Microsoft 365 Copilot Dropped to $21: Re-Run Your ROI Before the EA Renewal

On July 1, 2026, Microsoft cut the commercial list price of Microsoft 365 Copilot from $30 to $21 per user per month on an annual commitment. That is a 30% reduction on a line item that, at enterprise rollout scale, runs into seven figures. If your Copilot business case was written against the $30 number, it is now stale by a third, and the renewal you signed or are about to sign was priced against a ceiling Microsoft has since moved.

I want to be precise about what changed, because Microsoft attached several conditions to the cut, and one of them raises the base suite you have to license underneath Copilot in the first place.

Try the free calculator (15 seconds, no signup).

What actually changed on July 1

Four facts, all verified on Microsoft's own pricing pages on July 2:

• The Microsoft 365 Copilot annual list price is now $21 per user per month, down from $30.
• There is an $18 promotional rate running through September 30, 2026, available on the first year only. After year one, the seat reverts to $21.
• Paying month to month instead of on an annual commit costs $25.20 per user per month. The annual discount is real and worth locking if your seat count is stable.
• Copilot still requires a qualifying Microsoft 365 base plan. It is an add-on, not a standalone product. The consumer Copilot Pro at $20 per month is a different SKU aimed at individuals, not the commercial add-on your EA covers.

The same day the Copilot add-on fell, Microsoft raised the base suite. Microsoft 365 Business Standard went from $12.50 to $14 per user per month, a $1.50 increase, which is 12% on the underlying seat. Microsoft gave on the AI add-on and took on the suite it rides on. A CFO reading only the Copilot headline sees a 30% cut. A CFO reading the invoice sees a 30% cut on one meter and a 12% increase on another.

Why a 30% cut invalidates the old model

Copilot ROI models are productivity models. Nobody buys Copilot to compress seats. We are explicit about this in the catalog: the Microsoft 365 Copilot tool carries no seat-compression impacts, because a native AI add-on bundled into an existing suite has no clean traditional-SaaS seat to remove. The honest framing is that Copilot is a productivity lever, and the math is time saved per user against cost per user.

When cost per user falls 30%, the break-even hours drop with it. At $30 a month, a seat had to return $360 a year in recovered time to pay for itself. At $21 a month, the bar is $252 a year. That is a different approval decision. Rollouts that got shelved at $30 because the productivity payback was marginal deserve a fresh look at $21, and rollouts already approved at $30 are now over-provisioned in budget terms, which is money you can redirect or hand back.

The discipline here is the same one we apply to every agent deployment. Do not book the full theoretical productivity gain in year one. We use a 0.4 realization factor on deploy actions, and the reasoning is laid out in why deploy realization is 0.4, not 1.0. Adoption ramps, some licensed users barely touch the tool, and the vendor's "up to" figures are ceilings. A 30% price cut widens the margin for error, but it does not turn a productivity estimate into a certainty.

Worked example: 10,000-employee enterprise, 3,000 Copilot seats

Take a company with 10,000 employees that licensed Copilot for 3,000 knowledge workers, a reasonable 30% rollout to the roles where drafting, summarizing, and spreadsheet work dominate the day.

The move from $30 to $21 on 3,000 seats is $324,000 a year off the Copilot line. Take the $18 promo for the first year and you shave another $108,000 on top, landing year one at $648,000, though that seat climbs back to $21 in year two, so do not build the out-years on the promo rate.

The month-to-month trap is worth pricing. At $25.20 versus $21, paying monthly on the same 3,000 seats costs $151,200 more per year than committing annually. If your seat count is stable, the annual commit is free money. If you are mid-rollout and genuinely unsure how many seats stick, the monthly premium is the price of optionality, and $151,200 is what that option costs at this scale.

Now the base-suite offset. Copilot needs a qualifying base plan, and Microsoft raised the reference base rung the same day. SeatCompress models the M365 base seat at the $14 Business Standard list, the rung that moved from $12.50. Applied to the same 3,000 Copilot-enabled seats, the $1.50 increase adds back $54,000 a year. So the net improvement on the 3,000-seat Copilot line is $324,000 minus $54,000, or $270,000 a year, not the clean $324,000 the Copilot headline implies.

Across a full base estate the suite increase is a much larger absolute number. At 10,000 base seats, $1.50 a month is $180,000 a year. At that headcount you are almost certainly on E3 or E5 rather than Business Standard, and those list prices moved on a similar schedule, so treat the $180,000 as directional rather than exact. The point stands: the base creep partly eats the Copilot cut, and only the net belongs in your model.

What the CFO does before signing the EA

If your enterprise agreement renewal is in the next two quarters, this is the sequence.

First, reprice the Copilot line at $21, not $30. If your Microsoft rep is still quoting off the old number, that gap is yours to keep. The list moved; your negotiated rate should move with it or below it.

Second, separate the promo from the run rate. The $18 first-year rate is a real discount, but it expires September 30, 2026 and covers year one only. Model the recurring cost at $21 and treat the promo as a one-time credit, not a permanent rate. A renewal built on the promo number is a renewal that surprises you in year two.

Third, decide the commit term deliberately. The $25.20 monthly rate versus $21 annual is a 20% premium for the right to walk. On stable seats, commit and pocket the $151,200 per 3,000 seats. On seats you are still validating, the premium may be worth it, but name the number so the choice is conscious.

Fourth, net the base increase into the story. When you present the Copilot savings to the board, show the $270,000 net on the deployment line, not the $324,000 gross, and flag the estate-wide base creep separately. The same audit that catches Copilot over-provisioning catches base-suite waste, and dormant M365 seats are their own recovery. The bundled-AI-as-price-rise pattern is not unique to Microsoft; Google is running the same play on Workspace, which we break down in the Google Workspace Gemini bundle post.

Fifth, remember what Copilot is and is not. It is a native add-on priced close to the seat it augments, which is the same structural oddity we walked through in the Agentforce self-compression post, where the vendor sells you the tool to compress its own product. And when your EA rolls up into the broader E5 bundle, the seat-versus-bundle math gets its own treatment in the Microsoft 365 E5 renewal bundle post.

The 30% cut is good news. It is not automatic savings. It is a reason to reopen a model you closed at $30, net the base increase against it, and make sure the renewal in front of you is priced against the number Microsoft is charging today, not the one it charged last year.