Study

AI Compute Could Force 80% of SaaS and AI Companies to Increase Prices

co-written by newsrooms.ai09. September 2026, 13:23

Software as a Service. © Joan Gamell auf Unsplash

A new study by the pricing consultancy hy Consulting Group, the software review platform OMR Reviews and the market research firm Appinio maps how deeply artificial intelligence is reshaping the economics of the software industry. For the SaaS & AI Pricing Report 2027, the partners analysed 4,400 software profiles on OMR Reviews, surveyed 153 SaaS and AI companies and interviewed 23 industry experts.

The pressure to act is immediate. Four out of five of the software and AI companies surveyed are planning or actively evaluating a change to their pricing model within the next twelve months. The reason the report gives is a changed cost structure. Classic SaaS products scaled with almost no additional marginal cost, while every AI interaction consumes real compute and therefore real money. At the same time the source of value is shifting: customers increasingly buy completed work rather than access to a tool.

AI-Native Software Commands Four Times the SaaS Multiple

The shift shows up most sharply in valuations. According to VC funding data cited in the report (median EV/revenue for private rounds in the first quarter of 2026), legacy SaaS platforms trade at a revenue multiple of 5.5x. Products that have retrofitted AI features, described as AI-enabled, reach 8.5x. Purely AI-native architectures command 21.2x, close to four times the SaaS level.

The authors draw two conclusions from this. Investors pay substantial premiums for genuinely AI-native architecture, and even adding AI features to an existing platform produces a measurable valuation uplift. The figures come from sources including Finro, Windsor Drake, Eqvista and Aventis Advisors, and they cover private funding rounds rather than listed companies.

Hybrid Billing Models Take Over Within Two Years

Pure subscriptions remain the most common primary billing model for now, though their share is falling. Currently 40% of respondents rely on them, hybrid models that combine a fixed base fee with variable usage follow at 34%, licence and purchase models account for 22%, and purely usage-based billing sits at 4%.

Looking two years ahead, the picture reverses. A clear majority of 63% expect hybrid models to become the most relevant option for their company, only 22% still see pure subscriptions leading, and 13% bet on pure usage models.

Per-Seat Pricing Loses the Most Ground

The expected shift is even more pronounced at the level of the pricing metric, meaning the unit customers actually pay for. Asked which metric will gain importance in future, respondents put per-user pricing 31 percentage points lower than it stands today. Outcome-based metrics gain 20 percentage points, credits and tokens gain 14, API and consumption models gain 9, and workflow or task metrics gain 3. A quarter of participants have yet to define a target metric at all.

For autonomous agents, however, practice is more conservative than expectation. Currently 31% simply bill agents inside their existing packages, another 31% have no clear approach yet, 16% use credits, 8% each use outcome-based and usage-based models, and 7% sell per agent. Taken together, 62% of vendors either bundle agents into existing packages or have no monetisation strategy in place.

Pure pay-for-performance remains confined to niches with very clear measurability and attribution, the report finds, while task-based and workflow-based billing is establishing itself faster. One structural difference stands out: an AI feature competes for the software budget and is benchmarked against other tools, whereas an agent targets the personnel budget and is measured against FTE costs and daily rates.

AI features themselves reveal a split market. Half of all vendors include AI functionality in the core product free of charge, and 22% do not monetise it at all. More differentiated approaches such as paid add-ons (15%) or credit models (5%) remain the exception.

Product Discovery Becomes the Bottleneck

A separate chapter examines how software gets found in the first place. In the DACH region, 25% of software buyers already use AI systems for their research, and 22% of vendors name AI chat interfaces as a lead channel. Among those surveyed, 41% see LLM-based search as the biggest change in software buying and 16% point to AI agents acting as buyers. Three quarters, 76%, are already adapting their go-to-market approach in response.

The authors tie this directly back to pricing. Because 92% of B2B buyers ultimately choose a product that was on their initial shortlist, and that shortlist increasingly forms inside AI systems, machine readability now determines market access. Vendors who hide prices behind a “contact sales” button do not disappear from comparisons, they simply lose control over which price they are compared with.

A gap in current practice is striking. While 93% of vendors publish primarily on their own website, only 22% publish on review platforms. AI systems, though, validate their judgment largely through third-party sources, and 95% of citations come from non-paid sources according to the report. AI visibility is named as the top priority by 88% of the vendors interviewed, yet not one of them has a dedicated budget for it.

The study finds no universally correct AI pricing model. Its conclusion is that product, packaging and pricing need to be designed as a single monetisation system, and that companies should expect to keep adjusting their monetisation as products mature and customer value shifts.