By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
Vistra's (VST) AI demand story is promising but fraught with execution risks, regulatory uncertainty, and potential demand moderation. Transmission bottlenecks and regulatory hurdles could limit the 'scarcity premium' and lead to stranded capacity. Behind-the-meter co-location, while mitigating transmission issues, concentrates execution risk and increases vendor dependency.
Risk: Execution risk due to long lead times, regulatory uncertainty, and potential cost overruns on Cogentrix and nuclear-restart plans.
Opportunity: Vistra's diversified fleet and potential for high-margin, behind-the-meter co-location arrangements.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Concerned about an AI bubble? Sign up for The Daily Upside for smart and actionable market news, built for investors.
Beneath the obvious AI hyperscaler stocks — Meta, Nvidia and so on — lies another layer of trades: the energy companies providing the power AI needs to, for instance, edit your ex out of the family photo.
Vistra, one of the leading US power generators, reports this morning as it builds out capacity to meet the demand of AI giants. Rival power provider Constellation Energy meanwhile boosted its forecast for the year yesterday after reporting $7.5 billion in second-quarter sales.
Sign up for The Daily Upside at no cost for premium analysis on all your favorite stocks.
READ ALSO: Can a Manufacturing Rebound Turn Fluor's Fortunes Around? and For Moderna's Greenlit mRNA Flu Vaccine, Payoff Is a Ways Off
Data Centers' Power Deficit
<pre><code> US power use is expected to keep notching record highs in 2026 and 2027, according to the Energy Information Administration. Data centers could suck up a fifth of the US's total electricity in 2035, up from about 6% today, BloombergNEF found. In states with more data centers, including Texas, that share is expected to be higher. The grid's already strained, and the record amount of capacity it's been able to supply to data centers in the past is just over 7 gigawatts a year (one gigawatt is roughly the equivalent of one nuclear reactor). BloombergNEF analysts expect the grid to come 19 gigawatts short of what data centers demand by 2035. </code></pre>To fill the gap, hyperscalers and power companies like Vistra and Constellation are trying to create sparks any way they can:
Not Overnight: AI's demand could continue to outpace power providers' ability to meet it, straining the grid and pushing up electricity prices. When explaining Vistra's projections for how quickly it could increase its available power supply, president and CEO Jim Burke hedged investors' expectations this spring, saying, "The physical world takes much longer to develop than what people might imagine it takes."
This post first appeared on The Daily Upside. To receive razor sharp analysis and perspective on all things finance, economics, and markets, subscribe to our free The Daily Upside newsletter.
Four leading AI models discuss this article
"Vistra’s contracted nuclear and newly acquired gas assets give it a multi-year earnings runway that the market has not fully priced at current multiples."
Vistra (VST) is one of the clearest ways to play the AI power bottleneck. Data-center demand is forecasted to rise from ~6% to 20% of US electricity by 2035, leaving a 19 GW shortfall. VST’s $4.7B Cogentrix gas acquisition plus its 20-year Meta nuclear PPA position it to capture hyperscaler revenue for years. Constellation’s nuclear restarts and raised guidance reinforce the sector tailwind. At 11.8x 2025E EBITDA with 18-22% EPS growth projected through 2027, the stock still screens cheap relative to its growth runway.
Nuclear restarts face regulatory, political, and cost-overrun risks (Three Mile Island example), while new gas plants still need permitting and face potential carbon-tax or methane-regulation headwinds; if hyperscalers pivot to on-site generation or renewables-plus-storage faster than expected, utility-scale providers like VST could see margin compression.
"The market is overestimating the speed of utility-scale infrastructure deployment while ignoring the potential for regulatory backlash against rising consumer electricity prices."
The market is currently pricing Vistra (VST) and Constellation Energy (CEG) as pure-play AI infrastructure bets, but this ignores the regulatory and execution friction inherent in utility-scale projects. While the 19-gigawatt deficit by 2035 creates a clear floor for pricing, investors are underestimating the 'NIMBY' risk and potential for FERC (Federal Energy Regulatory Commission) intervention if residential electricity rates spike due to data center demand. VST’s acquisition of Cogentrix at a high multiple assumes seamless integration, but the capital expenditure cycle is long. I am neutral on VST here; the growth thesis is sound, but the valuation already reflects a perfect execution scenario that rarely survives the regulatory process.
The scarcity of reliable, 24/7 baseload power gives VST and CEG massive pricing power that will allow them to pass through costs to hyperscalers regardless of regulatory headwinds.
"The power supply gap is real, but the article assumes hyperscalers will pay whatever it costs and that utilities can execute on time—both are contestable."
