Camus Energy’s software uses flexible operating limits and grid-aware forecasting to let data centers draw conditional power while utilities complete upgrades, promising faster utility service and fewer temporary behind-the-meter solutions.
FlexConnect lets a data center connect sooner by combining a firm power commitment with conditional power governed by an operating envelope.
The platform monitors grid conditions and weather to predict rare moments when conditional power must be curtailed, giving operators advance notice to use batteries or defer compute.
Google selected Camus for its 2026 accelerator and will help advance FlexConnect demonstrations with utilities including Duke Energy, Edison International, and PG&E.
Camus Energy's FlexConnect software targets a frequent bottleneck for hyperscale AI data centers: long delays to interconnect to the utility grid. The company says its approach can shave three to five years off typical wait times by using flexible operating limits within existing infrastructure. Google selected Camus for its 2026 Google for Startups Accelerator focused on energy and AI, and Camus is demonstrating FlexConnect to large utilities including Duke Energy, Edison International, and PG&E.
Many data center developers either wait for full utility upgrades or build behind-the-meter generation as an interim fix. FlexConnect offers a third path: an earlier utility connection that provides firm power plus conditional incremental capacity. That can accelerate commercial operations for hyperscalers and reduce the need for temporary gas-fired or self‑generated power.
Operational trade-offs and claimed benefits
Google's accelerator delivered technical workshops, strategic consultations, and access to utility and investor networks. Camus is already working with Google on deployments and runs FlexConnect on Google Cloud. The company is demonstrating the system to major utilities to produce operational blueprints for interconnecting data centers using flexible operating limits while upgrades proceed.
Astrid Atkinson, CEO and co-founder, said the program helps hyperscalers move beyond an ''all-or-nothing approach to grid access,'' aiming to reliably unlock capacity while keeping the grid stable and affordable. David Bailey, who leads technical work at Camus, described the system as relatively simple: combine utility monitoring and weather forecasts, predict rare curtailment events, and relay an operating envelope so the data center can plan responses.
The demonstrations with utilities including Duke Energy, Edison International, and PG&E will show how FlexConnect integrates with utility operations and whether conditional operating envelopes can be standardized across territories. Outcomes to track include the frequency of conditional curtailments in practice, the quality and lead time of utility notices, and whether utilities adopt similar flexibility tools at scale.
FlexConnect proposes a practical compromise: partial, conditional access to utility power now, with upgrades delivered later. If demonstrations validate the claims, utilities and data center developers could significantly shorten time-to-power without immediate heavy capital outlays.
ComEd has showcased how utilities can collaborate with developers to fast-track flexible DER interconnections.
The grid has been straining under the weight of new AI data centers, and that has utilities courting fusion startups.
As grid connections are delayed, some operators turn to on-site generation. As racks grow denser, others upgrade electrical systems
AI mega-campuses are reshaping grid planning and exposing limits of traditional studies.
The market for commercial systems will grow 27% over the next five years, say American Clean Power and Wood Mackenzie.
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.