⏱️ Lectura: 9 min
China formally accused the United States of practicing AI hegemonism after Washington threatened to open an investigation into Moonshot AI, the Chinese startup behind the Kimi models, according to Times Now.
📑 En este artículo
The response came from Silicon Valley itself. Nvidia CEO Jensen Huang called Chinese models “excellent” and urged Washington to set aside “science fiction fears” about the technology, according to Yahoo Finance. The dispute exposes a real rift within the United States between the tech industry and those pushing for more restrictions.
TL;DR
- China formally accused the United States of practicing AI hegemonism after Washington threatened to investigate Moonshot AI.
- Moonshot AI is the Chinese startup behind the Kimi family of models, including Kimi K3.
- Nvidia CEO Jensen Huang called Chinese models excellent and urged setting aside science fiction fears.
- Huang stated that the United States misjudged the impact of Kimi on the open AI market.
- Nvidia and Microsoft separately urged Washington to avoid broad restrictions on open models.
- The dispute adds to existing tension between the United States and China over control of open source models.
- Kimi models already compete head to head with closed Western systems on several benchmarks.
The origin of the AI hegemonism accusation
The trigger was Washington’s threat to open a formal investigation into Moonshot AI, the Chinese company responsible for the Kimi family of models. According to Times Now, the Chinese government responded by accusing the United States of practicing AI hegemonism: using its dominant position in the sector to block foreign competitors instead of competing on technical merit.
Jensen Huang’s reaction was quick to join the debate. In remarks cited by Yahoo Finance, Nvidia’s CEO called Chinese models “excellent” and went further, stating that the United States “misjudged the impact of Kimi” on the global open AI market. It’s a striking stance, coming from the head of the company most dependent on the US-China trade relationship for selling AI chips.
This isn’t the first time US private industry has distanced itself from the government’s harder line. Days earlier, Nvidia and Microsoft had jointly asked Washington to avoid broad restrictions on open AI models, according to MSN. The current episode deepens that same line of tension, now with a concrete case at the center of the dispute.
Context and history
The debate over Chinese open models isn’t new. Since early 2026, labs like DeepSeek, Moonshot, Alibaba and Zhipu have been releasing increasingly competitive open-weight models against closed US systems. That pace of global adoption is exactly what worries the more conservative national security sectors in Washington.
The argument from the harder US side is that a model trained and controlled by a Chinese company could pose risks if integrated into critical infrastructure or widely used software supply chains outside China. The argument from the Chinese side, and now also from part of the US industry, is that restricting access to those models doesn’t stop China from continuing to develop them: it only reduces the competitiveness of US developers and companies that had already adopted them.
That’s the core of the AI hegemonism accusation: China argues that Washington is trying to preserve its leadership by blocking access to more open alternatives, instead of letting the market decide.
The following diagram summarizes how the dispute escalated over the past few days:
flowchart TD
A["Kimi gains adoption outside China"] --> B["US threatens to investigate Moonshot AI"]
B --> C["China accuses AI hegemonism"]
B --> D["Nvidia and Microsoft urge halting restrictions"]
C --> E["Diplomatic tension over open AI"]
D --> E
Technical details: what makes Chinese open models competitive
The dispute isn’t just diplomatic, it’s also technical. Moonshot AI publishes the weights of its Kimi models under modified open licenses, letting developers outside China download them, run them on their own infrastructure, or consume them via API without depending on a single provider. That openness is exactly what generates friction: the easier it is to adopt a Chinese model, the less control any government has over its global circulation.
The other technical axis is hardware. Models at the scale of the Kimi family are trained on large GPU clusters, and much of the controversy over restrictions revolves around whether Nvidia can keep selling AI chips to China without compromising US national security. It’s the same argument Nvidia and Microsoft raised in their joint request to Washington: restricting access to models already released openly doesn’t stop China, because those weights are already circulating freely, and it only reduces the competitiveness of US industry.
💭 Key point: once a model is released with open weights, no later restriction can undo its download. The real dispute is about who builds the next generation, not the one already available.
For a developer, the most objective way to assess whether a Chinese model like Kimi is competitive against closed alternatives isn’t to read any executive’s statements, but to run the same benchmarks against both systems and compare the results for the specific use case at hand.
