$2.52 trillion in forecast AI spending and $6.7 trillion in projected data center investment through 2030 are rewriting how enterprises allocate capital.1 GPU lead times now stretch 36–52 weeks, constraining deployment even as demand climbs.1
The bottleneck is not budgets — it is infrastructure. Hyperscalers are racing to build capacity, but supply chains cannot keep pace. For CFOs, that means capital commitments must be made quarters ahead of actual deployment.
The shift is also architectural. "Your existing tech stack was designed for human-operated, application-centric workflows," said Surojit Chatterjee in MIT Technology Review. "It needs to be reconsidered when the actor is an AI agent operating at machine speed across multiple systems simultaneously."2
Chatterjee argues the current wave of change is categorically different from prior technology cycles. "Digital transformation was about moving from paper to software. AI transformation was about adding artificial intelligence to existing processes. Co-pilot is about AI assisting in various human tasks. But ABT is something categorically different: It's the integration of AI agents into the fabric of the organization."2
Prasun Shah, also writing in MIT Technology Review, identified where competitive pressure will concentrate: AI agents function as "connective tissue moving across layers to coordinate tasks and contextualize data from multiple applications. That is where the next battleground will be."2
For investors, the financial signals are beginning to clarify. EXL generated nearly $300 million in free cash flow in 2025 and is pursuing a more balanced capital allocation strategy going forward.3 The company's agentic platforms and integrated data-and-operations model are aimed at converting AI experimentation into production use cases.3
The capital allocation question for enterprise leadership is no longer whether to invest in AI infrastructure — it is how to sequence commitments against constrained GPU supply, long infrastructure lead times, and the organizational redesign required to deploy autonomous agents at scale.
Sources:
1 Enterprise AI narrative data, Via News, May 2026
2 MIT Technology Review, May 26, 2026 (Surojit Chatterjee, Prasun Shah)
3 ExlService Holdings, Inc., finance.yahoo.com, May 17, 2026
