Enterprise Technology Research (ETR) data indicates that enterprise SaaS spending growth is decelerating even as interest in ML / AI solutions soars. Across traditional software categories, from enterprise applications to content management, net spending intention scores are declining, reflecting tighter budgets and optimization efforts. Yet in parallel, ML / AI platforms and tools maintain the highest net scores of any IT sector, with organizations reallocating resources toward AI initiatives. Early evidence in our AI Product Series and panel commentary points to a zero-sum dynamic: enterprises are funding new AI projects largely by squeezing existing software budgets rather than by increasing overall IT spend. Our analysis blends ETR survey data and panel insights to explain the factors behind the SaaS slowdown and how leading software providers (Salesforce, Workday, and ServiceNow) are navigating this transition. We also explore the potential outcomes of generative AI adoption on vendor business models, including seat license churn, shifting pricing models, and limits to expansion.

Key Takeaways

• Second straight decline in our survey. ETR’s July 2025 Macro Views survey pegs software budget growth at 3.5%, as spending plans have dropped by 100 basis points since January (just above October 2024 levels) amid macro‑driven belt tightening.
• SaaS higher on the chopping block than AI. 59% of organizations surveyed in our April 2025 C-Suite Spending Behavior Drill Down who are expecting budget shortfalls planned to cut software subscriptions first, versus only 18% intending to trim AI investments, underscoring wallet rotation toward AI.
• ML / AI retains the strongest Net Score of any IT sector. Despite some year‑over‑year cooling, ML / AI holds a Net Score about four times higher than enterprise applications, based on JUL25 TSIS data, as early‑stage pilots move to production scale.
• Readthrough to Salesforce, ServiceNow, and Workday. Each now touts substantial AI pipelines, yet with core seat growth indications slowing in our Macro Views survey data, higher average revenue per user (not volume), now underpins guidance. Success hinges on users accepting alternate models over legacy seat-based licensing.
• Watch the gap between AI and non‑AI Net Scores. As we look ahead to future ETR surveys, stabilization or convergence would signal equilibrium, while further divergence may imply deeper cannibalization of traditional SaaS budgets.

SaaS Spending Intentions Are on the Decline

Enterprise software spending is experiencing a clear downshift. ETR’s July 2025 Macro Views survey data reveals that organizations’ planned software (On-prem and SaaS) budget growth has fallen an additional 40 bps s/s to 3.5% spend-weighted growth and is down from 4.5% earlier in the year. This marks the second survey in a row showing softer forward-looking software spending, with the decline most pronounced at small enterprises, which have cut expectations by about half since January amid macroeconomic and geopolitical headwinds. In the words of one IT executive, companies are “staying on course with budget, but…shuffling spend between what is more important versus what is nice to have,” with AI and security now prioritized, while other areas see cuts.

Underlying this slowdown is a concerted effort by organizations to optimize existing licenses and rein in seat counts. ETR’s survey of seat-based software usage shows anticipated license seat growth has stalled at just 2.2% over the next six months, down from 3.0% earlier in 2025. Panelists from ETR’s SaaS Headwinds discussion confirmed this trend, as many have moved past the rapid SaaS land-grab phase and into a period of scrutiny and cost optimization. “We realized we were over-provisioning… we went combing through licenses,” said one technology leader, describing how his firm set up FinOps practices to identify underutilized software and ‘throttle’ license counts as needed. Others spoke of downgrading user tiers from full licenses to read-only where possible while consolidating overlapping tools. “Bottom line, we’re trying to trim down the cost on SaaS and reduce the license count,” explained one panelist.

ETR Data. ETR’s July 2025 Macro Views survey asked IT decision makers about total annual software spend and spend‑weighted software budget growth (N = 1713).

License Optimization as Seat Expansion Has Stalled

Rising list prices are colliding with cost‑discipline initiatives. According to ETR polling, 65% of enterprises experienced software vendors hiking fees in the past six months as compared to seven to twelve months ago. Those price moves coincide with CFO‑mandated software audits that aim to rationalize usage. Nearly one quarter of respondents expect no growth in license count in the next six months, on top of one-fifth expecting an outright decline. Several challenges explain this, including continued layoffs in technology and consumer sectors, hiring freezes amid uncertain demand, and a shift toward gig‑contractor or AI agent labor that falls outside traditional seat coverage. 

AI/ML Investment Stands Out Amid the Slowdown

While traditional enterprise software is facing growth challenges, artificial intelligence and machine learning remain a clear exception. ETR survey data consistently ranks the ML / AI sector with the highest Net Score among all IT categories, with significantly more organizations planning to add or increase AI spending than reduce or eliminate it. In our July 2025 survey, the ML / AI Net Score came in at over four times higher than other software categories despite a slight decline versus last year.

