Enterprise AI is hitting an unexpected wall: the data organizations built their AI strategies around is increasingly being locked behind vendor fees. A new analysis from Deloitte reveals that software-as-a-service (SaaS) providers are introducing "tollgating" charges, metering systems, and access restrictions on data that companies previously assumed they could use freely. This shift is forcing enterprises to rethink their entire approach to AI architecture, budgeting, and vendor relationships.

What Exactly Is Tollgating, and Why Should You Care?

Tollgating refers to fees that vendors impose for accessing data that enterprises created or stored within their systems. Think of it like this: a company uses a cloud-based customer relationship management (CRM) system to store customer data. In the past, that company could export that data freely for analytics or AI projects. Now, vendors are charging additional fees each time data leaves their platform. Alongside tollgating, vendors are also implementing "token metering," where they track and bill based on how frequently or how much data is accessed.

While data portability restrictions have technically existed in vendor contracts for years, what's new is enforcement. Enterprises that built their AI and analytics strategies on the assumption of open data access are now discovering unexpected costs mid-project. This represents a category of financial risk that most technology budgets and AI business cases never anticipated.

Why Are Vendors Doing This Now?

SaaS companies are under intense market pressure to monetize their artificial intelligence (AI) capabilities quickly. As vendors undergo their own "agentic transformation," meaning they're building AI agents that can autonomously perform tasks, they're looking for new revenue streams. The challenge is that many of these companies are caught between two competing pressures: reassuring customers that they can adopt AI without overhauling existing systems, while also trying to maintain control over data and extract value from it.

This dynamic is creating what Deloitte describes as two emerging enterprise architecture approaches. Some vendors are embedding AI agents directly within their core systems, while others are using application programming interfaces (APIs) and protocols like the Model Context Protocol (MCP) to sit above existing systems. The second approach, which promises more flexibility and interoperability, is ironically where tollgating becomes most disruptive because data must flow between multiple systems.

How Should Organizations Respond?

According to Deloitte's analysis, many C-suite leaders are already rethinking their technology strategies in response to this shift. The key is to make intentional architectural choices now, rather than discovering tollgating fees after you've committed to a vendor.

• Reassess vendor contracts: Review existing service agreements and data access terms carefully. Look for language around data portability, export restrictions, and any mention of future metering or usage-based pricing that could apply to AI use cases.
• Evaluate architecture options: Consider whether embedding AI agents within your core systems (like your enterprise resource planning or ERP platform) makes more sense than building separate AI layers that require constant data movement between systems.
• Plan for hybrid delivery models: Many organizations are combining internal teams with external partners to accelerate AI adoption, but this requires disciplined sourcing decisions that protect your control over core capabilities while using partners strategically.

The Broader Execution Gap in Enterprise AI

Tollgating is just one piece of a larger puzzle. A separate survey of 200 UK chief information officers (CIOs) found that while 79% report strong demand for AI from business teams, the technology function is struggling to deliver at scale. The survey, conducted by Wavestone in late 2025, reveals a significant gap between AI ambition and execution.

The most pressing challenge is skills. Despite AI-ready skills being the top IT challenge for 40% of organizations, only 14% are prioritizing investment in upskilling their workforce. Similarly, data maturity is a leading concern, yet only 23% of organizations are prioritizing investment in data management, governance, and compliance.

Perhaps most telling, 70% of organizations have yet to fully embed AI into their technology operating model, which is slowing their ability to move fast and lead AI transformation. This suggests that many companies are buying AI tools without fundamentally changing how their technology teams work.

"Many organisations treat AI as a technology upgrade rather than the operating model transformation required to embed AI at scale and drive sustained value," noted researchers analyzing the survey findings.

Wavestone, 2026 UK Tech Leaders Survey

What Does This Mean for Your AI ROI?

The convergence of tollgating, skills gaps, and operating model challenges creates a complex environment for enterprises trying to realize return on investment (ROI) from AI. Organizations that are already struggling to prove AI ROI now face the additional burden of unexpected vendor fees and architectural constraints they didn't anticipate.

The path forward requires more than just buying better tools. It requires rethinking how data flows through your organization, how your technology teams are structured and skilled, and how you partner with vendors. Companies that make these strategic decisions now, before tollgating becomes standard practice across their vendor ecosystem, will have a significant advantage over those that discover these costs mid-transformation.

For CIOs and technology leaders, the message is clear: AI is reshaping the technology function today, not in the future. The differentiator is no longer ambition; it's execution. Those that embed AI into their operating model while building strong internal capabilities and making intentional vendor architecture choices will be able to scale and realize sustained value.