Beijing’s open-source AI push echoes America’s personal computer boom
In the spring of 1977, a 21-year-old Steve Jobs stood before a crowd at the West Coast Computer Faire in San Francisco, unveiling the Apple II, a machine he insisted belonged not to corporations or universities, but to ordinary people. It was a radical idea in an era when computers filled rooms and cost as much as houses. Within a decade, that idea had restructured the American economy, spawned Silicon Valley, and changed the course of technological history. Steve Jobs later went on to describe the computer as “a bicycle for the mind”, and nearly fifty years later, the “Kingdom of Bicycles” is betting that artificial intelligence can do something similar so that this time, Beijing holds the controls.
The Open-Source Gambit
When DeepSeek, a Chinese AI lab backed by a quantitative hedge fund, released its R1 reasoning model in January 2025 under an open-source license, the reaction in Washington was something between alarm and disbelief. The model matched or exceeded the performance of OpenAI’s best offerings at a fraction of the cost to run. Within days, it had been downloaded millions of times. Within weeks, it was running on laptops in Iowa, servers in Jakarta, and smartphones in Lagos. China made a calculated decision: if it could not outspend the US on frontier AI, it would outflank it on distribution.
The logic echoes, in striking ways, a debate that consumed the American technology industry forty years ago. In the 1970s and early 1980s, the dominant players in computing were IBM, DEC, and Wang, who built closed, proprietary systems. Their customers were governments, banks, and Fortune 500 companies. Jobs, Wozniak, Bill Gates, and the Homebrew Computer Club crowd, as a generation of tinkerers and idealists, decided that computers should be for everyone. They opened the architecture. They sold kits. They wrote software anyone could afford, not just as a product revolution, but as a power transfer.
A Different Kind of Infrastructure
To understand what China is doing with AI, it helps to understand what the 1980s American computer revolution actually built. The personal computer did not succeed merely because it was cheap or small. It succeeded because it created an ecosystem: software developers who wrote programs for it, retailers who sold it, schools that taught it, and eventually an entire generation that grew up treating it as a native technology.
The government’s role in China’s AI strategy is something structurally similar to the role that the US government played in the market. Across China, local governments have begun embedding AI tools into the everyday bureaucratic infrastructure of ordinary life. Farmers in Heilongjiang province use AI-powered apps to diagnose crop diseases by photographing their fields with a smartphone. Delivery workers in Shenzhen use AI navigation systems that learn their routes over time. In Shanghai, AI tutoring programs built on open-source large language models have been deployed in public schools at a scale that no American edtech company has approached. The Chinese government understands that they need to build AI habits, not just build AI products. They want 1.4 billion people to become dependent on AI tools before those tools are shaped by any foreign power. This is something that a lot of Western observers miss.
The Homebrew Moment
In 1975, a small Albuquerque company called MITS released the Altair 8800, a build-it-yourself computer kit sold for $439 (worth about $2700 today). It had no keyboard, no screen, no software, yet it ignited something. Hundreds of hobbyist clubs sprang up across America. Kids in garages started writing code. A culture formed around the idea that the machine was yours to build, to break, and to understand. There was a level of freedom in steering toward the future.
China’s equivalent of the Homebrew Computer Club moment may be happening right now, in a less romantic but no less consequential form: the explosion of open-source AI tools that anyone can download, modify, and deploy. Models like Qwen, released by Alibaba’s research division, have been downloaded tens of millions of times globally. Baidu’s ERNIE (百度), once a closed corporate product, has steadily opened its APIs to developers. And DeepSeek, the unlikely upstart, has become a kind of Altair moment for AI, proof that you don’t need a $100 billion data center to build something remarkable. It’s genuinely disruptive when you can fine-tune a model on a consumer GPU, deploy it on a phone, and when a kid in a village can build an app on top of it. In the meantime, we in the US are making memes and fear-mongering over AI.
Parallels and Divergences
The comparison between China’s AI moment and America’s personal computer revolution is instructive, but it is not perfect. The 1980s boom in America was largely bottom-up, driven by entrepreneurs, hobbyists, and venture capital, with the government playing a supporting but secondary role. China’s AI push is conspicuously top-down, with Beijing setting targets, state-owned enterprises allocating resources, provincial governments providing subsidies, and, of course, the party shaping the narrative.
