Why Your 2027 Software Budget Needs Two Raise Lines
Forrester surveyed more than 2,600 business and technology decision-makers for its 2027 Budget Planning Guides. More than 80 percent of leaders expect their budgets to increase over the next twelve months. In the same planning year, Anthropic, OpenAI, and GitHub have already moved some services off a flat subscription onto a meter. Microsoft is selling a richer E7 bundle on top of E5. That is two raise lines in one budget: a higher catalog, and a usage charge.
That’s the problem.
This is not a token-unbundling recap, and it is not a Copilot-meter recap. Those events already happened. Forrester’s public point is that 2027 planning cannot be last year’s SKU mix plus an AI adjective. Vendors will not eat the infrastructure bill. They pass it through as a list-price hike, a usage charge, or both in the same cycle. Do not let the vendor frame planning as “AI is in the product now.” Which raise have you modeled, and which one still sits off the spreadsheet?
What Forrester actually published
Treat the figures as Forrester’s survey, as the 15 July 2026 press release printed them. This is not our book of work. More than 2,600 decision-makers. More than 80 percent expecting budgets up. As many as one-quarter anticipating growth of 10 percent or more. Among technology decision-makers, 82 percent expect an increase.
Sharyn Leaver, Forrester’s chief research officer, said: “The organizations that outperform in 2027 won’t be those that spend the most on AI. They’ll be the ones that invest in the foundations that make AI effective: trusted data, strong governance, organizational readiness, and the ability to continuously adapt as technology and customer behavior evolve.”
A bigger AI line without those foundations is still a raise you cannot defend.
Forrester’s own AI cost-management note is the operational half of the same story. Traditional FinOps struggles with token-based, usage-driven cost. Model training, inferencing, data pipelines, and specialized infrastructure are not reserved-instance math. A planning file that only inflates seats will miss the meter. A FinOps file that only prints tokens will miss the list-price floor that just moved.
E7 is the packaging example, not a SKU teardown. Copilot, Agent 365, and security tools bolted onto E5 prove the catalog can rise while a meter is added elsewhere. The cleanup file already exists. This is the budget forecast.
Do not invent a percentage and call it your uplift. Forrester did not publish your number. Your 2027 model is your current SKU mix against the next list, plus your own usage history against the meters you are already on. Two lines. One budget.
AI cost recovery is a double raise
If that feels familiar, the problem is not your team’s maturity. You have been building a 2027 software budget the way you always have: last year’s SKU mix, a list-price assumption, maybe a true-up. The vendor is now selling a higher catalog and a meter.
This is not a criticism of the teams building those files. Traditional SAM tools collect comprehensively and answer slowly. Neither SAM nor FinOps was designed to recast a double raise into one page a CFO can sign in planning season. The gap is interpretation work.
Picture your procurement director in 2027 budget lock. Finance wants a software number. The CIO has read that more than 80 percent of Forrester’s decision-makers expect spend to rise. The vendor will walk in with a list-price story and a usage story as two conversations. If SAM and FinOps need three weeks to put both on one page, you have already lost the meeting.
An analysis that arrives after the salesperson has named “AI cost recovery” is a document. One that arrives during planning is a decision input.
A “we held the seat price” talking point is not a position. A flat subscription can become a meter without looking like a classic increase. Forrester’s E7 example is the other half: a richer bundle that resets what “standard” costs. You can take both in the same year.
Staffing did not fall because AI arrived
Forrester told leaders to cut AI pilots that scale activity without governance, ownership, or a path to scale, and to fund the foundations that make the rest of the spend usable. That is not a headcount plan.
Do not treat a vendor layoff as your staffing model. The software bill and the people bill can both rise in the same cycle. If your 2027 model assumes AI spend substitutes for SAM, FinOps, or data roles, you are modeling a press release. Write the assumption down. If you cannot defend it in the same meeting as the software number, it is not in the budget.
KPMG’s Q2 Pulse, in a related file, found that nearly a third of leaders still struggle to control AI operating cost at scale. That is the visibility problem sitting under Forrester’s planning year. It is not a second survey to stack. It is the reason a 2027 number without a meter line will not survive first contact with an invoice.
What to model into the 2027 software budget
Do this in the next thirty days, as a planning pack rather than a project. One page the CFO can hold.
| Line you model | Why it is a raise | What you lock this week |
|---|---|---|
| List-price / SKU floor | Vendors increase catalog prices to recover AI cost | Current SKU mix vs the next published list. |
| Usage / token / credit charges | The meter on top of the floor | Your history on those meters. Floor, expected, spike. |
| Packaging that resets the baseline | A richer bundle sold as the new standard: E7 is the example, not the only one | Who needs the bolted-on products vs who moves with the bundle. |
| Staffing you were told AI would replace | Forrester said spend more on foundations, not less on people | Data, SAM, and FinOps roles. Write the assumption down. |
| FinOps controls for 2027 | Model routing, semantic caching, usage guardrails | Funded or not. Off the plan, the meter has no brake. |
Can a non-specialist brief the CIO in fifteen minutes from the pack you have today? If the answer is no, you do not have a 2027 software position. You have last year’s budget with an AI adjective.
The decision layer, not another forecast spreadsheet
You already have a planning file, a seat export, and a FinOps view of usage. The gap is not another inventory. The gap is one 2027 number a person can defend: list price, usage, packaging, and the staffing line you were told would shrink. LICENSEWARE sits on the inventory and ITSM tools you already run. It is not a rip-and-replace SAM suite. It is a decision layer: what matters, why it matters now, what should happen next. That is cost optimization work: list price, usage, and packaging on one page. NEO turns the double raise into a pack a non-specialist can brief. When the model is the commodity, the license question is how Copilot sits as a digital workforce, not which lab trained the weights.
If you are locking a 2027 software budget and your current tools still need three weeks to put the double raise on one page, book a Software Intelligence Review. You can also start on the free plan and run the analysis on your own data.
The question to walk in with
Do not let the vendor frame 2027 as “AI is in the product now.” The right question is: which part of this budget is a list-price pass-through, which part is a usage pass-through, what packaging resets the baseline, and whether the staffing line still moves up after the layoff headlines. That is a data question, not a sales question.
The vendor will walk in with an infrastructure story and a number. The only question is whether you have yours first: current, defensible, and tied to the contracts in front of you.
FAQ
What did Forrester say about 2027 budgets? The public 2027 Budget Planning Guides release surveyed more than 2,600 decision-makers. More than 80 percent expect budgets to rise. Eighty-two percent of technology decision-makers expect an increase.
What is a double raise? Two lines in one planning cycle: a higher list price and a usage charge. Your model needs both.
Why is Microsoft E7 in this story? Only as the packaging example: E5 plus Copilot plus Agent 365. It is proof the catalog can reset. It is not a SKU analysis.
What should FinOps fund for 2027? Runtime controls: model routing, semantic caching, usage guardrails. Traditional FinOps was not built for token meters. That team still has to leap in 2027.
