Consolidation Is Happening to Your Suppliers, Not Your Portfolio

The renewal that arrives with an AI line item

Picture the renewal meeting. The vendor's account director opens with a slide on new capabilities: a chat assistant in every application, analytics on how people use it, security additions. The price per seat is up. The procurement lead asks whether the organisation can take the suite without the AI. The answer is that the AI is now part of the suite.

Two of the largest suite vendors have done exactly this. Google announced in January 2025 that it would include its AI in Workspace Business and Enterprise plans "without the need to purchase an add-on" and stopped selling the Gemini add-ons³. Microsoft raised list prices across its commercial Microsoft 365 suites from 1 July 2026, with existing customers seeing the new prices at their next renewal, and cited advances in AI, security and IT management as the reason¹². Salesforce announced an average 6% list-price increase from 1 August 2025 alongside new AI add-ons⁵.

The tempting reading is consolidation: the suite swallows the AI, the AI swallows the point tools, the portfolio shrinks and the vendor holds more of the wallet. The public evidence does not show that. It shows something more awkward for a portfolio owner. The base-seat price is rising, and the portfolio is not shrinking.

What moved and what did not

Start with what moved. Microsoft's table of changes is specific: Office 365 E3 from $23.00 to $26.00 (13%), Microsoft 365 E3 from $36.00 to $39.00 (8%), Microsoft 365 E5 from $57.00 to $60.00 (5%), Business Basic from $6.00 to $7.00 (16%) and the frontline F1 licence from $2.25 to $3.00 (33%). Business Premium is unchanged¹. These are list prices. The percentage is largest where the seat is cheapest, which is where frontline-heavy organisations carry most of their headcount.

A composite illustration shows the scale. An organisation with 10,000 Microsoft 365 E3 seats pays $3.00 more per seat per month, which is $360,000 a year before any discount, whether or not any of those people use the assistant. That figure is arithmetic on list prices, not a report of anyone's bill.

Now what did not move. Zylo, a software-management vendor, reports that the average company manages 305 SaaS applications, a fall of 0.07% on the year, while spending on AI-native SaaS rose 108%⁷. Okta, which sees sign-in data across its customers, reports that large enterprises of 2,000 or more employees now average 259 integrated apps, up 5% from 247⁸. Okta's global average fell 3%, from 101 to 98, so some contraction exists among smaller customers. Both companies sell into this market and neither measures a portfolio from the buyer's side, so these are indicators rather than audits. They are, however, the only counts available, and they point the same way: portfolios are flat to rising for the organisations that spend most.

The reason lies in where discretionary AI money goes. Menlo Ventures, itself an investor in AI start-ups, estimates that companies spent $37 billion on generative AI in 2025, of which the application layer took $19 billion, and that start-ups captured 63% of the market, up from 36%⁹. Buyers are paying the suite for the AI it bundles and paying someone else for the AI they chose.

Consolidation is happening to your suppliers

So there are two stories running at once. On the supply side, consolidation is real and visible in the vendors' own results through acquisitions and packaging. On the buyer side, the same enterprises are adding vendors faster than they retire them. The bundle price rise does not purchase a smaller estate. It is a second payment layered on a portfolio that has not been simplified.

This is why the language of consolidation savings deserves suspicion in a business case. A saving exists only when an overlapping tool is switched off, with a date, an owner and a contract end. Bundled AI creates the option to retire a point tool. It does not retire it. Until someone does, the organisation carries both costs, and the vendor's price rise has been funded by the extra seat spend rather than by anything removed.

The mechanism is worth stating plainly, because it explains why the pattern persists. The suite is an anchor: identity, mail, documents and records sit inside it, so refusing the renewal is rarely a realistic option. The vendor folds AI into every base seat and raises the list price at renewal. The price applies to every seat, including those that will never use the AI. Meanwhile the teams that want better AI buy it from specialists, because the bundled version is not the one they chose. Both payments are made, the portfolio keeps its old tools, and the finance function sees total software spend rise on two lines at once. The decision that follows is not whether to accept the bundle but what to demand in return for it.

The strongest case against this reading

The sceptic's position deserves its full force, because it changes what a buyer should do. On this view, bundling is not a show of strength but an admission of weakness. Microsoft said in January 2026 that it had 15 million paid Microsoft 365 Copilot seats among 450 million commercial subscribers, roughly 3% of the base¹⁰. If customers would pay separately for the assistant at scale, there would be no need to fold AI into the seat price. Start-ups take most of the discretionary AI market. Seen this way, incumbents are harvesting an installed base that cannot easily refuse, not extending a lead.

There is a further complication: bundling does not always raise cost. Constellation Research reported that a Workspace Business Standard customer who had paid $32 per user per month with the Gemini add-on would pay $14, which is $2 more than Workspace without it⁴. For heavy AI adopters, the bundle was a saving.

This case is largely right, and it sharpens the conclusion rather than overturning it. If bundling reflects weak standalone demand, the vendor's leverage rests on the anchor function of the suite (identity, mail, documents, records), not on the AI. That makes the renewal the place where the buyer's evidence matters most. A buyer who can show per-group usage of the bundled AI, and which specialist tools outperform it, negotiates from data. A buyer who cannot has accepted the vendor's account of value. Regulators have shown the same pattern from a different direction: the European Commission accepted binding commitments from Microsoft in September 2025 that include offering suites without Teams at a reduced price¹¹. Tying in suites is treated as a matter for competition law, and it can be contested.

What the evidence does not show

The limits matter, because the claim above is narrower than the one many readers will expect.

• List, not net. Every price in the public record is a list price. No source shows the discounts large buyers achieve at renewal, the seats they remove or the tiers they move to. If discounts neutralise the rises, the repricing story largely collapses.
• No measurement of lock-in. Nothing here shows that bundling has raised switching costs, reduced churn or concentrated total spend on anchor suites. The AI spend data point the other way.
• Revenue per user is not attributed. Microsoft reported Microsoft 365 Commercial cloud revenue growth of 14% in constant currency against 6% seat growth in the quarter to June 2026⁶. The gap is about eight points of revenue per seat, but it predates the July increase and Microsoft does not say how much is price, mix or Copilot add-ons.
• Interested parties. Zylo, Okta and Menlo each sell something into this market. Their counts are consistent with one another, not independently verified.

A renewal test for portfolio owners

The practical consequence is a change in what the renewal is for. Treat the AI-bundled renewal as three separate decisions.

1. Price. Establish the net per-seat cost change by tier, and test which low-use groups can move to cheaper tiers. The percentage rises are largest on the cheapest seats.
2. Use. Measure how much of the bundled AI each user group actually uses before the renewal date, and compare it with the specialist tools bought for the same jobs.
3. Retirement. Make any approval of a new AI tool conditional on a dated retirement plan for the tool it overlaps, and track portfolio size and per-seat suite cost as separate metrics.

The distinction worth keeping is between supply-side consolidation and buyer-side simplification. A vendor can have the first without the buyer ever having the second. The test for any renewal is therefore simple and uncomfortable: which tool, on which date, stops being paid for because of this bundle? If nobody can name it, the organisation has not consolidated. It has been repriced.