Amazon and Alphabet just raised their AI build budgets past $400 billion — and say it still won’t be enough
None of that money earns a dollar until somebody delivers the power.
From the Desk of Robert Rapier Chief Investment Strategist, Utility Forecaster
Dear Reader,
On July 22, Alphabet raised its 2026 capital budget to $195–205 billion and told investors 2027 would climb further still. Eight days later, Amazon lifted its own to roughly $220 billion.
Then Amazon’s chief executive said something worth reading twice. Even at that number, “we will still not have enough capacity to meet all the demand we have in 2026.”
Alphabet posted the first negative free cash flow in its corporate history to fund the build.
These are the most powerful companies on earth. And they are short one thing they cannot code, cannot manufacture, and cannot buy their way to the front of the line for.
Power.
Every watt of it comes from a short list of companies nobody has wanted for a decade — paid on regulated rates and contracts locked years out, not on whether any particular AI model ever earns a dollar.
You don’t need the AI bet to pay off. You need it to continue.
And the contracts are already signed — by companies we have held for years.
A Southeastern electric utility · held since 1994
Signed a 25-year agreement to deliver 3.2 gigawatts to OpenAI’s first self-designed data-center campus — a $20 billion project, phased in from 2028. Announced July 22, 2026.
A Permian-anchored energy major · held since 2018
Signed a 20-year power agreement with Microsoft for roughly 2.67 gigawatts in West Texas — an oil producer turned power supplier, selling straight to a hyperscaler. Announced June 22, 2026.
A Gulf-region utility · portfolio holding
Building seven gas plants, three grid batteries and nuclear uprates to serve Meta’s 5-gigawatt Louisiana campus — a project north of $50 billion, approved by state regulators, and structured to return $2 billion in savings to existing customers. LPSC-approved.
Not forecasts. Signed contracts — with the companies writing the cheques named on them.
Sleep well
beta — about half the market’s swing
Income
yield, and growing
Growth
2026 year to date
Capital spending guidance: Alphabet and Amazon.com Q2 2026 earnings calls, July 22 and July 30, 2026. Contract details as announced by the operating utilities and their counterparties. Yield and beta as of July 30, 2026. Specific holdings are reserved for members. Past performance does not guarantee future results.
II · Why now
Any one of these alone would be bullish. Right now you’re stepping into all three converging at once — the kind of setup that defines an investment cycle, not a quarter.
01
Coal plants are retiring faster than replacements come online. Multiple regions now report sustained 5–7% load growth, with localized pockets in double digits — the first sustained growth in decades.
+5–7% / yr
02
Data centers consume an estimated 4–5% of US electricity and climbing. Hyperscalers are on track to spend ~$1.15 trillion through 2027 — and in July 2026 they raised it again. Alphabet posted the first negative free cash flow in its history to fund the build; Amazon called its booked 2028 demand “striking.”
~$1.15T AI capex
03
PJM’s capacity price cleared roughly 833% higher than the prior year — the total bill going from about $2.2 billion to over $16 billion in three years, with data centers driving ~40% of it. Companies that own firm capacity are being repriced as the irreplaceable infrastructure they are.
PJM: ~833% higher
III · The Trifecta
Most investments give you one. The good ones give you two. We’ve spent thirty-six years building a portfolio that delivers all three — and 2026 is making the case again.
| S&P 500 | Bonds | Utility Forecaster | |
|---|---|---|---|
| Safety | ✗Beta 1.00 | ✓Low volatility | ✓Beta 0.44 |
| Income | ✗1.3% yield | ~~4% (no growth) | ✓4.7% & growing |
| Growth | ~+8.6% (high risk) | ✗Near zero | ✓+14.2% |
The proof
All three at once, month by month — ahead of the market, and steadier getting there.
Higher return, half the drop. The Income Portfolio’s worst decline this year was 4.9% — the S&P 500’s was 8.9%. In March, when the S&P was down 4.4% for the year, the Income Portfolio was up 14.5%.
Safety, income, and growth from the same stocks — for the price of a dinner out.
See the membership plans →IV · Inside the Portfolio
Tickers stay private until you’re a member. But the shape of the portfolio is not a secret — these are real positions, with real entry dates and real total returns since we bought them.
Fee-based midstream giant
Held since 2000. A toll-free operator paid to move energy regardless of the commodity price, with the distribution raised every single year.
Gulf Coast multi-utility
Held since 1994, with a beta near 0.09. A 75-GW data-center pipeline through 2030 and an $81B five-year capital plan, with billions in federal financing secured.
Pure-play water utility
Held since 1994, 5-year beta near 0.10 — the most boring business imaginable, compounded into a thirty-three-fold return.
Coastal Virginia data-center utility
Held since 1989 and still paying 3.8% today. A $65B five-year capital plan, almost entirely tied to accelerating data-center demand.
Telecom infrastructure leader
Held since 1991. The connectivity layer underneath everything else, at a beta of 0.11.
Permian-anchored energy major
Held since 2018. Permian, deepwater Gulf, Australian LNG — and now a 20-year power agreement with Microsoft.
Six of more than 40 active holdings. Total return since our entry date, as of July 30, 2026; specific holdings and tickers are reserved for members. And we show our misses, too: of our last 12 closed positions, 8 were winners and 4 were losses — including one we cut at roughly −37% — because the discipline that protects the dividend means cutting what isn’t working.
The mistake
What everyone buys
The chipmakers, the hyperscalers, the names already up 300%. Crowded, expensive, and priced for perfection. You’re paying a fortune to bet on which AI company wins.
What almost no one buys
The unglamorous companies that own the power, the pipelines, and the grid the whole boom depends on — that already pay you a dividend while you wait, and historically fall less when markets turn.
