The dominant strategy among hyperscalers (Meta, X/Elon, Oracle, Crusoe) is to surround data centers with behind-the-meter natural gas generators. Shah's case against this isn't primarily environmental — it's engineering and economics. AI training workloads ramp from 20% to 80% load five times a minute, violating decades-old grid standards on load volatility. Natural gas turbines cycle in seconds or minutes, not milliseconds — they're simply not designed for this. The result: turbine shafts rated for 7 years are failing in 10 months. Lithium-ion batteries used as buffers degrade in 5-6 months under that cycling rate. To compensate, operators are over-building capacity by 50%, running one-third of turbines as spares for the broken ones. Shah calls this "literally the dumbest thing that human beings have ever attempted to do."

The alternative Shah champions—and traces to DOE Liftoff Reports and Tyler Norris's landmark Duke paper—is flexible interconnection. Data centers bring some on-site capacity, but connect to the grid with an agreement to curtail load for ~100 hours/year (or less). This means the data center uses grid power 99%+ of the time but can self-island when needed. The key insight from Norris's research: the US grid already has over 98 GW of "headroom" capacity baked in for peak planning. If new loads agree to flexibility, they can tap that headroom without needing massive new generation builds. Average curtailment events last ~2 hours — short enough that lithium-ion batteries can bridge it.

Grid utilization used to run around 70%; today it's below 50%. The culprit is the peak-valley problem: air conditioning created massive summer peaks starting in the 1970s, and deindustrialization stripped away 24/7 baseload. So we keep building more infrastructure for peaks that represent a small fraction of actual hours. The narrative Shah has coalesced around—via the Utilize Coalition—is "use what we've already paid for." This framing is technology-agnostic: batteries, grid-enhancing technologies (GETs), advanced conductors, and VPPs all serve the goal. The political stroke of genius is that increasing utilization can actually reduce rates for everyone while accommodating explosive new load.

Shah's theory of change runs through state governors, not federal regulators or utilities themselves. His evidence: Governor Shapiro capping Pennsylvania's capacity market, Governor Spanberger making grid utilization "her bill" in Virginia (it passed near-unanimously), Governor Sherrill in New Jersey running on electricity as a top-two issue. Deploy Action's model is to pass laws in priority states that require grid utilization measurement (Virginia), moving toward mandates (utilization portfolio standards, analogous to RPS). Shah teases that a "utilization portfolio standard" — forcing utilities to raise utilization from 50% to 60% by X date — is on the roadmap. Michigan's PUC is already requiring VPP evaluations alongside distribution projects.

IOUs (investor-owned utilities) make money by building capital infrastructure — the rate-base model. Asking them to improve grid utilization is asking them to cannibalize their own revenue model. Shah acknowledges this directly. But here's the surprising turn: Wall Street is increasingly nervous about utilities' expansion plans. Utility CapEx went from $20B in 2003 to $178B in 2025. Raising that much new capital worries investors: if rates keep climbing 9%/year, 1-in-5 households fall behind on bills, defaults spread, and eventually the political blowback makes the utility itself a bad investment. IOUs also now cost ~30% more per kWh than public power utilities. Shah's pitch: Wall Street wants utilities to "achieve some balance here," making them an unusual ally for grid utilization reform.

The Lego-block framing is Shah's most memorable: WeaveGrid proved software can manage distribution substation loading (something no distribution engineer believed possible before). Spark Fund is proving that aggregated distributed resources can be trusted by a utility (Xcel's pending approval). Tesla Powerwalls dispatched during California grid crises at $2/kWh. Base Power is scaling residential batteries to 100 MW+ in Texas. Each of these is a building block. No one has yet brought a full basket of distributed capacity to a utility and gotten a large data center interconnected on that basis — but Shah thinks the cascade is imminent once the building blocks are certified.

Shah's most cutting cultural critique: hyperscalers are making engineering decisions based on status signaling, not technical analysis. A big natural gas turbine "feels manly." A spreadsheet full of distributed capacity does not. Google is the lone counterexample — they actually hired the expert talent (Tyler Norris, DOE alumni Lucy Etienne and Campbell Howe), did the grid analysis themselves when Xcel couldn't, identified the Form Energy battery solution, and presented it to Xcel for approval. The Minnesota deal closed. Shah's verdict on the rest: "They're like kids in a soccer game."

The final section's "speed round" grades various constituencies. Key verdict: the clean energy industry has massively underinvested in political power. Oil/gas spent $450M in the last election; clean energy spent $15M. Abby Hopper scraped together $2.5M for her PAC from a $110B/year CapEx industry. Shah's prescription: give money to every county council member near every project, regardless of party. Stop outsourcing political engagement to green groups (whose environmental framing creates a coded resistance). One bright spot: the investment community has actually played politics effectively, including protecting IRA credits during OBBB.

The conversation ends with a sobering geopolitical frame. Physical oil markets are disrupted for at least 3-4 years. Gasoline likely above $4-5/gallon at the midterms. Iran may effectively "own the toll booth" on the Strait of Hormuz — geopolitical leverage approaching nuclear deterrence. Global South nations face humanitarian catastrophe: fertilizer shortages cutting crop yields 50%, NICUs running without power, LPG unavailable for cooking. China positioned to benefit from global realignment. The US-aligned coalition has shrunk to Russia (which Shah calls "the most shocking level of incompetence I've ever seen"). The implicit conclusion: this accelerates the case for electrification and energy independence — but the transition costs will land hardest on the world's poorest.