Meta's Zuckerberg Warns US Risks Losing AI Race to China, Calls for Policy Overhaul
Meta Platforms CEO Mark Zuckerberg has called on the United States government to cut regulations and barriers to the development of open-source artificial intelligence (AI). The move is deemed crucial so that American technology companies can compete fiercely with their rivals from China.
The statement was made to coincide with the launch of Meta’s latest open-weight model, Muse Glimmer, this week. Meta also confirmed that it will release more similar AI models in the near future.
Unlike giant AI models in general, Muse Glimmer is more compact in size. The model is specifically designed to run agentic tasks directly on Mac or PC computers using just a single graphics card (GPU). The move was taken to capture a market that wants local AI systems without having to rely on cloud servers.
“We have much larger models that will be launching soon,” Zuckerberg said in a video post accompanying his 14-page essay titled “The Future is for Everyone”, as quoted by Reuters on Tuesday (11/8/2026).
In the essay, Zuckerberg stressed the importance of democratising AI technology rather than allowing it to be controlled by a handful of giant corporations. “The idea that AI is so dangerous that the only safe path is extreme centralisation of power is, in my view, problematic thinking,” he asserted.
Zuckerberg’s statement reflects an industry trend that is beginning to look towards open-weight AI. Many global companies are starting to switch to this model due to the ballooning subscription costs of closed AI, as well as widespread concerns over cybersecurity incidents that have involved systems belonging to Anthropic, OpenAI, and Meta itself.
Open-weight systems generally offer far cheaper operational costs as well as flexibility for companies to modify their source code—something that cannot be done with closed models made by OpenAI or Anthropic.
The Threat of Chinese Dominance
Currently, Chinese technology startups are recorded as leading the open-weight AI competition map. Models such as Kimi K3 from Moonshot, Qwen3.8-Max made by Alibaba, and V4-Flash from DeepSeek are reported to be able to match the performance of top-tier US AI systems. In contrast, US giants such as OpenAI, Anthropic, and Google are still holding on to closed-source models.
Yet closed models require far higher costs than open models. This reason has also given rise to the phenomenon of many US companies switching to Chinese AI models and abandoning US-made AI models.
The market responded positively to Meta’s manoeuvre. Meta shares (META) recorded a nearly 3% increase in pre-market trading on Monday, although on a year-to-date basis they are still down around 10%.
In addition to Muse Glimmer, Meta is preparing to release the code weights for Muse Spark 1.2, the most advanced model being developed by their superintelligence team. This project is a major bet for Meta to reclaim the throne in the global AI race.
Touching on Data Centre Issues and US Policy
Not only launching a new model, Zuckerberg also announced a US$1 billion grant fund (around Rp17 trillion) to support local communities around Meta’s data centres. The massive construction of data centre infrastructure often triggers rejection from local residents and has now become a hot political issue in the US.
“One major disadvantage the US has compared to countries like China is the difficulty of building infrastructure here,” Zuckerberg said. This year alone, Meta has allocated a capital expenditure (capex) budget of up to US$145 billion, most of which is absorbed by AI infrastructure.
Zuckerberg urged the US government to overhaul policies, especially regarding rules on data use and distillation methods—a technique for training smaller AI models using the output of larger AI models to be more efficient in computing power. According to him, the strict limits applied in the US actually provide an opening for foreign AI labs to advance more quickly.
“US policy must reduce these barriers if American open-source models are to continue leading the market,” he concluded.