Bill Ackman disclosed that Pershing Square has built a “core holding” long position in Microsoft (MSFT). He said the position was accumulated after a multi-month pullback, around 21x forward earnings.

It sounds soothing: world-class business, “market multiple” valuation, buy the dip, go back to ignoring it for five years.

But Microsoft isn’t being questioned because investors suddenly doubt Word and Excel. The tension is simpler – Microsoft is spending like an infrastructure company right when the market wants software-company certainty.

Scoreboard (what happened, with numbers)

Here are the hard anchors we can rely on as of today (May 15, 2026): Microsoft’s most recent reported results are for fiscal Q3 2026 (quarter ended March 31, 2026; results released April 29, 2026).

• Microsoft Cloud revenue: $54.5B, up 29% year over year (25% in constant currency).
• Capital return: $10.2B returned to shareholders in the quarter via dividends and repurchases.
• Capex guidance: Microsoft said it expects to invest roughly $190B in capital expenditures in calendar year 2026, including about $25B attributed to higher component pricing.

Those three numbers – $54.5B cloud, $10.2B returned, $190B capex – are the entire argument in one snapshot. Growth is strong. Shareholder returns are intact. Spending is enormous.

The real reason this is even a debate

Ackman’s thesis (as he described it) leans on Microsoft 365 being deeply embedded and underappreciated. That’s fair. The franchise is sticky, habit-forming, and procurement-friendly in a way that most “AI winners” simply aren’t.

Here’s where I’m at: the stock in 2026 isn’t a referendum on whether Microsoft is a great company. It’s a question about timing. How quickly does the AI build-out convert into earnings and cash flow that feel normal again?

Slight tangent, but it matters: for most businesses, capex is a line item. For hyperscalers, it’s the product. When Microsoft tells you “~$190B,” you’re no longer just buying software renewal rates. You’re buying data centers, power availability, specialized chips, and the messy reality of building physical capacity at speed.

Deep dive: what Microsoft is selling now

Microsoft’s money machine is diversified, but it rhymes:

• Productivity & collaboration: Microsoft 365 is still the default for a lot of knowledge work. The key is not “new customers,” it’s higher value per seat over time (security bundles, AI assistants, premium tiers).
• Cloud platform: Azure and related cloud services are where a lot of AI demand expresses itself – both training and inference – and where capacity constraints can become real constraints.
• Adjacencies: Security, LinkedIn, Dynamics, Windows, gaming, search/ads. These don’t all grow the same way, but together they stabilize the overall earnings profile.

Data to watch (not vibes)

• Microsoft Cloud growth: Does that 29% keep holding up as they add capacity?
• Capex follow-through: Do they stick to ~190B for 2026, or does the number creep?
• Capital returns: Does buyback pace stay steady while capex is that heavy?

Also: listen for management language around getting data centers “revenue ready” faster. That phrase matters because it’s the bridge between spending and monetization. It’s encouraging… but it’s not the same as “spending is about to slow.”

Is it cheap?

“~21x forward earnings” is not cheap in isolation. It’s cheap for Microsoft if you believe two things: (1) Cloud demand stays strong enough to absorb the capacity coming online, and (2) the spending wave eventually creates operating leverage again.

If either of those slips, a low-20s multiple can still compress. Great company, not immune stock.

Bull / Base / Bear

• Bull: Azure demand + AI services scale cleanly, utilization stays high, Microsoft 365 monetizes AI features broadly, and investors regain confidence that capex is building future gross profit (not just bigger bills).
• Base: Growth stays good, but capex stays loud longer than people want. Earnings rise, but the multiple doesn’t expand much until spending feels less uncertain.
• Bear: Competition pushes pricing, AI features don’t monetize as fast as hoped, and Microsoft ends up with heavy depreciation and a longer payback period on the build-out.

Action plan (cost-conscious, not heroic)

If you’re conservative: treat MSFT as a hold / slow-add. Scale in across 3–5 purchases, and only get more aggressive if Cloud growth stays strong and you see clearer proof that new capacity is becoming revenue-producing quickly.

If you’re aggressive: you can lean into Ackman’s “core holding” framing, but you should be comfortable owning it through messy quarters where spending headlines overshadow fundamentals. That’s the price of admission.

Cheap Investor scorecard

• Microsoft Cloud growth rate vs last quarter
• Calendar 2026 capex expectations (still ~190B?)
• Share repurchases vs prior quarters (steady or slowing?)
• Any commentary on capacity constraints easing
• Any signs of margin pressure tied to AI infrastructure (gross margin, operating margin)

Bottom line

If Microsoft converts this capex surge into clearly monetized capacity and sustained Cloud growth, then “~21x forward earnings” can look like a rare moment of sanity for a premium franchise.

If the spending stays high and the payoff stays harder to see quarter-to-quarter, you can still be right long term – and still feel early. Worth a look, but don’t confuse “great business” with “easy ride.”