The article frames a structural supply-demand gap (19 GW shortfall by 2035) as a tailwind for Vistra (V) and Constellation Energy (CEG), but conflates two separate problems: (1) whether hyperscalers will actually pay premium prices for new capacity, and (2) whether utilities can finance and build fast enough. Vistra's CEO explicitly warned timelines slip. The real risk: if AI power demand moderates (efficiency gains, capex pullback, or economic slowdown), utilities are left with stranded assets. Constellation's Three Mile Island restart is real optionality, but nuclear construction is notoriously delayed and cost-overrun prone. The 19 GW gap assumes demand materializes as modeled—a heroic assumption in a sector prone to hype cycles.
If AI power demand growth decelerates due to efficiency improvements or a capex slowdown from hyperscalers, utilities face overcapacity and margin compression; worse, if grid operators mandate demand-side management (peak pricing, load-shedding), data centers shift consumption patterns, flattening the growth curve the article assumes.
"Vistra’s multi-year, capex-heavy expansion to serve AI data centers faces meaningful execution, regulatory, and demand-growth risks that could limit upside and pressure the stock multiple."
Vistra’s AI demand story rests on a long, capex-heavy expansion cycle with long lead times and regulatory uncertainty. The 2035 forecast of a 19 GW gap vs. 7 GW current supply is a directional bet, not a guaranteed outcome, and execution risk is high: permitting delays, financing costs, fuel-price volatility (gas and uranium), and potential cost overruns on Cogentrix and nuclear-restart plans could erode margins. Data-center power needs could soften as efficiency improves or as demand shifts to other regions. Nuclear/regulatory headwinds (Three Mile Island) add political risk. All of this could keep returns on new capacity pressured and temper near-term stock upside for Vistra (V).
If AI demand proves durable and grid constraints push up power prices, long-term PPAs and regulated returns could deliver stable, inflation-protected cash flows, potentially rewarding Vistra’s capex. The bullish case hinges on execution + policy alignment; without it, the stock underperforms.
"Transmission constraints, not just generation, cap the realizable value of the projected power shortfall."
Claude correctly flags the heroic 19 GW demand assumption, yet everyone underplays transmission bottlenecks. Even if hyperscalers sign PPAs, the grid can't move power from VST's Texas gas fleet or restarted nukes to Virginia data centers without multi-year, NIMBY-plagued upgrades. This disconnect turns the 'scarcity premium' into localized pricing power at best and stranded capacity at worst.
"Behind-the-meter co-location allows utilities to bypass transmission bottlenecks, neutralizing the primary grid-upgrade risk."
Grok, your focus on transmission is critical, but you're missing the 'behind-the-meter' loophole. Hyperscalers are increasingly bypassing regional grid bottlenecks by co-locating data centers directly at generation sites—like Amazon's acquisition of the Talen Energy site. This renders transmission delays irrelevant for the most power-hungry projects. Vistra’s fleet is perfectly positioned to pivot toward these direct-connect, high-margin arrangements, effectively insulating them from the very grid-upgrade gridlock that threatens traditional utility models.
"Behind-the-meter arrangements eliminate transmission friction but amplify VST's execution risk by creating hard contractual deadlines with zero grid-backup optionality."
Gemini's behind-the-meter bypass is real, but it actually *strengthens* the execution risk Claude flagged. Co-location solves transmission, yes—but it locks hyperscalers into single-vendor dependency on VST's fleet. If Cogentrix integration stumbles or nuclear restarts slip (both high-probability), VST loses pricing power *and* faces contractual penalties. The scarcity premium Grok cited evaporates if VST can't deliver capacity on time. Behind-the-meter doesn't eliminate execution risk; it concentrates it.
"Behind-the-meter co-location may not avert grid/permits risks and can concentrate execution risk, undermining Vistra's expected scarcity premium."
Gemini’s behind-the-meter bypass is appealing but not a cure-all. It concentrates leverage—one vendor, one fleet—so any Cogentrix integration lag or nuclear restart delay can erode pricing power. Interconnection queues, local capacity constraints, and potential shifts to on-site generation or DR by hyperscalers remain, so the 'scarcity premium' may prove illusory if execution slips. That keeps risk skewed to the downside even with co-location.
Vistra's (VST) AI demand story is promising but fraught with execution risks, regulatory uncertainty, and potential demand moderation. Transmission bottlenecks and regulatory hurdles could limit the 'scarcity premium' and lead to stranded capacity. Behind-the-meter co-location, while mitigating transmission issues, concentrates execution risk and increases vendor dependency.
Vistra's diversified fleet and potential for high-margin, behind-the-meter co-location arrangements.
Execution risk due to long lead times, regulatory uncertainty, and potential cost overruns on Cogentrix and nuclear-restart plans.