Tech giants Alphabet and Nvidia have reportedly invested in Safe Superintelligence (SSI), the AI startup launched months ago by former OpenAI chief scientist Ilya Sutskever, according to a source familiar with the matter who spoke with Reuters.
The investments place Alphabet and Nvidia alongside prominent venture capital firms backing SSI, which has rapidly achieved a significant valuation, solidifying its status as a major player in the AI landscape.
The source indicated that SSI was recently valued at approximately $32 billion in a funding round led by venture capital firm Greenoaks. This high valuation highlights the intense interest in the company, largely driven by Sutskever’s esteemed reputation and track record in anticipating key advancements in AI development.
SSI focuses on AI model research, a field known for its massive demand for computational resources, particularly advanced chips.
These investments signal a renewed push by major technology and infrastructure companies to strategically align with cutting-edge AI startups that require substantial computing power.
The exact financial terms of Alphabet’s and Nvidia’s investments in SSI could not be determined by Reuters.
Alphabet’s involvement with SSI extends beyond just investment. Earlier this week, its cloud computing division announced a significant deal to supply SSI with its proprietary Tensor Processing Units (TPUs). This agreement allows SSI access to Google’s in-house AI chips to fuel its intensive research and development efforts.
This dual approach – corporate investment coupled with cloud service provision – highlights an evolution in Google’s AI hardware strategy.
Darren Mowry, a managing director overseeing Google’s startup partnerships, told Reuters in an interview this week that Google is actively expanding the external sales of its TPUs, which were initially reserved for internal use.
While Nvidia’s Graphics Processing Units (GPUs) have historically dominated the AI chip market, holding over 80% share, sources indicate that SSI is currently primarily utilizing Google’s TPUs for its AI research.
Google Cloud offers both Nvidia GPUs and its own TPUs. The TPUs are designed to excel at specific AI tasks and are often considered more efficient for large-scale model building than general-purpose GPUs.
Other notable companies like Apple and Anthropic, a major OpenAI competitor backed by billions from Google and Amazon, have also employed TPUs.
The competitive landscape includes Amazon Web Services (AWS), which is developing its own custom AI chips, Trainium and Inferentia.
As far back as 2023, Amazon stated that Anthropic would develop technology using these chips, and in December, it announced Anthropic would be the first customer for a massive supercomputer powered by hundreds of thousands of its proprietary processors.
Despite this, two sources confirmed that Anthropic continues to heavily use Google’s TPUs for its ongoing AI development and has not reduced its spending on Google’s chip infrastructure.
