Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Nvidia has agreed to pay about $20 billion to acquire assets from AI chip startup Grok, marking the company’s biggest deal to date and a continuation of its strategy to absorb potential competitors before they can challenge its market dominance.
The chipmaker’s latest licensing deal mirrors a similar deal just three months ago, supporting the narrative that decentralized AI infrastructure may be the only alternative to Nvidia’s growing dominance.
Sponsored
Sponsored
Triple bonus in three months with Trump Jr
The deal closed just three months after Grok raised $750 million at a $6.9 billion valuation. BlackRock, Samsung, Cisco, and 1789 Capital, where Donald Trump Jr. serving as partner, participated this time. Nvidia acquired nearly all of the company’s assets, with the exception of its cloud computing business, although Groke described the deal as a “non-exclusive license agreement.”
Grok CEO Jonathan Ross, a former Google engineer who helped develop the search giant’s Tensor processing unit, will join Nvidia along with chairman Sunny Madra and other senior executives. The startup will continue to operate independently under the leadership of CFO Simon Edwards, who will become its new CEO.
Recurring plan
Grok’s deal follows a pattern set by Nvidia just three months ago. In September, the company paid more than $900 million to hire Infabrica’s CEO and employees while licensing its startup technology. Both bids rely on licensing structures rather than outright acquisitions, likely in an attempt to avoid antitrust scrutiny stalling Nvidia’s $40 billion bid to acquire ARM Holdings in 2022.
Our strategy is “We will buy you before you can compete with us,” Kobesi’s letter said directly.
Sponsored
Sponsored
Technical advantage and competitive pressure
The GPU uses on-board SRAM instead of external DRAM, which allows the company to achieve 10x better energy efficiency as claimed. This architecture excels at real-time temporal inference, but limits the size of the model – a trade-off that Nvidia can now explore in its wider ecosystem.
Timing took on an obvious importance. Google recently unveiled its seventh-generation tensor processor, codenamed Ironwood, and introduced Gemini 3, built entirely on tensor modules, to the top of the benchmark ranking. Nvidia responded via Platform When the biggest players start issuing reassuring messages like these, it becomes clear that the competitive pressures are clearly increasing.
Impacts on decentralized artificial intelligence
Although it does not have a direct impact on cryptocurrency markets, it supports the narrative that drives decentralized AI computing projects. Platforms like io.net stand out as Alternatives to centralized AI infrastructure.
Jack Collier, head of growth at io.net, explained that people can put their own supplies on the network, either data centers or even through your laptop, contributing the power of your available GPU and getting enough compensation with Tokinomics. The platform claims that institutional customers, including Leonardo.ai and UC Berkeley, have achieved significant cost savings.
There is still a wide gap between narrative and reality. Nvidia’s acquisition of low-latency GROC technology further strengthened its technological leadership, making it difficult for any alternative to offer competitive performance.
The move also raised questions about the independent development of AI chips. Cerebral Systems, another Nvidia competitor preparing to go public, may eventually face similar pressures. It remains to be seen if it can maintain its independence or succumb to the financial pull of Nvidia.
