Nvidia Absorbs Enfabrica: Massive Investment in AI Technology

Nvidia, the graphics processing giant, has just closed a deal in which it spent more than $900 million (approximately CZK 20.7 billion) to hire Enfabrica CEO Rochan Sankar and other employees of the artificial intelligence startup. The transaction includes both cash and stock and was finalized last week. Rochan Sankar has thus joined Nvidia, according to sources close to the deal. This move resembles similar talent acquisitions carried out by companies such as Meta and Google, where the focus is more on people and technology than on the entire company.

Enfabrica, founded in 2019, specializes in developing technology capable of connecting more than 100,000 graphics processing units (GPUs) into a single functional system. Its solution helps Nvidia offer integrated systems built around its own chips, allowing large GPU clusters to operate as a single computer. The company says its hardware addresses challenges in artificial intelligence data centers where enormous amounts of data are processed. Nvidia had already invested in Enfabrica in 2023 as part of a $125 million funding round (approximately CZK 2.875 billion) led by Atreides Management. At the time, Enfabrica did not disclose its valuation but said it had increased fivefold compared with the previous round.

What Does Enfabrica Do and Why Is It Important for AI?

Enfabrica focuses on hardware and software for the next generation of computing, particularly for artificial intelligence and data centers. Its key product is a switch called the Accelerated Compute Fabric, which enables large groups of GPUs and machine-learning accelerators to be connected efficiently. This technology combines functions such as PCIe, CXL, RDMA-NIC, and Ethernet, bridging servers and networks in hyperscale data centers. The company is headquartered in Mountain View, California, employs around 77 people, and generates annual revenue of approximately $20.9 million (CZK 480 million).

Specifically, Enfabrica develops chips that enable clusters to scale to as many as 100,000 GPUs, making them ideal for training large language models such as the one from OpenAI. Nvidia sees this technology as a way to strengthen its position in the AI boom that began with the launch of ChatGPT in late 2022. Its GPUs are sold in massive clusters to cloud providers that offer AI services to clients. Late last year, Enfabrica raised another $115 million (CZK 2.645 billion) from investors including Spark Capital, Arm, Samsung, and Cisco, bringing its valuation to approximately $600 million (CZK 13.8 billion), according to PitchBook.

Deal Details and Comparison with Other Acquisitions

This Nvidia deal is not a traditional acquisition of an entire company, but rather an "acquihire" – the acquisition of talent and a technology license. Nvidia is paying in both cash and stock, allowing it to integrate key personnel quickly without regulatory obstacles. Meta took a similar approach in June, spending $14.3 billion (CZK 329 billion) on Scale AI founder Alexandr Wang and others and acquiring a 49 percent stake. In July, Google closed a $2.4 billion deal (CZK 55.2 billion) for Windsurf CEO Varun Mohan and his team, including licensing fees.

Nvidia is not known for major acquisitions, but this ranks among its largest. For example, in 2019 it acquired Israeli company Mellanox for $6.9 billion (CZK 158.7 billion), strengthening its networking technology for current products such as Blackwell. Its attempt to acquire Arm failed in 2022 due to regulatory issues. This year, Nvidia completed its acquisition of Run:ai for $700 million (CZK 16.1 billion) to optimize AI infrastructure. In addition, this week it announced a $5 billion investment (CZK 115 billion) in Intel and a partnership on AI processors, as well as an investment of nearly $700 million (CZK 16.1 billion) in British data center startup Nscale.

Why Is Nvidia Betting on Moves Like This?

Nvidia is benefiting from the AI boom, in which its GPUs power model training in data centers, such as the $4 billion facility (CZK 92 billion) in Wisconsin announced by Microsoft on Thursday. Older chips such as the A100 were standalone, but new products come in racks with 72 GPUs working together. Enfabrica helps extend this to a massive scale, giving Nvidia an advantage in competing with companies such as Microsoft, Amazon, and Google, which are also hunting for AI talent. This deal underscores how Nvidia is strengthening its position in the artificial intelligence ecosystem through smart investments and acquisitions.