The promise of co-locating renewable generation with battery storage has captured the imagination of project developers and investors alike. On paper, the logic is compelling: pair solar or wind with a battery to smooth output, capture arbitrage opportunities, and reduce grid connection costs. Yet as Daniel Moore-Oats of Arenko reminds us in a recent analysis, the operational reality is far messier than the spreadsheet models suggest. The core tension lies in a simple, uncomfortable truth: forecasts are always wrong, and successful co-location hinges on real-time data and seamless integration, not just predictive algorithms.
The industry has learned through hard experience that weather forecasts, wholesale price projections, and grid constraint predictions carry inherent uncertainty. When a wind farm and a battery share a grid connection point, that uncertainty compounds. A battery responding to a day-ahead forecast may charge when the wind is actually dying, or discharge just as a cloud passes over a solar field, missing the real opportunity. Moore-Oats argues that the only way to navigate this volatility is through a control layer that ingests live telemetry from the generator, the battery, the meter, and the grid operator simultaneously. Without that real-time feedback loop, co-location becomes a gamble rather than a strategy.
This insight arrives at a pivotal moment for the energy transition. Co-located projects are proliferating in markets from the United Kingdom to California, driven by interconnection queue backlogs and the need to maximise land use. Yet many of these projects are being built with siloed control systems—separate software for the renewables plant and the battery, each optimising for its own objective. The result is suboptimal performance, curtailed energy, and missed revenue. Developers are waking up to the fact that integration is not a hardware problem but a software and data problem. The most valuable asset in a co-located site is not the MW of battery capacity, but the ability to process and act on data in milliseconds.
The implications for investors are clear. Due diligence on a co-located project should scrutinise not just the equipment and the PPA, but the control architecture and the data pipeline. Projects that rely solely on heuristic or static forecast models will underperform relative to those built around adaptive, real-time optimisation. For grid operators, the lesson is that co-location should be incentivised through market mechanisms that reward flexibility and real-time responsiveness, not just firm capacity. As the energy system becomes more distributed and weather-dependent, the gap between forecast and reality will only widen. Bridging that gap requires a fundamental shift from planning to live orchestration.
Moore-Oats’s argument serves as a necessary corrective to the oversimplified narratives that have sometimes accompanied the co-location boom. Real value exists, but it is unlocked only when operational complexity is confronted head-on. The future belongs not to the best forecast, but to the most responsive system.
Read the full report at Energy Storage News.