Intel's acquisition of SambaNova for 1.6 billion likely close to completion
According to several US reports, the deal is almost complete: Intel wants to fully acquire the AI chip startup.
(Image: Hepha1st0s/Shutterstock.com)
According to consistent reports from US media, Intel is close to acquiring the AI startup SambaNova. Initial rumors about such a deal were already circulating in November 2025. As early as mid-2021, the company is said to have raised over a billion US dollars in investments, and was valued at five billion US dollars in the same year – yet Intel is now reportedly set to secure the deal for only 1.6 billion dollars.
This is reported by Bloomberg, among others. Shortly before that, Wired stated that the two companies had agreed on a term sheet for the acquisition, but did not specify a purchase price. Both media outlets refer to unnamed sources but do not cite each other. According to Bloomberg, the transaction is expected to be completed next month. The CES in Las Vegas, the largest industry event for the US high-tech industry, will also take place then.
Intel and its partners are already invested in SambaNova
The fact that Intel is likely to acquire SambaNova's know-how relatively cheaply is probably due to the entanglements between the two companies. Intel CEO Lip-Bu Tan is also the Chairman of SambaNova. Furthermore, Intel Capital has invested in the startup, and Intel is currently trying to spin off its investment division. Another investor in SambaNova is Softbank, which in turn invested two billion in Intel in 2025. Thus, part of the value that SambaNova represents today may already indirectly belong to Intel.
Videos by heise
The acquisition target was founded in 2017 and develops chips for AI applications, primarily for inferencing in data centers. SambaNova already offers its solutions through partners such as Hugging Face. The chip developer is among a number of companies aiming to accelerate parts of artificial intelligence calculations with novel circuits. The architecture, specialized for specific areas of AI development, is said to differ significantly from GPUs like those from AMD and Nvidia, which have so far been leading in AI accelerators.
(nie)
