Pacific Intelligent Technologies unveiled factory-built data center modules this week designed to deploy in days on stalled construction sites, the kind of limbo that has become increasingly common as the AI boom strains infrastructure capacity. The Y Combinator-backed startup has disclosed customer commitments, though the figures vary between sources: its Y Combinator profile lists $95 million in early customer commitments for hosted computing power and another $27 million in letters of intent for outright pod purchases, while its LinkedIn page shows $70 million and $23 million respectively.
The pitch hinges on a peculiar inefficiency in the data center buildout race: construction delays and transformer shortages mean valuable grid connections often sit idle while operators wait months or years for permanent facilities to materialize. Pacific's modules, which the company calls "pods," are meant to bridge that gap by monetizing stranded power almost immediately.
Each unit houses up to 32 Nvidia HGX B300 nodes with 983 terabytes of raw NVMe storage and configurable networking fabric, targeting operational readiness within three months of pre-payment. The startup positions the approach as something between traditional cloud infrastructure and purpose-built GPU clusters, with an on-premises twist that distinguishes it from ventures like CoreWeave and Lambda Labs.
Industry data suggests Pacific may have timed its entry shrewdly. JLL's Global Data Center Outlook from earlier this year found that 57 percent of data center projects experienced delays of at least three months, citing labor shortages. TD Economics noted in a subsequent report that only a fraction of announced gigawatt capacity was actually under construction, pointing to GE Vernova's sold-out transformer capacity stretching through 2028 as a bottleneck.
The financing model Pacific uses borrows from aircraft leasing, structuring pod purchases as redeployable equipment that the company says can secure debt at 5 to 6 percent, well below the 11 to 15 percent range typical for construction loans. In its launch announcement, the startup mentioned a first factory "standing up over the next few months, potentially funded by the state of Missouri," though this remains speculative and publicly available state records didn't show a specific program or award naming Pacific as of early fall.
Harper Dammann-Smith, who studied aerospace engineering at the University of Bath and previously founded a student accelerator, leads the three-person team. Pacific operates as a Delaware corporation with headquarters in San Francisco.
On the technical side, production units rely on direct-to-chip liquid cooling. The startup has also promoted "diamond wafers in the cooling system" for extreme-climate operation, though its website clarifies that advanced heat-spreader materials including diamond remain "active R&D, not a shipped feature." Pacific says it plans to vertically integrate supply-constrained components such as transformers, switchgear, and power distribution, an ambitious goal for a startup with limited capital compared to established infrastructure players.
The company's first two clusters were slated to go live in Missouri around the time of the launch announcement in mid-August. Pacific declined to name customers but indicated it was working with large aerospace companies on deployments requiring on-premises security controls, a use case that sidesteps some of the compliance headaches inherent in public cloud arrangements.
One curious wrinkle: Pacific's LinkedIn page listed different letter-of-intent totals than its Y Combinator launch announcement. The social media profile showed $70 million in hosted compute and $23 million in direct pod sales, totaling $93 million, compared to the $122 million sum disclosed in the launch post. The company did not clarify which figure reflected current commitments or whether any letters of intent had converted to binding contracts, a distinction that matters considerably when evaluating early-stage traction.
"We mass-produce modular data centers that deploy on those sites in days and turn their unused power into revenue," the company wrote on its Y Combinator profile, distilling the value proposition to its essence.
Whether Pacific can execute on that promise at scale remains an open question. The infrastructure demands are substantial, the competitive landscape includes well-capitalized incumbents, and the gap between letters of intent and revenue can be wide. But if construction delays persist and power constraints tighten, the startup's bet on stranded capacity may prove more prescient than it first appears.
