Open Source AI Labs Challenge China: Forbes Analysis

The AI model scoreboard used to look like a cloud bill with bragging rights attached: bigger training runs, tighter access, higher benchmark claims. Forbes is pointing at a different scoreboard now, one where Kimi, Deepseek, and Qwen are described as dominant AI tools, while Poolside, Mira Murati’s Thinking Machines, and Reflection AI try to answer from the American open source side. That is a product strategy story wearing a model race jersey. If developers pick the tool they can inspect, adapt, and spread, closed model scale starts to look less like a moat and more like a very expensive castle wall. ## Forbes Names The Challengers Forbes frames the current matchup cleanly: Chinese open source models like Kimi, Deepseek, and Qwen are dominant AI tools, and American upstarts including Jason Warner’s Poolside, Thinking Machines, and Reflection AI are trying to challenge them. That is the real launch analysis here, even if it is not a single product demo with confetti. Poolside is useful as the anchor because it signals the bet American labs are making: developer trust and architecture may matter as much as raw model size. The temptation is to read this as another America versus China leaderboard fight. That misses the product motion. Open source AI distribution behaves like a developer tool marketplace: the model that gets forked, fine tuned, benchmarked, embedded, and argued about earns compound interest in public. Closed models can still win enterprise accounts, but open models can win the hallway conversation before procurement even gets a login. ## USCC Explains The Flywheel The U.S. China Economic and Security Review Commission gives the clearest explanation of why China’s strategy has traction. Its March 23, 2026 paper says China has gone all in on an open source AI approach, with most Chinese labs publishing source code and weights, while also charging far less for high end products than global competitors. The same report says Alibaba’s Qwen models had more than 100,000 derivatives on Hugging Face as of publication, which is not just usage, it is ecosystem surface area. The Daily Herald, reporting on the same advisory body, described China’s open source dominance as creating a "self-reinforcing competitive advantage" and said cheaper Chinese large language models from firms including Alibaba, Moonshot, and MiniMax dominate worldwide usage rankings on HuggingFace and OpenRouter. That is the flywheel founders should notice: low friction drives adoption, adoption drives feedback, feedback drives iteration, and iteration drives more adoption. It is the software equivalent of putting the free samples at the front of the store, except the samples learn from everyone who takes one home. ## Rest Of World Shows The Product Choice Rest of World captures the engineering philosophy split through Tiezhen Wang, the former head of the Asia Pacific ecosystem at Hugging Face. The publication writes that OpenAI and Anthropic favor a closed source approach, keeping proprietary model code behind a commercial interface, while Chinese AI labs aggressively release open source models that developers can download, inspect, and customize. That difference is not academic. It changes who gets to build adjacent products, who sets defaults, and who becomes the platform other teams quietly depend on. The Wall Street Journal adds another useful data point from the product launch side: Moonshot AI said it planned to fully open source Kimi K3 by late this month, and said the model has 2.8 trillion parameters. The parameter count gets the headline, but the planned release model is the more strategic detail. A giant model behind a locked interface is a vendor; a giant model that developers can adapt starts acting like infrastructure. ## New Scientist Shows The Panic Is Fading New Scientist reported that DeepSeek’s open source R1 model, released in January 2025, made global headlines because it was free to download and was reported to rival powerful U.S. company models. The publication also noted that a trillion dollars was wiped off the value of U.S. tech companies after that release, while Z.ai’s GLM-5.2 launch last month drew similar performance claims without the same panic. That normalization matters. Once the market stops treating every strong Chinese open model as a surprise, American labs have to compete on durable adoption rather than novelty. For founders, the lesson is not to staple an open license onto a weak product and call it strategy. The lesson is to decide where your compounding loop lives. If it lives in enterprise data contracts, closed access may be rational. If it lives in developer mindshare, architecture, customization, and cheap experimentation, open source is not a charity move, it is distribution with a compiler attached. The next logical move for Poolside, Thinking Machines, and Reflection AI is not merely to publish impressive benchmark tables. It is to make their models easy to adopt, easy to extend, and hard to replace once they sit inside developer workflows. Watch the forks, the derivatives, the pricing pages, and the integrations. In platform fights, the winner is often the one everyone builds around before anyone agrees the fight has started. ## Sources - American Open-Source Labs Think They Can Beat China’s Best AI Startups

• How China's Open AI Strategy Reinforces Its Industrial Dominance
• The Daily Herald - China's open-source dominance threatens American AI lead, US advisory body warns
• OpenAI or Deepseek? Role of China’s open-source in the AI race - Rest of World
• China's Moonshot AI Releases Model to Challenge Top ...
• The US-China AI arms race has taken an unexpected turn | New Scientist

