Follow the updates on LinkedIn
OpenAI’s latest $40B funding was the largest private investment round in history. To put that in perspective, that single investment is ~4x what the Department of Energy has annually budgeted to make its electric grid more resilient.1 The grid has been systematically underinvested and continues to get squeezed. Now, that same strained grid is now expected to fuel the next wave of AI growth, which is already compounding at venture speed. Global data center consumption is projected to double to ~945 TW by 2030, and AI-optimized loads are expected to quadruple in that same window.
This collision has fueled the rise of a reliability premium, an uncapped price on guaranteed 24/7 power. That premium shows up both directly and indirectly. It’s visible in price spikes during grid stress, like when Texas hit the $5,000/MWh cap during a heat wave. It also appears more subtly too, like in higher insurance premiums or premium rents for buildings with better power resilience.
Big tech has already demonstrated willingness to pay for this premium. Just last month, Meta signed a 20-year power purchase agreement with Constellation to underwrite the 1.1 GW Clinton nuclear plant in Illinois. While terms weren’t disclosed, Meta effectively locked in a guaranteed block of always-on, zero-carbon megawatts through 2047.
The Capital Flow Towards Reliability
So why is so much capital pouring into compute while the grid that powers it remains underfunded?
1. Capital chases scarcity. In 2025, investors see compute as the rarest, high-value bottleneck commodity, and are funding it accordingly. That’s why billions flow into AI startups.
2. More compute means more capacity. Every dollar invested in LLMs drives downstream demand for chips, cooling, and megawatts. The digital boom creates a heavy physical burden (e.g. dense racks, thermal loads, backup systems) that lands on an aging grid.
3. Legacy infrastructure strains under pressure. As demand, extreme weather, and price volatility increases, so do the costs of stress events. Utilities pass these costs along through higher customer rates, regulatory bailouts, or rising insurance premiums for outage losses.
4. Rising reliability costs open opportunity (if packaged right). As power costs rise and uptime becomes a business critical need, capital will shift toward those who turn reliability into a product. The market rewards startups that monetize certainty. In this case, that is those building the safety net for a stressed grid.
This logic chain points toward a coming wave of reliability-as-a-service innovation. Seizing that opportunity requires navigating a critical and growing market barrier, friction.
The Friction Gap Is Widening
Digital systems still scale at lightning speed. An LLM fine-tune might take 10 minutes and requires no permit. Meanwhile, even energy projects face delays that stretch from months to years. For high-voltage transmission, the average timeline to secure a “notice to proceed” is 7.5 years. Construction takes another 1.5 to 3.
The more friction a system has, the more advantage incumbents hold. High-friction sectors, like utility-scale energy, are dominated by players who’ve already cleared regulatory moats. Once a permit is secured and steel is in the ground, the position becomes more defensible. Margins follow, because replication is hard. At Equal, we know this firsthand. Utilities and insurers, two sectors we know well, thrive in high-friction environments. And as the friction gap grows, it becomes relatively easier to launch digital products than physical ones. That’s why we believe the most compelling reliability startups will weld software margins onto hard infrastructure. In short: use bits to overcome the drag of atoms.
Some Factors We Consider When Investing in Reliability
Underwrite time, not just TAM. In high-friction markets, speed creates sustainable moats. We look for startups that compress deployment or permitting cycles. If a company can cut a five-year interconnection slog down to five months, that’s value and a pricing lever. Don’t just chase big addressable markets; look at time saved in the bottlenecks.
Plan for higher power costs. Electricity is no longer a stable, cheap input, especially for compute-heavy industries. Hyperscalers already sign power deals well above wholesale rates to guarantee uptime. In diligence, we often ask ourselves: What happens if electricity costs 2–3× more, or uptime drops? If unit economics break, it’s a red flag. Conversely, startups that benefit when power gets more expensive (like storage, efficiency, or flexible load platforms) deserve a closer look.
Sell uninterrupted operations, not hardware. Customers ultimately want certainty that their critical systems stay online, so bundle the battery or generator with a performance guarantee that backs that outcome. The offer then becomes: “Your facility gets 99.9% assured power, or we compensate you,” shifting the conversation from product specs to guaranteed uptime. This outcome-oriented model simplifies the sale, builds recurring service revenue on top of equipment margins, and locks in long-term trust.
Mind the AI cost curve. While power prices rise, the cost of AI compute is falling fast. Startups that rely on expensive AI as a differentiator will face margin compression as models commoditize. Track token-price deflation, and pressure-test gross margins under future cost curves.
Demand robust gross margin, even in asset-heavy plays. Hardware and infrastructure often come with thinner margins. Winning companies find ways to boost blended margins through software layers, financing, or operational leverage. If a company looks more like a contracting firm than a high-growth platform, it’s probably not a venture scalable play.
Compute gets cheaper every time OpenAI cuts token prices while certainty gets more expensive every time the thermostat hits 99 °F and the grid groans. We are eager to invest where those lines cross. The founders who weld software margins onto hard-infrastructure pain points will own the compounding machines of the next decade. If you are building in the space, please don’t hesitate to reach out to sophia@equal.vc and grace@equal.vc.
Note: Many thanks to Kyla Scanlon for her astute thoughts. This post was inspired by her post “The Most Valuable Commodity In The World Is Friction”
Meanwhile, utilities spent a record $179B last year, but less than half of that went toward upgrading the transmission and distribution lines that actually power data centers