This creates both advantages and constraints. China can mobilize resources and drive adoption in ways that no private market can match, for an advantage in speed and scale. The limit, however, is that systems built to serve state priorities may not serve human creativity in the same way that the American garage-inventor culture did. The personal computer revolution produced not just useful tools but genuinely new forms of expression, such as desktop publishing, digital music, and eventually the internet itself. Whether China’s AI ecosystem will produce equivalent creative results remains an open question.
There is also the matter of trust. The personal computer succeeded in part because users believed, for right or wrong, that the machine was theirs to inspect, modify, and control. Open-source AI models offer something similar in technical terms, but the political context in China complicates the picture. When your AI assistant is built on infrastructure that the state has access to, the concept of a personal AI takes on a different meaning. The 1980s were about individual empowerment, so whether that’s really what China’s AI democratization is about remains to be seen in the coming decades.
On the other side of the picture, these Chinese models, which are free for anyone to use with Ollama, are seen as biased, reigniting the drive for US models such as Nvidia’s nemotron and Google’s gemma4. Even Meta, very problematic in its own ways, is promising to eventually open-source some recently developed Muse models, after spending obscene amounts of money on a poaching spree of AI developers following the LLaMa model’s forgettable performance.
What the Rest of the World Watches
For countries outside the US and China, the race to put AI in the hands of ordinary people looks less like a geopolitical contest and more like a development opportunity. In Southeast Asia, in Africa, in Latin America, Chinese AI tools are arriving first as cheaper, often better adapted to local languages, and distributed through the kind of grassroots channels that American companies have been slow to cultivate.
Nigeria’s largest mobile operator has integrated Chinese AI tools into its customer service infrastructure. Indonesian developers are building agricultural apps on top of open-source Chinese models. In Brazil, a startup has built a legal aid tool that helps low-income citizens navigate the court system, using a model that traces its lineage to Alibaba’s research division. The 1980s computer revolution mostly happened in America and Western Europe first, while this one is going on everywhere simultaneously, with tools mostly coming from China.
To be blunt, given the US government’s abuses of AI chat logs, there is also a very real perception that American tools can no longer be trusted. By the nature of the rivalry, China understood this from the very beginning. When weighing these two sides, the world now sees the US as the devil they know all too well, and are willing to take a bet on China. Meanwhile, in other areas, such as Europe, the push for sovereign AI development has been given an impetus it would never have otherwise had. Mistral, EuroLLM, Helsing, LightOn, Aleph Alpha, Black Forest Labs, and DeepL all have a stake in the race to build a competitive frontier AI that fully complies with the EU AI Act.
The Unfinished Revolution
In 1984, Apple aired its famous Super Bowl commercial depicting a lone rebel hurling a hammer at a screen showing a totalitarian figure addressing the masses. The metaphor was about IBM. The promise was liberation through personal computing. Forty years later, the most aggressive effort to put powerful AI tools in the hands of billions of ordinary people is coming from a country whose government thoroughly opposes Apple’s enthusiasm for the rebel with the hammer.
The 1980s computer revolution changed the world in ways that its architects did not fully anticipate and could not fully control. It empowered individuals and corporations alike, created new forms of inequality alongside new forms of opportunity, and ultimately reshaped global power in ways that are still unfolding. China’s AI push may bring something stranger, more troubling, genuinely new, or even just the same. What seems clear is that the age of AI as an elite technology that stays locked up in corporate data centers, accessible only to those who could afford premium subscriptions, may be ending faster than anyone in Silicon Valley expected. The computer revolution taught the world that once a technology escapes into the hands of ordinary people, it may be impossible to contain. Beijing has apparently learned that lesson, whether the rest of the world is ready or not for what comes next.
Mobilizing One-Person AI Startups
China’s AI strategy is moving so far as to have local governments in Chinese cities offering benefits like free apartments, free office space, discounted computing power, and converting idle data centers into incubators as they compete to attract “one-person company” startups run by a single founder with the help of AI. With tools like vibe-coding agents and video generators, individuals are empowered to build tech products independently.
Suzhou’s high-tech hub, for example, is developing 30 of these one-person company (OPC) communities, aiming to create 1,000 such enterprises by 2028. Other regions, such as Pudong, provide substantial coverage of computing costs for AI startups. AI adoption across China is being propelled by government support, facilitating grassroots AI entrepreneurship. This multifaceted ecosystem mirrors the organizational mobilization that helped develop American computing culture decades ago, but is turbocharged to seize the future. Waiting idly by in the hopes that AI culture gains traction is not a winning strategy. AI education needs to take place across the landscape immediately, or we risk being a generation behind in a race where every month matters.
Playing with Sharp Objects. 