Block's Q4: workforce cut by more than 40% — Jack Dorsey's 'intelligence-native' bet lands on Square sellers

Friday-morning coffee, two screens, the Block release on one and the call transcript on the other. The number I underlined twice wasn’t the gross-profit beat. It was the line in the middle of the prepared remarks: Block, going forward, runs with fewer than 6,000 employees, down from a little north of 10,000 at year-end. That is more than a 40% cut to the headcount, effective immediately, announced alongside an earnings print that itself was good — Q4 gross profit $2.87B and up 24%, full-year $10.36B and up 17%, Q4 adjusted operating income $588M and up 46% (Motley Fool transcript, Feb 27 2026; AOL summary of the release, Feb 26 2026). Companies that cut 40% of the workforce do not usually do it on top of a beat-and-raise. Block did.

So this isn’t a layoff story dressed up as a transformation story. The cost line will follow — Amrita Ahuja told analysts the operating-income benefit lands “primarily in the second half of 2026” (Motley Fool transcript) — but the cost line is not why Dorsey did this. Read his own phrasing: Block will be “significantly more valuable as a smaller, faster, intelligence-native company.” Intelligence-native is the operative word. The thesis is that the tooling has gotten capable enough — Dorsey said models became “an order of magnitude more capable” in December — to change the shape of the org chart itself. That’s a bet on how the firm operates, not on what it spends.

I want to model what intelligence-native means for the actual paying customers — Square restaurant sellers — because that’s where this bet either compounds or collapses over the next four quarters.

The print, briefly, so the rest of this column makes sense

The Block release Thursday after the close gave the Street four numbers worth holding in your head together:

• Q4 gross profit $2.87B, up 24% year over year.
• FY 2025 gross profit $10.36B, up 17%.
• Q4 adjusted operating income $588M, up 46%.
• 2026 guide: $12.2B gross profit (~18% growth), $3.2B adjusted operating income (54% growth), $3.66 adjusted diluted EPS (54% growth) (Motley Fool transcript; AOL, Feb 26 2026).

Square specifically: gross payment volume up 10.3% in Q4 (Motley Fool transcript). Square gross profit growth was more modest — roughly 7% in the quarter — which is the number the bears will lean on. Cash App monthly actives 59M as of December, with primary-banking actives 9.3M, up 22% (AOL, Feb 26 2026).

Now the part that doesn’t look like an earnings line: Block will operate going forward with fewer than 6,000 employees, against roughly 10,000-plus at year-end 2025 (AOL, Feb 26 2026; the company’s own framing as reported in the transcript). That’s the headline structural change. The cost benefit hits late in 2026. The organizational change is immediate.

Why this is not the usual tech-layoff

The 2023-2024 pattern was different: cut to hit a margin number, frame it as “focus,” return to the same operating model with fewer chairs. Dorsey didn’t describe that. The language on the call:

“Intelligence tools have changed what it means to build and run a company.”

“Intelligence will be at the core of how the entire company works.”

“We can actually prompt our customers, and we can do it at the right time.”

That last sentence is the one a Square seller should circle. The first two are organizational claims; the third is a product claim. The same intelligence layer that lets Block run on roughly 60% of its prior workforce is the layer Dorsey says will let Square sellers get prompted by their POS at the right moment — and that’s the moat thesis.

The internal-productivity data backs the org claim with one hard number: Ahuja said “greater than 40% increase in production code shipped per engineer since September” (Motley Fool transcript). Six months, a discipline where outputs are countable. If half of that gain is real and durable — assume normal AI-tooling decay — engineering runs at well under 60% of prior headcount and ships more code. Apply half the leverage to support, finance, ops, and you arrive near Dorsey’s workforce number.

Judgment, flagged: I’ll underwrite the internal productivity claim because the measurement is concrete. I’m not yet underwriting the external seller-facing intelligence claim at the same confidence — the product is days old.

Square AI is live in every market. That’s the unlock — and the trapdoor.

Block said on the call that Square AI has launched to all markets (Motley Fool transcript). Not a pilot — the rollout. The next-tier capability Dorsey named is ManagerBot for sellers, with a Cash App analog called MoneyBot: the proactive intelligence layer that doesn’t wait for the seller to open the app, but pings them when a number is moving the wrong way.

What does that look like on a restaurant floor at its functional ceiling? The POS notices Tuesday-dinner labor running 240bps above your trailing four-week average by 6:45 p.m., covers tracking 9% below forecast, and pushes a one-line note to the manager-on-duty: cut a runner, push the bar promo, absorb the night. Today that judgment lives in a manager’s head. Tomorrow, if ManagerBot ships as Dorsey describes, the judgment is software and the manager is the executor.

If you operate on Toast you’ve heard most of this — I covered Toast IQ Grow and bundle compression, and the Toast IQ launch arc. The Toast pitch is “bundle the marketing agent for $499/month and squeeze the point solutions.” The Square pitch as of Friday is different in kind: the operating intelligence is the platform, included with the rails. No $499/month surcharge announced, and Square historically bundles into existing fees. If that’s how this lands — pricing detail unconfirmed — Square’s small-format restaurant take-rate dynamic shifts meaningfully.

