%20(3).png){kind=small}
![TFD [LOGO] (10).png](https://static.wixstatic.com/media/bea252_c1775b2fb69c4411abe5f0d27e15b130~mv2.png/v1/crop/x_150,y_143,w_1221,h_1193/fill/w_179,h_176,al_c,q_85,usm_0.66_1.00_0.01,enc_avif,quality_auto/TFD%20%5BLOGO%5D%20(10).png)
Nvidia’s Next Act: Making Money From the AI Chips Challenging Its Dominance
- By The Financial District
- 2 hours ago
- 2 min read
Custom AI chips are emerging as one of the biggest threats to Nvidia’s dominance. CEO Jensen Huang’s answer is to participate in that market rather than fight it.
![Nvidia CEO Jensen Huang is pursuing a broader AI infrastructure strategy as customers increasingly develop custom AI chips. [Photo: MediaTek Facebook]](https://static.wixstatic.com/media/1c4fd3_f34a4bfb3b6f4e53ae2bd0a7856e5b92~mv2.jpg/v1/fill/w_980,h_515,al_c,q_85,usm_0.66_1.00_0.01,enc_avif,quality_auto/1c4fd3_f34a4bfb3b6f4e53ae2bd0a7856e5b92~mv2.jpg)
Nvidia recently invested $3.5 billion in Taiwanese chipmaker MediaTek through convertible bonds, strengthening a partnership that will help customers develop custom AI infrastructure capable of integrating with Nvidia-based data centers, Jared Blikre reported for Yahoo Finance.
Meanwhile, Anthropic has committed billions of dollars to secure Nvidia-powered computing capacity from cloud providers, underscoring the growing demand for AI infrastructure.
The figures involved in such deals are not directly comparable with Nvidia’s MediaTek investment: one is an investment by Nvidia, while the other represents spending commitments for computing capacity.
The deals nevertheless point to a broader strategy. Huang wants Nvidia to become the underlying architecture for AI systems rather than remain primarily a supplier of GPUs.
At its most basic level, an AI factory takes in electricity and data and produces tokens, the units generated by AI models. Huang has described the process as: “The input is electrons, the output is tokens.”
Nvidia’s strategy is to sell more of the infrastructure surrounding its GPUs.
The company estimates that its revenue opportunity for each gigawatt of AI-factory power capacity has increased from roughly $18 billion with Hopper to $25 billion with Blackwell and $40 billion with Vera Rubin.
On Nvidia’s latest earnings call, Chief Financial Officer Colette Kress said the Vera Rubin platform spans the Vera CPU, Rubin GPU, NVLink, InfiniBand or Ethernet networking and Groq’s LPU.
The increasing revenue opportunity reflects Nvidia’s expansion beyond GPUs into CPUs, networking, systems and other components of the AI factory.
