Summary in Plain Language
Recently, the news of Nvidia being investigated by the US Department of Justice has gone viral. On the surface, it seems like another ordinary antitrust case, but in reality, regulators have finally realized that Jensen Huang doesn’t want to just be a boss of a company that sells AI chips anymore. Over the past year, he has made four major moves that go far beyond the scope of a chip company: securing power quotas, gaining access to AI developers, leveraging hundreds of billions of dollars to lend to the entire industry to buy its own chips, and even getting directly involved in developing open-source models. His goal is to control the entire AI industry, from computing power and traffic to financial resources and technical standards, and to become a “private AI government” that can define global AI rules. However, this seemingly invincible “Godfather Plan” has a very fragile foundation—all of its power relies on the industry’s consensus that “Nvidia is the best option.” Now, from governments to competitors around the world, they are quietly undermining this consensus, and his ambitions have already run into numerous obstacles.
---
Detailed Analysis
1. This DOJ investigation is not an ordinary antitrust case; it’s Nvidia’s “underhanded tactics” that have angered the regulators
Many people think the investigation is because Nvidia has acquired another competitor in violation of regulations. But this move is much more sneaky than a direct acquisition. For $17 billion, Nvidia didn’t buy the AI chip startup Groq; instead, it acquired non-exclusive rights to its chip technology and also poached the founder and the core research team. It’s like using money to eliminate its potential strongest rival without the label of “malicious acquisition.” Antitrust regulators have traditionally focused on direct company acquisitions, but Nvidia has adopted a new strategy of “draining” its competitors. The DOJ’s action is a clear warning to Huang: don’t try to quietly eliminate all potential chip competitors; it won’t work.
2. Instead of selling chips, Nvidia is spending billions on power: The next critical issue for AI is no longer just the chips
Many people are puzzled by why a chip company would invest heavily in power companies. The reason is simple: in the past, when the AI industry was in short of GPUs, chip deliveries usually took weeks or months, but now, even if you buy Nvidia’s chips at a high price, building a data center and connecting it to the power grid can take 3-5 years—there are long waiting lists for transformers, transmission lines, and government permits. The speed of US power grid construction cannot keep up with the demand from AI data centers. Nvidia has bought out power developers with existing grid connections and intermediary companies that can secure land for data centers. This means that even if you get Nvidia’s chips, without the necessary power quotas, your systems won’t be able to operate. It’s like everyone was competing for cooking pots, only to find that Nvidia had already stockpiled all the gas cylinders in the city, essentially blocking the next critical step in the AI industry’s development.
3. From a chip seller to the “property manager + central bank” of the AI industry: Nvidia’s profits come from more than just one-time sales
Huang spent $12.9 billion to acquire Hugging Face, the largest AI platform in the world, with a price-earnings ratio of 86 times. He didn’t buy it for its annual revenue of $150 million but for its 13 million developers and its position as the go-to platform for open-source models. Now, everyone in the AI industry uses this platform to release models and share code. Nvidia doesn’t need to directly suppress other chip manufacturers; by making its GPUs the most user-friendly and prominently displaying its products on the platform, developers will naturally choose Nvidia. Additionally, Nvidia has gathered a financing pool of $500 billion with top financial firms like BlackRock and Goldman Sachs to lend to cloud companies and AI startups to buy its chips, allowing it to share in the profits generated by this computing power. It’s like going from selling a shovel to lending the money for using that shovel, turning a one-time sale into a permanent rental arrangement, similar to a central bank in the AI industry. However, this approach carries significant risks: if the AI applications developed using Nvidia’s technology don’t generate profits, the massive debt could lead to a collapse, with risks many times greater than those of Enron’s collapse.
4. Getting directly involved in open-source models: Protecting against both Chinese competition and client defection
Nvidia is investing $20 billion in developing its own open-source models. This is a dual strategy: on one hand, Chinese open-source models are becoming more cost-effective, and domestic AI chips are making rapid progress. If the industry switches to a combination of domestic chips and Chinese models, Nvidia’s high-end GPUs and CUDA ecosystem will be obsolete. By developing its own models, Nvidia can secure the open-source space and prevent Chinese models from gaining traction. On the other hand, major clients like OpenAI and Google are investing in their own chip development. If these giants stop using Nvidia’s chips, Nvidia will lose half of its revenue. By developing its own models, Nvidia can ensure that small developers still use its chips, even if major clients switch. However, this also puts it in direct competition with all its clients. Currently, no one in the industry fully trusts Nvidia.
5. A facade of being an AI “Godfather”; the foundation is a bubble of consensus
Nvidia’s current dominance seems daunting: it controls over 70% of the global AI chip market, sets software standards, controls access to developers, and holds significant lending power. These powers were traditionally reserved for governments. However, the entire foundation of this “AI Godfather” model is based on a fragile consensus. Countries are actively developing their own chips and ecosystems, and the EU is requiring all AI platforms to be compatible. Google, Amazon, and OpenAI are also investing in their own chip development. If this consensus breaks down and people realize they can develop AI without Nvidia, Nvidia will no longer be the global AI authority but just a larger chip company. The DOJ’s investigation is just the first warning; there are many more pitfalls ahead for Nvidia.