Enterprise decision maker commentary reinforces this trend. CIOs and IT executives describe an intentional shift toward AI, often with dedicated budget lines for the first time. One panelist shared that their organization required every department to allocate funding to AI for both new innovations and platform enhancements, expecting vendors to charge more for AI features. Across the board, evaluation and pilot activity has surged, including strong interest in dedicated model providers like OpenAI, Azure OpenAI, Anthropic, along with industry-specific copilots. Even traditionally slower industries such as financial services are embracing AI. As one executive put it, they remain cautious on broader spending but are “very optimistic on the AI side.”

ETR Data: The July 2025 Macro Views survey also asked users about their expected licensed‑seat count growth on a relative basis to their current number over the next six months (N = 1713). 

Reallocating Budgets from SaaS to AI Initiatives

When organizations tighten IT dollars, they are currently reluctant to cut AI-related investments. Conversely, when extra budget is available, AI is a top beneficiary. As Figure 6 illustrates, nearly three-fifths enterprises that expect to spend below initial plan this year are trimming software licensing costs, far more than those cutting cloud infrastructure, headcount, or security. Only 18% said they would reduce AI spend, making it one of the least likely areas to face cuts. A follow-up ETR survey question probed which specific software categories would see reduced spending, with a majority of those organizations planning to dial back on enterprise applications (ERP, CRM, HCM, etc.) and office productivity suites if needed to save money. These are foundational SaaS domains that had seen years of growth but are now being scrutinized for possible pruning. Developer toolchains and data/analytics platforms are also on the list for some, but notably, very few companies plan to cut cybersecurity. This underscores security as a largely non-discretionary area, whereas HR, collaboration, and even CRM might be asked to “do more with less” or defer upgrades.

Generative AI Adoption: Strong Trajectory and Questions Around Future Outcomes

Plotting sector results vs. one year ago show divergent trajectories. ML / AI sustains a Net Score north of 50%, still far above any other sector, while Enterprise applications fell from 17% to 13% and Diversified applications fell slightly to 10%. The widening gap underscores that AI products capture incremental enthusiasm and budget. As they become embedded in enterprise software, several questions will answer whether it expands IT budgets or merely redistributes spend. Three areas we are looking at around this question include:

• Seat Churn vs. Expansion. Per our AI Product Series, AI automation is already reducing reliance on human end-users in areas like support and operations. As AI agents come to handle even more tasks, companies may reduce license counts, leading to more potential seat churn.
• Recurring vs. One-Time Revenue. Much of today’s AI revenue is still exploratory, including pilot projects, proofs of concept, and short-term cloud usage, as opposed to than true annual recurring revenue. The durability of this revenue depends on trial-to-paid conversion, renewal rates, and deal scale-up over the next 12–18 months.
• Pricing Models and Value Realization. Generative AI does not fit uniformly into current SaaS pricing. Vendors are experimenting with per-user fees, usage-based models, and outcome-based pricing. High surcharges can deter adoption, and usage-based plans may create unpredictability. Tiered offerings may bundle basic AI while charging a premium for other features. Vendors that demonstrate clear ROI will gain advantage while others may fold AI into existing licenses.

ETR Data: Our April 2025 C-Suite Spending Behavior Drill Down found that within software spending reductions, Enterprise Applications (72% of respondents) and Productivity/Collaboration apps (59%) are the top targets for cuts, followed by DevOps tools and analytics. Security software (7%) is relatively insulated from cuts, reflecting its criticality (N = 58).

Impact on Leading Horizontal Vendors

Large‑cap SaaS companies are often bellwethers for the broader market and Salesforce, Workday, and ServiceNow (discussed in the full report) each launched multiple AI‑embedded modules over the past 18 months. Their experiences reveal common themes, like how AI revenue is material but still a fraction of core subscription bases, AI tools attach most easily to large enterprises already paying for premium tiers, and how rollouts extend sales cycles as buyers demand clear ROI.

Are Vendor-Developed AI Tools Offsetting or Exacerbating Growth Issues?

With these vendor dynamics in mind, do new AI offerings help the SaaS vendors rejuvenate growth or end up cannibalizing existing sales? The answer so far appears to be a mix of both, dependent on the vendor’s market position and execution. For the market leaders like Microsoft, Salesforce, and ServiceNow, early indications are that AI features provide an upsell opportunity that is adding revenue (at least in the short term). These players have the “land and expand” foothold and deep customer relationships to push new modules or editions with AI. As one panelist put it, “best-of-breed SaaS will benefit enormously [from AI upsells]… players like Microsoft and Salesforce have enormous leverage” now. Customers seem willing to pay a premium for AI capabilities from these trusted platforms, evidenced by Salesforce’s thousands of AI add-on deals and ServiceNow’s big uptick in “Pro+” SKU adoption.

However, even among leaders, customers are cost-sensitive and selective in deploying pricey AI add-ons. Panelists cautioned that they are starting small with AI licenses due to high costs. “We have started investing in Microsoft Copilot, adding selected users… but have not opened up the floodgates… because of the cost,” one IT decision maker stated. His organization chose to provision Copilot for a limited group of developers and then actively looked to cut other software to offset the extra licensing cost. In the Copilot case, the panelist eliminated some QA testing software, reasoning that Copilot’s capabilities made those tools redundant.