Eclipse Ventures leads 70 million Series B for data center microgrid startup as AI power crunch deepens
Sofia Reyes ·
Vernal Power, a startup that develops modular natural gas and battery microgrids for data centers, said Monday it raised 70 million dollars in Series B financing led by Eclipse Ventures. DCVC and Lowercarbon Capital joined the round, which the company will use to accelerate deployments adjacent to AI clusters in Texas and the Southeast where grid interconnection queues are longest.
The timing signals an investor pivot toward the inputs that govern AI margins, specifically access to reliable megawatts at predictable prices. Compute buyers need power in months, not years, and on site generation can be contracted faster than waiting for new transmission.
Why investors are moving into power for AI
AI inference costs are increasingly shaped by energy price and availability, which can account for a meaningful share of cost per token. LLM operators and cloud providers have been signing bespoke power deals, but most greenfield sites face multi year utility timelines. Microgrids that colocate with data centers can compress time to power and hedge peak rates.
Eclipse has leaned into industrial and energy hardware in prior bets, and DCVC and Lowercarbon have a track record in deep tech and climate. Their participation indicates that top tier firms see power as a venture problem with defensible technology and execution, not only a project finance exercise.
Vernal sells five to seven year offtake contracts that bundle generation and storage with controls software. The company says its units can be installed in under nine months and priced below peak grid tariffs, which can stabilize AI workload costs and improve server utilization by reducing curtailment.
Competition includes established microgrid providers and emerging data center power specialists pursuing gas, solar plus storage, and waste heat strategies. The main risks involve permitting, gas price volatility, and potential policy shifts on emissions for data center energy, which could raise compliance costs or constrain siting.
For venture portfolios that rode the first wave of model and application funding, Monday’s round reflects a second order allocation into energy, land, and latency. If power remains the bottleneck, the edge may accrue to startups that can deliver fast, contracted megawatts that turn GPUs into revenue sooner.