• THE GRID
The Forecast That Couldn't Keep Up
Last December, BloombergNEF told us U.S. data centers would need 106 gigawatts by 2035.
That number lasted seven months.
Last week, BNEF tore it up. The new figure: 194 gigawatts. That's an 83% jump from their own call made just before Christmas.
| BNEF Dec 2025 106 GW | BNEF Jul 2026 194 GW | |
| DC Share Today 5.9% | DC Share 2035 ~20% | |
When Starlink Completes Its Global Takeover...
It's set to connect 3 billion people who've never had reliable internet.
That's the biggest connectivity explosion in human history.
Wall Street is focused on SpaceX.
But the smartest money is looking at a backdoor play almost no one's talking about.
A Potential IPO company already positioned to profit from every single new Starlink user — in 170 countries — before they ever go public.
And if you act fast, you can invest now at $0.52/share.
But don't wait.
Their last round sold out with 59,000+ total investors. Once word spreads — it could be too late.
Read that grid again. One in five watts of U.S. power will flow to data centers by 2035. Up from roughly one in seventeen today.
But the uncertainty underneath that headline is staggering. At the BNEF report presentation last week, co-author Nathalie Limandibhratha showed a single slide: the gap between the highest and lowest forecast from major research firms is now 100 gigawatts. For context, the entire U.S. nuclear fleet is 98 gigawatts.
"This is quite significant uncertainty."
These aren't small revisions. The forecasters keep getting lapped by the builders. Every time an analyst prints a number, a hyperscaler breaks ground on another campus and makes it stale.
And BNEF isn't alone. EPRI has more than doubled its 2024 estimate. S&P raised its projection by more than a third between October and April. Every expert model keeps breaking against the same wall: AI demand grows faster than the models can track.
The U.S. now has over 3,100 data centers running. More than 1,800 more are in some stage of build. BNEF says even a record pace of grid hookups every year through 2035 won't be fast enough — the shortfall is 19 gigawatts even in the best case.
I've been doing this long enough to know: when the models can't keep up with the real world, the real world is moving faster than the money thinks.
If the best forecasters on the planet can't hold a number for seven months, you should assume every grid stock, every utility plan, and every power plant timeline is still priced too low.
• RESISTANCE
Henrico County Got 37 Data Centers. Now It Can't Afford the Lights.
Henrico County, Virginia, spent years luring data centers. Big tax breaks. Fast permits. Open arms.
It worked. The county now hosts 37 of them.
On July 1, the electricity rate for Henrico's county government and school facilities jumped 24.9%. County leaders estimate an extra $5 million in power costs this year. Their fix? They asked teachers and county staff to unplug phone chargers and turn off the lights.
$18.6 billion in electric and gas rate increases were filed by U.S. utilities in the first half of 2026, per PowerLines — down from the same period last year. More than 56 million customer accounts are in the crosshairs.
Henrico isn't alone. Wholesale power prices near Virginia's data center hubs have climbed as much as 267% in five years. The county that rolled out the red carpet is now rationing watts.
This pattern is spreading. Utilities filed a record $9.2 billion in rate hikes in Q2 alone. The cost of powering the AI boom is landing on family bills coast to coast.
Henrico is the preview. The county gave Big Tech cheap land and fast approvals — and got a 24.9% rate hike in return. Other towns are watching.
• BLACKOUT WATCH
Three Risks Nobody Wants to Price
The AI power trade is crowded. Everyone sees the demand curve. Almost nobody is looking at what breaks if the curve bends.
The BIS just compared AI capex to the dot-com bubble. The Bank for International Settlements — the central bank of central banks — said in late June that combined hyperscaler capex will top $1 trillion from 2025 through 2026. Their word for it: a potential "investment bust." They put it alongside the railway mania and canal boom as historical parallels. That's not a fringe blog. That's Basel.
The 2027 capex cliff is getting steeper. Consensus estimates see hyperscaler spending growth dropping from 84% this year to just 22% in 2027 — though Goldman Sachs argues even that forecast is too conservative and expects higher spending. Some analysts peg it closer to 15%. When growth falls that fast, the supply chain — turbine makers, cable crews, transformer plants — loses pricing power overnight. Stocks that rode the capex wave up will ride it back down.
NERC flagged elevated summer risk in New England and the Pacific Northwest. Both regions face possible supply shortfalls under extreme heat. NERC's spring assessment said fewer areas face elevated risk than last year, dropping from six to three. Meanwhile, over 58 GW of new generation came online across North America since last summer — and it still may not be enough.
The market is pricing the AI power boom as a straight line. History says the line always bends — and nobody rings a bell when it does.
Disclaimer*: Please read the offering circular and related risks at invest.modemobile.com. This is a paid advertisement for Mode Mobile’s Regulation A+ Offering.
Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur.
The Deloitte rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period.
Pro forma revenue and EBITDA, includes full year numbers of the businesses acquired throughout 2025.
