of the leading AI labs, tried hard to obfuscate their origins: any references to China were scrubbed from their websites and apps, their founders kept a low profile and avoided media interviews, and some even chose to relocate to Singapore, Canada and the US to appear more globally minded. Just take a cursory look at AI video generator Kling AI and its social media profiles, and you’ll see no mention of Kuaishou, the Chinese company that built it, or its founders. The same is true for MiniMax and their Hailuo AI platform: despite the company being headquartered in Shanghai, the platform identifies itself as based in San Francisco on social media.
Ironically, under pressure from tough domestic regulation, these companies did more than fake their location; they looked to Silicon Valley for inspiration on how to survive. Many of them adopted a flat hierarchy, moved fast despite fewer resources, and prioritized open scientific and engineering breakthroughs over closed, incremental product optimization.
Meanwhile, American Big Tech, once defined as an anti-establishment force for disruptive progress, became bureaucratic, risk-averse, and self-referential. American companies are now more focused on copying each other rather than pioneering genuinely novel breakthroughs.
Against this backdrop, DeepSeek was a rude awakening for many living in North America and Europe who were not aware of the dynamics shaping the Chinese tech ecosystem. Rather than relying heavily on tens of billions of dollars of investments and sprawling managerial structures, DeepSeek—just like its peers—operated with a leaner, younger team and more experimental approaches. This agility allowed them to iterate rapidly, deploying AI models at a speed that surprised people who were used with the iterative product cycles of American incumbents.
As someone who was worked for large Chinese and American tech companies, DeepSeek-R1 came as no surprise. When I last visited China in 2023, I spoke with many scientists and researchers who, while finding the chip export controls annoying, were more determined than ever to advance the field of AI and build awesome products.
But what surprised me this week was how quick everyone has been to adopt the language of militarization, especially the people who should know better. So let’s be clear: this is an ideological choice, not an inevitability. The framing of AI as a Cold War-like race between two superpowers fundamentally distorts the reality of technological progress, which has historically thrived on international collaboration and open scientific exchange. When American firms and policymakers stoke fears of China’s AI advancements and vice versa, they are escalating geopolitical tensions and creating an environment where innovation is constrained by artificial barriers.
Rather than defaulting to Cold War comparisons, the focus should be on creating an AI research ecosystem that prioritizes a healthy, global competition which ultimately produces shared benefits. The real challenge ahead is not which country "wins the AI war” or which model is better, but how AI research translates into products that benefit humanity as a whole. Treating every technological breakthrough as a battlefield maneuver risks stifling progress, which in turn leads to worse products and reinforces a world order where science is a weapon rather than a tool for advancement.
The "Sputnik moment" narrative does not serve AI research—it serves entrenched corporate and political interests. If we continue down this path, the biggest casualty will not be a geopolitical adversary, but the spirit of scientific discovery itself.
Speaking of Sputnik, tracking the first Russian satellite provided American scientists with valuable information. They were able to deduce the density of the upper atmosphere by calculating Sputnik 1’s drag on the orbit, and the propagation of its radio signals gave data about the ionosphere. In other words, American scientists studied Sputnik 1, learned from it, and built better technology. Similarly, DeepSeek’s researchers learned from Meta’s Llama, and built a better model. What do you think the rest of the AI research community, including Meta’s researchers, are doing right now?
And now, here are this week’s news:
❤️Computer loves
Our top news picks for the week - your essential reading from the world of AI
DeepSeek
FT: How small Chinese AI start-up DeepSeek shocked Silicon Valley
Wired: Chinese AI App DeepSeek Soars in Popularity, Startling Rivals
TechCrunch: How DeepSeek’s efficient AI could stall the nuclear renaissance
The Information: Meta Scrambles After Chinese AI Equals Its Own, Upending Silicon Valley
Bloomberg: DeepSeek Challenges Everyone’s Assumptions About AI Costs
Business Insider: Meta's chief AI scientist says DeepSeek's success shows that 'open source models are surpassing proprietary ones'
Bloomberg: Nvidia Calls DeepSeek ‘Excellent’ AI Advance, Dismisses Concerns
Wired: How Chinese AI Startup DeepSeek Made a Model that Rivals OpenAI
TechCrunch: Viral AI company DeepSeek releases new image model family
The Guardian: The DeepSeek panic reveals an AI world ready to blow
FT: China’s emboldened AI industry releases flurry of model updates
Bloomberg: DeepSeek Tests Meta’s Ambition for US Open-Source AI Dominance
CNBC: ASML CEO sees low-cost AI models like DeepSeek driving more demand — not less
Forbes: Why These AI Chip Startups Are Rejoicing Over The DeepSeek Freakout
FT: AI leaders clash over safety and $100bn Stargate project
Business Insider: A lead Block engineer said the fintech's new AI agent writes better code than him — and the fintech is open-sourcing it to competitors
The Atlantic: OpenAI Goes MAGA
Fortune: Which issues will be top of mind for leaders in 2025? AI, of course, but it’s just the beginning
Bloomberg: An AI-Powered Robot and Gaming Are Helping Scientists Identify New Deep-Sea Species
Bloomberg: India Races to Build Own AI Models as DeepSeek Leaps Ahead
Fortune: Amid the AI arms race, these IPOs could be poised for long-term investor success
WSJ: Why ‘Distillation’ Has Become the Scariest Word for AI Companies
Keep reading with a 7-day free trial
Subscribe to Computerspeak by Alexandru Voica to keep reading this post and get 7 days of free access to the full post archives.