General Compute Secures $100M Series A to Scale Inference-First Cloud Infrastructure

General Compute has closed a $100 million Series A round led by Blackbird Ventures and Abstract Ventures, lifting its post-money valuation to $800 million — up from $60 million after a May seed round. The raise comes amid a turbulent narrative around "neoclouds," with CoreWeave's stock under pressure and Elon Musk predicting mass consolidation. CEO Finn Puklowski dismisses the pessimism, citing relentless inbound demand for any available compute, whether TPUs, GPUs, or alternative accelerators.

Unlike peers chasing Nvidia GPU clusters, General Compute has built its strategy around inference-optimized silicon from Nvidia rivals. A $400 million debt-financed purchase agreement with SambaNova anchors the hardware roadmap, now supplemented by Positron and Cerebras. This diversification insulates the company from single-vendor supply risk while addressing the growing inference workload segment.

Why it matters for GPU / AI infrastructure

The startup's physical deployment model is equally distinctive: rather than constructing greenfield data centers, it leverages smaller power envelopes, colocation facilities, and repurposed bitcoin mining sites. This approach yields faster time-to-revenue and avoids the multi-year permitting and power procurement cycles that constrain hyperscalers and well-funded neoclouds alike. Puklowski targets 75 MW deployed by 2027, with line of sight to 200 MW of accessible power.

General Compute's customer base consists of inference-specialized cloud providers that need bare-metal servers without software abstraction layers. Blackbird's conviction stems partly from its investment in Baseten, an inference cloud experiencing rapid growth. As open-weight models proliferate, demand for cost-efficient, high-throughput inference infrastructure is accelerating — creating a natural distribution channel for alternative chipmakers seeking scale beyond hyperscaler partnerships.