Everyone chasing the AI theme already found Palo Alto Networks. The stock hit an all-time high last week, trading near 78 times free cash flow.

Equinix doesn't rent server racks. It rents the room where everyone's servers meet.

Same AI infrastructure theme. Nothing like the same price. And 13F season just confirmed it: Stanley Druckenmiller's Duquesne Family Office opened a brand new position here this quarter, not a small one.

The one-line thesis

That opening line is the whole thesis. Slow down on it for a second. A wholesale data center landlord sells raw space and power by the megawatt. Equinix sells that plus the interconnection layer: the physical cross-connects where a bank's trading system talks to an exchange, where a retailer's app talks to AWS, where an AI company's inference cluster talks to its customers. Once a company wires its infrastructure into an Equinix facility, ripping it out means re-plumbing every one of those connections. That is the moat. It's not the building. It's the switchboard inside it.

The number that proves it: Equinix runs a 51% EBITDA margin. Wholesale data center REITs run closer to 40%. Same industry, same physical inputs, an 11-point gap because one company rents megawatts and the other rents megawatts plus the meeting room.

Why now

The first wave of the AI buildout was about training. Massive clusters, built once, running flat out in a handful of hyperscaler campuses. That wave favored wholesale landlords: Meta, Microsoft, and OpenAI didn't need interconnection density, they needed raw square footage and a power substation.

The second wave is inference. Inference workloads sit close to users, plug into multiple clouds at once, and need to talk to the rest of the internet in real time. That is Equinix's exact design brief. Management didn't hint at this. They put it in writing on the earnings call and raised long-term guidance to match: capex of $5 billion to $7 billion a year through 2029, revenue growth of 10% to 13% annually, and AFFO already growing 20% year over year in the most recent quarter.

If negative free cash flow just made you want to close the tab, hold on. AFFO, not free cash flow, is the number to actually watch here. Reported free cash flow is negative right now because Equinix is pouring capital into new capacity ahead of demand it has already signed. That is a reinvestment cycle, not a weak business. It doesn't carry the stale smell of a REIT coasting on a portfolio it finished building years ago. Adjusted funds from operations strips out the accounting non-cash charges and the growth capex, and it is the closer read on what this business actually earns.

The math

Equinix trades at $1,043 a share. 2026 guided AFFO per share: $42.69 to $43.29. Call it $43.

$1,043 divided by $43 is roughly 24 times AFFO.

Chew on that number next to Palo Alto Networks' 78 times for a second. Same theme, a third of the price.

For a REIT growing AFFO 20% a year, with a structural moat and a multi-year demand runway already contracted, 24 times is not a rich multiple. It's closer to a fair one. Compare it to Palo Alto Networks: roughly 78 times free cash flow for 14% revenue growth. The market has already paid up for the AI security story. It hasn't yet paid up for the AI landlord underneath it.

By the time this shows up in a screener next to Palo Alto Networks as "the other AI infrastructure name," 24 times probably won't still be 24 times.

Five scenarios

Super Bull. AI inference demand outruns even the raised guidance. Power and land constraints let Equinix push pricing further. AFFO per share compounds at 25% a year for five years, reaching roughly $131. The market re-rates the stock to 28 times on sustained growth REIT premium. Price target: roughly $3,650. Five-year annualized return: near 29%.

Bull. Growth holds at the current guided pace: 20% AFFO per share growth a year, multiple steady at 24 times. Five years out, AFFO per share near $107, price near $2,570. Annualized return: roughly 20%.

Base case. Growth normalizes toward the low end of management's own 2027-2029 guidance, call it 14% AFFO per share growth a year, with the multiple compressing slightly to 22 times as the market gets used to the story. AFFO per share near $83, price near $1,820. Annualized return: roughly 12%.

Bear. Hyperscalers shift AI infrastructure in-house or to wholesale capacity, bypassing the interconnection layer for at least the training half of their spend. AFFO per share growth slows to 6% a year, the multiple compresses to 18 times, in line with average REIT. AFFO per share near $58, price near $1,035. That's roughly flat to today, five years from now.

Super bear. The AI infrastructure buildout genuinely overbuilds, and Equinix is caught mid-cycle with capex committed and demand that doesn't show up. Growth stalls, the multiple craters to 15 times on a shrunk AFFO base near $38. Management goes on the earnings call and says the word every REIT holder dreads: the dividend is being resized to protect the balance sheet. Price target: roughly $570. That's a 45% drawdown from today, and it is the case that would actually break the thesis.

What breaks the thesis

Two things, watch both. First: does the AI-driven interconnection demand Equinix has already guided to actually show up in the next four to six quarters, or does it stay a slide in an investor deck. Second: does the capex run ahead of contracted demand for too long. A REIT can carry negative free cash flow through a reinvestment cycle. It cannot carry it forever without the market losing patience with the multiple.

The takeaway

Two of the market's smartest concentrated investors bought AI-adjacent names this quarter. One of them, Gavin Baker, already owned Palo Alto Networks and added to it at 78 times free cash flow. The other, Stanley Druckenmiller, opened a brand new position in Equinix at 24 times AFFO growing 20% a year.

Same theme. Very different price to admission. You're reading about the cheap one before it's a consensus screener trade, not after.

If you're tracking this one, the number to check every quarter is AFFO per share against the 20% guide, not the stock price. That's the number that tells you whether the thesis is still intact before the multiple moves. We'll pull that number back up on the show each quarter it reports.

Want to build your own repeatable investment process? Free tools https://www.strategysprints.com/tools

Ready to accelerate now? Book a Discovery Call: https://calendly.com/strategysprint/discovery-call

Also from Simon: The Sales Show (B2B sales acceleration for founders, 800+ episodes). Subscribe https://thesalesshow.beehiiv.com

Happy hunting.
Simon & The Sprinters 🐬⚡️🐆