General Compute secures $15 million seed to deploy SambaNova AI inference chips
FUSE VC leads funding round as General Compute places $300 million order for SambaNova SN50 chips, betting on air-cooled efficiency and speed in a fragmented AI market.
General Compute, a new inference-focused neocloud, has secured a $15 million seed round at a $60 million post-money valuation. The funding was led by FUSE VC, with participation from Carya Venture Partners and Village Global Ventures. The company aims to address the growing demand for AI processing power by deploying SambaNova’s specialised SN50 inference chips, which it claims offer superior speed and efficiency compared to standard graphics processing units.
The investment comes as the AI ecosystem faces significant capacity constraints at major hardware providers. General Compute co-founders, CEO Finn Puklowski and CTO Jason Goodison, identified SambaNova as a strategic alternative to competitors such as Nvidia, Groq, and Cerebras. While Nvidia’s $20 billion acquisition of Groq in December and Cerebras’ recent $57 billion IPO highlight intense interest in specialised AI hardware, General Compute is positioning itself to bypass these bottlenecks by leveraging SambaNova’s architecture.
SambaNova, an Intel-backed chipmaker, claims its new architecture delivers between 600 and 700 tokens per second, significantly outperforming the approximately 250 tokens per second typical of standard GPUs. The company has placed an order for $300 million worth of SN50 chips and intends to be the first neocloud to deploy them. This move reflects a broader industry shift towards optimising inference speed and cost, a trend underscored by OpenRouter’s recent $113 million Series B raise.
To reduce infrastructure costs, General Compute plans to install the air-cooled, energy-efficient chips in existing data centres without requiring significant upgrades. This strategy includes pursuing colocation deals with cryptocurrency mining facilities, where the cost of producing Bitcoin has frequently exceeded its market price, creating an opportunity to repurpose infrastructure for AI workloads. The company launched its cloud offering last week, claiming to be the fastest provider for the MiniMax 2.7 open-source large language model.
Joe Hasselmann, who launched Evercrest Capital Partners this year after investing in Groq in 2021, made General Compute his first investment. Hasselmann draws parallels between this partnership and the relationship between Coreweave and Nvidia, suggesting that both the cloud provider and the chipmaker are making mutually dependent bets on a future where no single provider dominates the AI landscape.
