How to reduce your AI software spend

Nearly 42% of all software spend now goes towards AI, and the share is still rising, according to Vertice’s AI Market Intelligence Index.

But that growth comes at a cost. Average AI-native spend per company has tripled in just 12 months, as vendors continue to add AI premiums and switch to token-based pricing models. And Vertice’s pricing benchmarks show that companies are overpaying for these AI tools by 17%.

Without control over what you buy, what you pay and how much you use, AI spend will keep outrunning your budget.

This guide covers where AI spend currently stands, which tools companies are paying for and how you can keep your AI software and token costs down without losing the tools your teams rely on.

Where AI software spend stands in 2026

AI spend is rising on two fronts: more companies buy standalone AI tools and established SaaS vendors charge more for the AI built into their products. Vertice tracks these as two categories:

• AI-native: AI is the product (for example, OpenAI or Glean)
• Applied AI: Paid AI capabilities added to an established software tool (for example, Microsoft Copilot or Salesforce Einstein),

Four pricing shifts sit behind these numbers:

• Price rises at renewal: AI-native vendors that have proven their value are moving into the “expand” phase, raising prices once customers depend on them.
• AI premiums: Almost 63% of enterprise SaaS products now include Applied AI features, and 79% of those vendors charge an AI premium for it, sometimes doubling the original subscription.
• Token-based pricing: Vendors are increasingly replacing per-user licenses with usage-based pricing, which tracks their own infrastructure costs and pushes spend up across SaaS.
• Shorter contracts: The average AI-native contract runs 15 months, nine months shorter than the average SaaS contract. More renewals mean more chances for price increases.

Overspend on AI-native tools did fall from 22% in Q2 2026 to 17% in Q3, suggesting some buyers are starting to push back. But that only brings it in line with traditional software, and given how quickly AI pricing can escalate, buyers can’t afford to settle there.

The most popular AI tools in 2026

AI budgets are growing faster than any other software category and finance leaders are under pressure to show a return. By following these five steps, organizations can cut waste without cutting the tools that deliver.

1. Benchmark every AI quote and renewal

A 17% average overpayment means many companies pay far more. AI pricing is new and inconsistent, so without peer data you have no way to know whether a quote is fair, and you risk paying a first-mover premium.

Check every AI quote and renewal against what similar companies pay for the same product. Vertice's pricing benchmarks draw on more than 2 million pricing points from 250,000+ contracts.

2. Bring shadow AI under central intake

According to Vertice's Buying AI playbook, 44% of AI spending happens without procurement oversight. Teams sign up for tools on a card, usage grows and the spend surfaces only at renewal.

Route every new AI request through one intake process that checks it against tools you already own and your security standards before anyone commits.

3. Cut shelfware and right-size licenses

13% of AI applications go entirely unused, and nearly half are underused. Waste is concentrated in certain functions and industries:

• HR has the lowest AI usage rate of any department, despite heavy investment.
• Marketing shows middling usage from a tech stack where 20% of tools are AI-based.
• Sales has the most AI-embedded stack of any function but relatively low adoption.
• Healthcare, Education, Construction, Retail and Logistics all spend above average on AI with low adoption.

Track usage per tool and per team, then reclaim or downgrade unused licenses before renewal. Low usage often signals a training gap, so fix adoption before you cut.

4. Consolidate overlapping tools and challenge AI premiums

37% of companies pay for both OpenAI and Anthropic, and many also pay for Applied AI features such as Copilot or Gemini that cover similar ground. Some overlap is deliberate, as teams use different models for different tasks. Make sure it's a decision, not an accident.

When an existing vendor adds an AI premium at renewal, ask what your teams will use. If the answer is "not much", negotiate the AI add-on out or price it on usage.

5. Negotiate AI-specific contract protections

With AI-native contracts averaging 15 months, renewals come round fast. Use each one to lock in protections:

• Remove auto-renewal and cap annual price increases at 3–5%.
• Limit overage charges on usage-based plans.
• Opt out of model training on your data to protect your IP.
• Add performance guarantees, with credits for accuracy or latency shortfalls.
• Agree exit and migration terms in case a rival tool overtakes yours.

Keep every term and renewal date in one contract repository so no termination window is missed.

How to control AI token and usage-based costs

Token-based pricing makes AI spend harder to predict than seat-based SaaS. Vertice's How to Control Your AI Token Costs report found that, compared with seat-based pricing:

• Budgets can swing by up to 40% from forecast.
• Cost per user is 37% higher.
• The average AI contract costs 23% more than an equivalent seat-based contract.

Controlling these costs requires finance, procurement and IT to work together. This involves:

1. Allocating spend by provider, model and team. You can't cut what you can't attribute. Find the teams and models driving cost.
2. Moving low-complexity tasks to cheaper models. Not every prompt needs a frontier model. Review usage regularly and route routine work to lower-cost options.
3. Setting spend thresholds and alerts. Catch usage surges when they happen, not at month-end close.
4. Tracking burn against every commitment. Compare actual spend with committed-use agreements so you know which providers are on track, at risk or over the limit.
5. Forecasting 90 days ahead. Projected overages give you weeks to adjust usage or renegotiate before the bill arrives.
6. Negotiating usage terms. Push for volume discounts, overage caps and milestone-based payments that reflect real consumption.

For more on why consumption pricing drives costs up, read our investigation into token-based pricing models.

Control your AI costs with Vertice

Vertice’s AI Cost Optimization gives finance and IT one view of AI spend, usage and commitments across every provider. The platform specifically:

• Recommends savings, such as switching models for low-complexity tasks, right-sizing commitments and flagging overspending teams
• Breaks down spend by provider, model, department and date range
• Forecasts daily and annual spend 90 days ahead, accounting for weekday patterns and recent usage
• Tracks spend against every committed-use agreement in real time, with on-track, at-risk and over-limit flags

For AI subscriptions and renewals, Vertice benchmarks your pricing against the world’s largest procurement intelligence dataset: over $75bn in processed spend, 250,000+ negotiated contracts and 32,000+ vendors. Customers typically save 20% on software with Vertice and see a 7x average ROI.

See for yourself how Vertice can support AI cost reduction by taking a self-guided tour of the platform or by scheduling a demo at a time that suits you.