The Services Budget Is AI’s Biggest Prize

The Services Budget Is AI’s Biggest Prize

Enterprise software budgets are large. Enterprise services budgets are larger. A company might spend $10,000 licensing accounting software and 10 times that on the firm that closes the books. For most of the last decade, artificial intelligence companies were building for the software seat. The larger bet is the services contract.

A new generation of AI companies is skipping the tool sale entirely, deploying autonomous agents that run back-office finance, insurance brokerage, legal drafting and revenue cycle operations with minimal human oversight. The prize is substantially bigger than the market they came from. For every dollar companies spend on software, they spend roughly six on services, Sequoia Capital noted.

The categories most vulnerable to substitution share a common profile: high volume, rule-based execution and vendor relationships already structured around outcomes rather than hours. Insurance brokerage, healthcare billing, legal transactional work, accounting and IT managed services all fit. If work is already outsourced to a third party, a budget line exists, the buyer is purchasing a result, and swapping one vendor for an AI-native provider is a procurement decision rather than a corporate restructuring.

The contracts that govern those relationships are already changing. PYMNTS reported in that contracts for agentic AI deployments increasingly resemble managed services agreements rather than traditional SaaS subscriptions, with performance measured by outcomes like accuracy rates and financial error rates rather than uptime percentages. Vendors are pricing by transactions processed, workflows automated or a percentage of savings generated. CIOs are renegotiating agreements to add clauses around data use, IP ownership and productivity-sharing provisions that prevent providers from capturing efficiency gains internally.

Gartner predicted that 40% of enterprise applications will embed AI agents by the end of 2026, up from 5% in 2025. The research firm projected agentic AI could drive 30% of all enterprise application software sales by 2035. The companies winning enterprise AI contracts are betting the money is somewhere else.

Capturing that market requires a different sales motion than enterprise software has used. It requires technical staff embedded inside client organizations, people who understand specific workflows well enough to rebuild them around an AI agent. That role now has a name and a proliferating job description.

Advertisement: Scroll to Continue

Forward-deployed engineers sit inside client operations, build evaluation frameworks that catch errors before they reach production, and translate between what a model can do and what a business process actually demands. Enterprise AI pilots fail the vast majority of the time not because the models are weak but because deployment is broken. The forward-deployed engineer exists to fix the deployment side of that failure rate.

PYMNTS reported in March that monthly job listings for the role grew more than 800% between January and September 2025. The forward-deployed engineer exists to fix the deployment side of enterprise AI’s failure rate. OpenAI and Anthropic each launched enterprise deployment ventures in May, backed by private equity and built around engineers placed on-site at corporate clients.

PYMNTS reported that Anthropic crossed a $30 billion annualized revenue run rate in April, up from $9 billion at the end of 2025, driven primarily by enterprise and startup API usage. More than 300,000 business customers accounted for roughly 80% of that figure. The companies at the top of that spending distribution are not buying access to a model. They are replacing a services contract, and the dependencies that come with a rebuilt operation make renewal a structurally different conversation than a software seat.

Volume automation is the entry point. The durable business is built on the judgment layer, the cases that require context, history, and institutional knowledge that takes years to accumulate. That is where the services budget lives, and it is what every major AI firm is now trying to reach.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.