AI Could Increase Enterprise IT Budgets 75% by 2033

AI's true cost emerges in infrastructure, security, and staffing

Bain & Company projects artificial intelligence could drive enterprise IT costs up by as much as 75% within the next decade, even when organizations invest strategically. The forecast, first reported by CIO Dive, points to a convergence of cost pressures that procurement and IT operations teams are already navigating: expanded infrastructure requirements, broader security surfaces, and competition for scarce technical talent.

The budget impact is showing up in three operational areas that leaders can address now: how AI workloads are architected across different model types, how autonomous AI agents are governed and certified, and how long organizations must wait to deploy hardware. Recent moves by Deloitte, KPMG, the U.S. Department of Transportation, and Dell illustrate how these pressures are playing out in practice.

Why it matters

The 75% cost increase is not speculative—it reflects budget pressures already visible in vendor contracts, capacity planning, and governance workflows. Organizations that treat AI as a contained line item will face surprise costs in storage, networking, platform engineering, and incident response. Understanding where the money actually goes gives procurement and IT leaders leverage to negotiate better terms and build more realistic roadmaps.

Multi-model strategies create contract complexity

Deloitte is launching an open model engineering practice to help enterprises balance proprietary and open-source AI models, CIO Dive reported. This hybrid approach treats model selection as an ongoing portfolio decision rather than a single vendor commitment.

For procurement teams, multi-model strategies introduce contract sprawl unless governance is designed upfront. Organizations need separate agreements for hosting, fine-tuning, data retention, evaluation tools, and safety controls. The expensive work often lies in the integration layer—model gateways, logging systems, policy enforcement, and repeatable evaluation frameworks—rather than in the model licenses themselves.

Autonomous agents push assurance into procurement

As AI tools move from generating content to taking actions, buyers are demanding proof of built-in guardrails. KPMG obtained certification for its agentic tool covering security and reliability, CIO Dive reported, as enterprises confront growing risks from AI systems that can act without human approval.

Certification gives procurement teams concrete negotiation points: defined control objectives, test artifacts, operational limits, and escalation procedures. Implementation work shifts toward governance workflows—approvals, auditing, change control, and role-based permissions that align with how the business operates.

Hardware backlogs turn deployment into a scheduling problem

Dell reported a $95 billion AI infrastructure backlog, CIO noted, with shortages spanning servers, storage, and related components. Supply constraints are lengthening timelines, complicating refresh cycles, and forcing interim architectures.

Organizations are extending rental periods, standardizing on fewer configurations to secure priority allocation, and treating infrastructure procurement like supply chain planning—with lead times, approved alternatives, and pre-authorized substitutions. The operational advantage goes to teams that plan for delays rather than react to them.

Federal continuity signals operational priorities

The U.S. Department of Transportation named Jack Albright acting chief digital and information officer and deputy CIO for IT shared services, FedScoop reported. Albright, who has served as DOT's deputy CIO since December 2020, succeeds Pavan Pidugu.

The appointment underscores how AI-era planning depends on shared services foundations—identity management, endpoint security, collaboration platforms, and data center operations—that determine whether AI can be deployed safely and repeatedly across agencies or business units.

Questions for AI planning now

When vendors offer agent capabilities, what documented limits exist on actions, approvals, and audit logging—and are those controls included in base pricing? For hybrid model strategies, what is the exit path if a model changes price or policy, and which integration layer will own routing and evaluation? If hardware lead times slip, what is the approved fallback, and which team owns that decision?

These details were first reported by CIO Dive, FedScoop, and CIO, drawing on analysis from Bain & Company and developments at Deloitte, KPMG, Dell, and the U.S. Department of Transportation.