How to try it: accessing Kimi from outside China
Moonshot AI exposes its models through an API compatible with the OpenAI standard, which simplifies integration for any developer already using that request format. The first step is to generate an API key on the Moonshot platform and confirm the endpoint responds correctly.
curl https://api.moonshot.ai/v1/models \
-H "Authorization: Bearer $MOONSHOT_API_KEY"That command lists the models available to the account and confirms the API key is valid before spending tokens on an actual query.
from openai import OpenAI
client = OpenAI(
api_key="YOUR_MOONSHOT_API_KEY",
base_url="https://api.moonshot.ai/v1",
)
response = client.chat.completions.create(
model="kimi-k3-code",
messages=[
{"role": "user", "content": "Explain what a race condition is in three lines"}
],
)
print(response.choices[0].message.content)
print(response.model)The response.model field in the response confirms exactly which model version processed the query, which is useful for verifying there was no silent version change between tests.
💡 Tip: run the same query against a closed Western model and a Moonshot model with the same prompt to compare response time and quality for your own use case, before drawing conclusions from other people’s statements.
Impact and analysis
The episode reveals a rift within the United States. On one side, part of the government is pushing for more control over which Chinese AI models can circulate or be integrated into US products. On the other, companies like Nvidia and Microsoft depend on access to the Chinese market and on the global developer ecosystem that already uses open models, Chinese ones included.
| Actor | Stance on Chinese open models | Main argument |
|---|---|---|
| Chinese government | Accuses the US of AI hegemonism | Sees the restrictions as geopolitical protectionism, not a technical measure |
| Nvidia (Jensen Huang) | Publicly backs the quality of Chinese models | Depends on the Chinese market for AI chip sales |
| Microsoft | Calls for avoiding broad restrictions | Warns that limiting access stifles innovation within the US |
| US national security sectors | Push for more control and investigations | Consider it a risk to integrate models controlled by Chinese companies |
That rift isn’t just rhetorical. If Washington moves forward with a formal investigation into Moonshot AI, the outcome could directly affect US companies that have already integrated Kimi models into their products or internal workflows, creating regulatory uncertainty just as adoption of Chinese open AI had been growing.
⚠️ Heads up: integrating an API hosted in China into production means reviewing where data is processed and which jurisdiction applies, beyond whether the model is technically competitive.
What’s next
There’s still no formal decision on whether the investigation into Moonshot AI will move forward or under what legal framework. The likely outcome depends on how much weight the lobbying of companies like Nvidia and Microsoft carries against the national security sectors within the US administration.
China, for its part, has already made its discursive strategy clear: every new restriction or investigation threat is framed as an act of AI hegemonism, a narrative that serves it both domestically and in international technology governance forums.
What’s concrete for developers and companies is that access to models like Kimi remains available today. Any regulatory change, if it happens, will likely take weeks or months to formalize, not days.
📖 Summary on Telegram: View summary
Try it yourself: generate an API key on the Moonshot platform and run the first curl command from this article to confirm within minutes whether your account already has access to Kimi K3.
Frequently Asked Questions
What is Moonshot AI?
It’s a Chinese artificial intelligence startup behind the Kimi family of models, which includes variants geared toward general use, coding and agents.
What does it mean that China accuses the US of “AI hegemonism”?
It’s the accusation that the United States uses its dominant position in the sector to block Chinese competitors through regulatory restrictions or investigations, instead of competing solely on technical merit.
Why did Jensen Huang defend Chinese models?
Because Nvidia depends commercially on the Chinese market for AI chip sales, and broad restrictions on Chinese open models could affect that relationship as well as the global adoption of models already used by its own customers.
Can Kimi be used outside China?
Yes, Moonshot AI offers an API compatible with the OpenAI format, accessible by generating an API key on its platform and pointing the client to the Moonshot endpoint.
What would happen if the US opens a formal investigation into Moonshot AI?
It could lead to restrictions on the use or integration of its models in US products, although so far it remains a threat and not a formalized measure.
Does this dispute affect other Chinese labs like DeepSeek or Zhipu?
Not directly yet, but any regulatory framework arising from this case could set a precedent for how Washington treats other Chinese-origin open models.
References
- Times Now: report on China’s AI hegemonism accusation and the investigation threat against Moonshot AI.
- Yahoo Finance: Jensen Huang’s remarks backing Chinese models and his comment on Kimi’s impact.
- MSN / Fox Business: joint request from Nvidia and Microsoft to Washington to avoid broad restrictions on open AI models.
📱 Like this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day. @programacion
Imagen destacada: Foto de Ewan Kennedy en Unsplash
0 Comments