PaleBlueDot AI, a Silicon Valley GPU infrastructure provider, announced on October 1 that it has closed a $200 million Series C round led by ComputeCore at a $3.2 billion valuation. Existing investor B Capital and other global investors joined. The round caps a steep climb: the company's $150 million Series B, announced in January, valued it at just over $1 billion.
Founded in 2024 and headquartered in Palo Alto, the company operates what it calls a Super Intelligence Infrastructure Platform built on three businesses: self-owned GPU clusters that it customizes for large or specialized workloads, a marketplace that brokers on-demand capacity from a global network of supply partners, and a serverless inference service for developers and enterprises.
The commercial numbers are the substance behind the re-rating. PaleBlueDot says it had signed over $5 billion in customer contracts as of the end of September, with customers in the United States and Japan together generating more than half of its monthly revenue. At its January Series B it also claimed revenue had grown more than tenfold over the preceding year - a company-reported figure that has not been independently audited.
A key credential is its NVIDIA B300 cluster in Japan, which recently earned NVIDIA Exemplar Cloud status. According to the company, every test run exceeded 98 percent of NVIDIA reference performance across six large-model training workloads - DeepSeek-V3, GPT-OSS, Nemotron-H, Qwen3 and two Llama 3.1 configurations - against the program's 95 percent threshold. Each node carries eight Blackwell Ultra GPUs connected via NVLink, with Quantum-X800 InfiniBand delivering up to 6.4 Tb/s of aggregate bandwidth per node and 63.36 TB of local NVMe cache. The company says it also completed a week-long continuous full-load stability test on the cluster.
The financing also leans on debt capacity raised earlier this year. In July, PaleBlueDot closed a $255 million credit refinancing facility structured as a three-year private note, provided by Brookfield Asset Management and Tor Investment Management with JPMorgan acting as placement agent. Supplier financing and structured credit have become standard tooling for mid-tier GPU clouds racing to add capacity against hyperscalers with far larger balance sheets.
CEO Stephen Watts said the company will use the Series C to fund additional compute capacity across more locations and hardware options, and to invest in full-stack and go-to-market teams, with a focus on frontier labs, "Neolabs" and enterprises in the US. The bet investors are making is that an NVIDIA-validated footprint plus a $5 billion contract book converts into revenue fast enough to justify a valuation that has tripled in nine months - and that the marketplace layer can smooth over the capacity shortfalls that individual clusters cannot.
Comments (0)
Log in to join the discussion
Log InNo comments yet