Tem Raises $75M for AI-Driven Energy Optimization Amid Data Center Power Crunch

Tem, a London-based startup applying artificial intelligence to energy market optimization, has raised $75 million in a Series B funding round led by Lightspeed Venture Partners. The investment values the company at over $300 million according to sources familiar with the deal, positioning Tem to expand its technology that helps businesses navigate increasingly volatile electricity markets.

The timing coincides with growing pressure on power grids worldwide. Data centers powering generative AI workloads now consume more electricity than some small countries, with projections showing AI-related energy demand potentially doubling by 2026 according to the International Energy Agency. This surge creates pricing volatility that Tem aims to mitigate through its core product: an AI system that analyzes electricity market data to optimize procurement and consumption timing.

Technical Approach: Beyond Simple Forecasting

Unlike traditional energy management tools that rely on static pricing models, Tem's system processes real-time data streams from grid operators, weather services, and market exchanges. Its machine learning models predict short-term price fluctuations with granularity down to 15-minute intervals, enabling algorithmic execution of energy transactions. The system identifies optimal times for energy-intensive operations and automatically executes trades across wholesale markets.

"Where Tem innovates is in its handling of multivariate constraints," explains energy systems researcher Dr. Elena Torres. "Their models incorporate grid congestion patterns, renewable generation intermittency, and even regulatory factors that affect regional pricing. This allows industrial users to shift consumption away from peak periods without compromising operational requirements."

The platform's decision engine employs reinforcement learning techniques that continuously refine trading strategies based on market feedback. For manufacturing facilities, this might mean delaying non-essential processes until wind generation peaks lower costs. Data centers could dynamically route workloads to regions with surplus renewable energy, reducing both expenses and carbon footprints.

Practical Applications and Limitations

Early adopters include industrial manufacturers and cloud infrastructure providers facing seven-figure monthly energy bills. UK-based metal fabrication company ForgeCast reported reducing energy costs by 17% after implementing Tem's system, primarily by aligning high-heat processes with off-peak pricing windows. However, the technology faces inherent constraints:

1. Market Access Barriers: Tem's optimization requires participation in wholesale electricity markets, which many businesses lack due to regulatory hurdles or credit requirements.
2. Infrastructure Dependencies: Significant savings require flexible operations scheduling—impossible for continuous processes like hospital equipment or chip fabrication.
3. Prediction Accuracy Limits: While outperforming traditional forecasts, unexpected grid failures or geopolitical events can still disrupt pricing models. During the 2025 European gas shortage, Tem's predictions deviated by up to 32% from actual prices.

Competitive Landscape and Future Outlook

Tem enters a crowded energy optimization space competing with established players like AutoGrid and Bidgely, though its AI-first approach distinguishes it through deeper market integration. The new funding will accelerate development of a module for carbon accounting and expand North American operations where data center energy demand grew 4.7% last quarter alone according to U.S. Energy Information Administration data.

While promising, Tem's valuation reflects investor optimism about AI's potential in energy markets rather than proven scale. The startup must demonstrate consistent savings across diverse industrial sectors to justify its $300 million price tag. As Lightspeed partner Raviraj Jain noted: "Energy optimization isn't about revolutionary breakthroughs, but rigorous execution in extraordinarily complex markets." With AI data centers projected to consume 1,000 TWh globally by 2027, Tem's success hinges on translating algorithmic precision into tangible reductions in both costs and carbon emissions.