The demand side is crowded and expensive. The supply side is scarce, essential, and still valued like it’s 2015 — and that’s exactly the inefficiency this portfolio is built to own.
V · The Math of Patience
Buy a $50 stock yielding 4%. Hold it while the dividend grows 7% a year. Three decades later that same share is paying you 30.4% — on your original cost. It is not a trick. It is what our long-term holdings have actually done.
Year 10
yield on cost · $3.94 / yr
Year 20
yield on cost · $7.74 / yr
Year 30
yield on cost · $15.22 / yr
VI · How We Pick
Going forward, the essential-infrastructure sector will not move as one. Some companies will compound through this cycle. Others will lag, and many will cut their dividend before they admit it. Here is what separates them.
Constructive jurisdictions for the regulated names; long-term, fee-based contracts for the midstream and infrastructure names. Without one or the other, capital plans don’t translate into earnings.
Investment-grade credit. Capacity to fund a multi-billion-dollar capital plan without leaning on dividend cuts or shareholder dilution.
Dispatchable generation: natural gas, nuclear, hydro. Storage that scales. Whoever owns firm capacity is being paid premium rates for the first time in decades.
Long-term agreements with creditworthy data center and industrial customers. Regulated transmission pipelines.
You don’t have to guess which company wins. You just have to own the ones that get paid either way.
Start my membership →VII · Membership
Every recommendation passes through the same machinery — the screens, rankings and early-warning models Robert has refined over decades. It’s the difference between a tip and a system, and it’s what you’re really paying for.
All 40+ holdings, every entry price, every live rating — Robert’s exact positioning, ready to mirror. No guessing, no hunting.
The names Robert would buy right now — force-ranked with buy-under prices, five income and five growth. When you have cash to put to work, you’re never guessing where it goes.
Robert in your inbox every Friday — portfolio moves, new recommendations, income strategies and monthly Income Spotlights, so you’re never the last to know.
An 8-point credit check on every dividend — payout, debt, cash-flow coverage, regulatory footing — re-scored monthly against each company’s own industry.
Every month
Owning the portfolio is one thing. Knowing where the next dollar should go is another. Every month Robert force-ranks every company in both portfolios and publishes the top five in each — the Best Buys. The ranking is a composite of three independent sources: Fidelity’s Equity Summary Score, FactSet ratings, and valuation data from S&P Global Market Intelligence. Nothing under a billion dollars in market value is eligible.
Then he overrides the ranking when judgment says to — and tells you he did. He kept one power producer off the list despite a number-three rank, on the view that “we are in an AI bubble that will pop at some point.”
| Year | Best Buys | That portfolio | Difference |
|---|---|---|---|
| 2019 · Growth | +38.6% | +27.6% | +11.0 pts |
| 2024 · Growth | +27.1% | +14.7% | +12.4 pts |
| 2024 · Income | +26.6% | +22.8% | +3.8 pts |
In the three years Robert totaled them, the January Best Buys beat their own portfolio every time. The 2024 Income list did it at a beta of 0.57 — roughly half the market’s volatility.
Figures as published in Robert Rapier’s year-end portfolio reviews. Returns are equal-weighted averages of the five holdings on each January list, held twelve months. Past performance does not guarantee future results.
Members
“90% of your recommendations that I have acted on, I have made money on.”
John Armstrong · Investing 10+ years · 2021 subscriber survey
“If you are a conservative investor and want to experience steady annual gains without the volatility and stress, Utility Forecaster is for you.”
“Robert Rapier avoids the sensationalism of many investing publications. He will admit to mistakes, is willing to change as the landscape moves against positions or industries, and looks to make reasonable returns with a generally conservative approach.”
“I really trust Robert Rapier. I believe he puts the investor’s best interest first, he does extensive research, and he wants the investor to be successful.”
Comments from Utility Forecaster subscribers, 2021 subscriber survey. Star ratings are each subscriber’s own response to “I would recommend Utility Forecaster to my friends and family,” rated 1–5. Individual results vary; past performance does not guarantee future results.
Bonus briefings
A $199 retail value — each one turns the strategy into something you can act on now. All four are included with your one-year membership at the new-member rate.
Report · No. I
Your income portfolio, pre-built and ready to buy.
Report · No. II
Where the 2026 income is headed — and why.
Report · No. III
The supply side almost no one is buying.
Report · No. IV
Turn today’s yield into double-digit income on cost.
Membership
Every membership is a full year of Utility Forecaster — the complete portfolio, weekly issues, and the 90-day money-back guarantee. The difference is what you pay. New members get the first year for $49 — 50% off our regular $99 rate — and all four bonus reports are included. After your first year it renews at our standard $99/year; we’ll always email you first, and you can cancel in one click.
Regular Rate · 1 Year
12 months · our standard membership rate
New-Member Rate · 1 Year
12 months · about $4 a month · 50% off the $99 regular rate
Choose your plan below · backed by the 90-day money-back guarantee
Take ninety days. Read the issues, look at the portfolio, follow a few positions. If it isn’t for you, tell us and you pay nothing — and you keep the reports.
I’d rather you spend your time on the portfolio than on this page. So here’s the simplest version: the structural setup for essential-infrastructure stocks is as favorable today as I’ve seen in 36 years. Ninety days and a no-questions guarantee mean you can verify that for yourself — and if it isn’t for you, you pay nothing and you keep the reports.
If it is — welcome.
Robert Rapier
Chief Investment Strategist · Utility Forecaster
P.S. The stocks everyone’s chasing are the demand side of the AI-power trade — crowded and expensive. Utility Forecaster is about the supply side: the essential companies that get paid no matter which AI company wins, that pay you a dividend while you wait, and that fall less when the market turns. There’s no countdown clock here — just the fact that the repricing is happening now, and early is where the advantage is. Start today, with 90 days to decide.