The trapdoor: a 40%-smaller workforce is a thinner customer-success bench. If ManagerBot ships before support staffing catches up, the failure mode is sellers getting prompted by a model they can’t reach a human about. Watch Q1-Q2 support-NPS commentary as a leading indicator. Dorsey didn’t speak to that on the call.

What it means for the M&A read

Three things, in order of confidence.

One: Square’s restaurant SMB book just got more defensible, faster than consensus thinks. I’ve argued for months that strategic acquirers — DoorDash, Thoma Bravo via Olo, Amex via Resy/Tock — pay the platform premium and operators cannot. See the deal-pattern read and Oliver’s buy-side counter. What changes: Block is now explicitly playing the same game — buy/build the intelligence layer, include it in the rails, lock the operator in via take rate rather than feature gating. If Square AI plus ManagerBot ship on Dorsey’s cadence — visible to sellers in 2H 2026 — Square’s restaurant book gets harder to migrate off, and the LBO on any Square-anchored group needs a higher re-platforming reserve. Add a unit to my $5–15K-per-unit reserve from the buy-side note.

Two: the workforce-cut math is the new comp for “intelligence-native” CIM pitches. Every banker over the next two quarters will reference Block. Some honestly — the asset re-architected and can show a concrete per-function productivity number the way Ahuja showed +40% engineer velocity. Most as theater. The diligence question: show me the per-function output metric six months before and after the tooling change. If they can’t, the “intelligence-native” line is decoration.

Three: the modest 7% Square gross-profit number is the bear case, and it’s not nothing. Square AI is everywhere, GPV grew 10%+, but Square gross profit grew about 7%. Transcript framing is mix and launch costs, with operating leverage showing up later in 2026. The skeptical read: Square’s core seller economics are GDP-plus, not the AI-platform multiple the stock now implicitly carries. Bull and bear cases both turn on 2H 2026 Square gross-profit re-acceleration.

The 2026 guide forces the question. $12.2B gross profit, $3.2B adjusted OI at 54% operating-income growth cannot work without (a) the headcount cost savings materializing on schedule and (b) Square gross profit re-accelerating. Both have to land. If only the cost piece lands, the multiple compresses because growth didn’t show up. If only growth lands, the multiple compresses because the org bet didn’t pay for itself. Dorsey has to deliver both in the same year with 40% fewer people.

What I’d watch over the next two prints

Concrete checks, in the order I’d run them on a desk:

1. Square gross profit growth in Q1 and Q2 2026. The Q1 guide is $2.8B gross profit at the company level, +22%. Square’s contribution inside that number, and whether the segment growth rate breaks 8-9% on a clean comp, is the single best read on whether intelligence-native is showing up in seller spend.

2. Square AI adoption disclosure. Block has not committed to a specific MAU-style metric for Square AI, but the natural one is the percentage of active Square restaurant sellers using the assistant week-over-week. If it gets disclosed and it’s anything north of 30% by mid-year, the moat thesis is real.

3. Support-NPS commentary on the Q1 call. Cut 40% of the company, ship a proactive-intelligence layer, and the surface area where it breaks first is seller support. If Q1 commentary is silent on support quality, I get more cautious, not less.

4. The ManagerBot ship date. Dorsey framed ManagerBot as under construction, not as shipped. If it slips past Q3, the 2H 2026 operating-leverage story has a hole in it.

5. Engineer-velocity update. Ahuja gave the +40% number with a September baseline. If that figure is still climbing in the Q1 call, the internal productivity story compounds. If it’s flat or rolled back, the org bet looks worse than it does today.

The trade for the indie operator

You’re not buying Block stock. You’re either on Square or you’re not.

If you’re on Square, the cost of staying just went down in expected value — the intelligence layer is included and seller-success investment is now the platform’s problem. The risk is migration: the longer Square AI and ManagerBot are live, the more your operating data sits inside them and the harder a swap to Toast or Stripe-Lightspeed becomes. Lock-in by velocity.

If you’re shopping, the comparison just sharpened. Toast bundles features at $499/month. Square is betting the platform includes them. The diligence question is no longer “which has the better feature set today” — both will converge inside twelve months — but “which is the cheaper place to be when both have shipped ManagerBots?” That math may not favor the incumbent in your market.

If you’re a buy-side underwriter looking at Square-anchored groups: raise the re-platforming reserve, raise the data-asset value, discount the “AI tooling” CIM line the way I discount it on Toast-anchored groups — 30-50% on the productivity claim, because the platform captures some of it via take rate.

What I’m not doing yet: putting a premium on Block-stack groups at sale. Dorsey announced the bet Thursday. The print on whether it worked lands two quarters out. The Bottom Line will read both numbers when they come.

— Marcus runs The Bottom Line and gets the deal flow before the brokers. Tips: [email protected].