Sources

• Neo notches $100M, a sponsor-led REIT deal and Karman's space buy - Axios
• American Open-Source Labs Think They Can Beat China’s Best AI Startups
• How China's Open AI Strategy Reinforces Its Industrial Dominance
• The Daily Herald - China's open-source dominance threatens American AI lead, US advisory body warns
• Andrew Ng on X: "There is now a path for China to surpass the U.S. in AI. Even though the U.S. is still ahead, China has tremendous momentum with its vibrant open-weights model ecosystem and aggressive moves in semiconductor design and manufacturing. In the startup world, we know momentum matters: Even if a company is small today, a high rate of growth compounded for a few years quickly becomes an unstoppable force. This is why a small, scrappy team with high growth can threaten even behemoths. While both the U.S. and China are behemoths, China’s hypercompetitive business landscape and rapid diffusion of knowledge give it tremendous momentum. The White House’s AI Action Plan released last week, which explicitly champions open source (among other things), is a very positive step for the U.S., but by itself it won’t be sufficient to sustain the U.S. lead. Now, AI isn’t a single, monolithic technology, and different countries are ahead in different areas. For example, even before Generative AI, the U.S. had long been ahead in scaled cloud AI implementations, while China has long been ahead in surveillance technology. These translate to different advantages in economic growth as well as both soft and hard power. Even though nontechnical pundits talk about “the race to AGI” as if AGI were a discrete technology to be invented, the reality is that AI technology will progress continuously, and there is no single finish line. If a company or nation declares that it has achieved AGI, I expect that declaration to be less a technology milestone than a marketing milestone. A slight speed advantage in the Olympic 100m dash translates to a dramatic difference between winning a gold medal versus a silver medal. An advantage in AI prowess translates into a proportionate advantage in economic growth and national power; while the impact won’t be a binary one of either winning or losing everything, these advantages nonetheless matter. Looking at Artificial Analysis and LMArena leaderboards, the top proprietary models were developed in the U.S., but the top open models come from China. Google’s Gemini 2.5 Pro, OpenAI’s o4, Anthropic’s Claude 4 Opus, and Grok 4 are all strong models. But open alternatives from China such as DeepSeek R1-0528, Kimi K2 (designed for agentic reasoning), Qwen3 variations (including Qwen3-Coder, which is strong at coding) and Zhipu’s GLM 4.5 (whose post-training software was released as open source) are close behind, and many are ahead of Meta’s Llama 4 and Google’s Gemma 3 — the U.S.’ best open-weights offerings. Because many U.S. companies have taken a secretive approach to developing foundation models — a reasonable business strategy — the leading companies spend huge numbers of dollars to recruit key team members from each other who might know the “secret sauce“ that enabled a competitor to develop certain capabilities. So knowledge does circulate, but at high cost and slowly. In contrast, in China’s open AI ecosystem, many advanced foundation model companies undercut each other on pricing, make bold PR announcements, and poach each others’ employees and customers. This Darwinian life-or-death struggle will lead to the demise of many of the existing players, but the intense competition breeds strong companies. In semiconductors, too, China is making progress. Huawei’s CloudMatrix 384 aims to compete with Nvidia’s GB200 high-performance computing system. While China has struggled to develop GPUs with a similar capability as Nvidia’s top-of-the-line B200, Huawei is trying to build a competitive system by combining a larger number (384 instead of 72) of lower-capability chips. China’s automotive sector once struggled to compete with U.S. and European internal combustion engine vehicles, but leapfrogged ahead by betting on electric vehicles. It remains to be seen how effective Huawei’s alternative architectures prove to be, but the U.S. export restrictions have given Huawei and other Chinese businesses a strong incentive to invest heavily in developing their own technology. Further, if China were to develop its domestic semiconductor manufacturing capabilities while the U.S. remained reliant on TSMC in Taiwan, then the U.S.’ AI roadmap would be much more vulnerable to a disruption of the Taiwan supply chain (perhaps due to a blockade or, worse, a hot war). With the rise of electricity, the internet, and other general-purpose technologies, there was room for many nations to benefit, and the benefit to one nation hasn’t come at the expense of another. I know of businesses that, many months back, planned for a future in which China dominates open models (indeed, we are there at this moment, although the future depends on our actions). Given the transformative impact of AI, I hope all nations — especially democracies with a strong respect for human rights and the rule of law — will clear roadblocks from AI progress and invest in open science and technology to increase the odds that this technology will support democracy and benefit the greatest possible number of people. [Full text: https://t.co/jn0KNi3gmA ]" / X
• The US-China AI arms race has taken an unexpected turn | New Scientist
• American Open-Source Labs Think They Can Beat China's ...
• OpenAI or Deepseek? Role of China’s open-source in the AI race - Rest of World
• China's Moonshot AI Releases Model to Challenge Top ...
• On China's open source AI trajectory - by Nathan Lambert
• Why aren't any American open-source AI labs even close ...