London-based energy startup Tem has secured $75 million to accelerate an AI-driven overhaul of how electricity is bought and sold. The company is building a transaction infrastructure that compresses layers of middlemen in power markets, promising businesses prices that hew much closer to wholesale. Tem says more than 2,600 U.K. customers are already on its platform and that some are saving up to 30% on bills, according to company materials.
An AI Engine For Power Transactions in Wholesale Electricity Markets
At the core of Tem’s approach is Rosso, a transaction engine that uses machine learning and large language models to forecast supply and demand, clear trades, and manage risk across day-ahead, intraday, and balancing windows. Instead of separate desks for forecasting, trading, settlements, and risk, Rosso aims to unify these functions in software — shrinking operational costs and the bid-ask spread that quietly inflates customer bills.
Electricity markets are notoriously complex: generators offer capacity into wholesale auctions, suppliers buy on behalf of end users, and grid operators fine-tune frequency and reserves in real time. Each step introduces intermediaries and fees. Tem’s thesis is simple but radical for this sector — if algorithms can continuously price volatility, transmission constraints, and balancing risks, suppliers can transact closer to the clearing price and pass savings through to customers.
Rosso also targets one of the silent cost drivers in retail power: imbalance charges. In Great Britain, suppliers that miss their forecasts pay penalties when the system operator must correct shortfalls or surpluses. By tightening intraday forecasts and automating repositioning, Tem is trying to slash those imbalance costs, which can spike dramatically during scarcity events.
A Neo-Utility To Prove The Model In Live Power Trading
Because incumbents were slow to adopt its infrastructure, Tem built RED, a “neo-utility” that runs entirely on Rosso. RED functions as a conventional supplier for businesses but taps Rosso for forecasting, trading, and settlement. The utility arm gives Tem real transaction flow and a proving ground to refine the engine before opening it to other suppliers.
Tem seeded its marketplace with renewable generators and small-to-midsize enterprises, where distributed assets and variable loads sharpen the algorithms. The customer list spans sectors — including Boohoo Group, Fever-Tree, and Newcastle United FC — a signal that the model can scale from retail and hospitality to heavier commercial demand profiles.
Why This Matters As AI Load Surges And Grids Tighten
Electricity demand is tightening as AI and data centers expand. The International Energy Agency estimates electricity use from data centers, AI, and crypto could roughly double to around 1,000 TWh by 2026 from about 460 TWh in 2022. National Grid ESO has also flagged rising flexibility needs as electrification and digital infrastructure stack onto the system. During stress events, Britain’s system has seen imbalance prices soar and constraint payments add billions to consumer costs in peak years.
In that context, shaving even a few percentage points off procurement and imbalance costs is material. For a mid-market business with a seven-figure annual electricity bill, a 10% improvement can be transformative for margins. Tem’s claim of up to 30% savings won’t hold for every customer, but the mechanism — algorithmic trading closer to wholesale, lower overhead, and better forecast accuracy — attacks the right cost buckets.
Market Design Tailwinds And Hurdles For Tem’s Strategy
Policy shifts could amplify or complicate Tem’s trajectory. The U.K.’s Review of Electricity Market Arrangements is weighing reforms such as nodal or zonal pricing, deeper flexibility markets, and evolved capacity mechanisms. If locational signals strengthen, software-native traders that can digest network constraints in real time should benefit. Conversely, changing settlement rules and tighter prudential requirements could challenge new entrants’ capital efficiency.
Regulators like Ofgem will also expect robust governance around algorithmic decision-making, stress testing, and model auditability. Power markets can turn brutal in rare events; models must handle scarcity pricing, negative prices during oversupply, and the U.K.’s Balancing Mechanism dynamics without blowing through credit limits. Tem’s biggest technical moat may be less about fancy models and more about reliable data engineering, resilient controls, and compliance-grade backtesting.
Where The $75M Likely Goes To Build Utility-Grade Systems
Building a utility-grade transaction stack is capital intensive. Expect Tem to deploy funds on market integrations with exchanges and system operators, credit and collateral optimization, and a broader supply portfolio that includes storage and flexible demand. Expansion to additional European markets with liquid intraday trading — for example, the Nordics and parts of continental Europe — appears logical once the U.K. book scales.
The other milestone is platformization. Today, RED is the exclusive consumer of Rosso. Over time, Tem intends to let other suppliers use the engine, charging for access much like a payments or cloud infrastructure provider. If that happens, Tem’s growth will hinge less on winning retail accounts and more on capturing transaction flow across the market.
Electricity markets are moving from static tariffs to software-defined trading. With $75 million in new fuel and a working utility to validate its stack, Tem is betting that AI can compress the path from generator to meter — and in the process, redraw how value is shared across the power system.
