Computing costs emerge as barrier to agentic AI adoption in software teams
Half of software engineering teams plan to prioritize agentic AI investments this year, with 84% expecting it to lead their spending within three years. Yet only 12% report widespread deployment today, according to a SoftServe report conducted with MIT Technology Review that surveyed 300 technology leaders in December and January.
The gap between intention and implementation points to a concrete problem: computing costs and system integration challenges are holding teams back. Organizations are spending more on computing infrastructure than on hiring to support agentic AI rollouts.
The spending mismatch
More than half of data and analytics leaders are increasing warehouse and compute spending to handle agentic AI demand. Only 36% are expanding their team budgets to match, according to a separate State of Analytics Engineering report from Dbt Labs surveying 363 practitioners and leaders.
The cost burden reflects a wider reality: agentic AI requires substantial computational resources that existing infrastructure often cannot support without upgrades.
Integration and quality concerns
Teams expect agentic AI to eventually handle code generation, testing, refactoring, and deployment across software lifecycles. Within three years, early adopter fields like software engineering anticipate these agents managing entire development workflows.
But technical barriers remain. Data quality and governance present distinct challenges. When traditional systems fail, the failure is obvious. When an AI agent produces a plausible but incorrect answer-like a revenue calculation for a board presentation-the error can spread before anyone catches it.
Data leaders who treat governance as foundational infrastructure rather than an afterthought are better positioned to approve and deploy agentic AI faster and with more confidence.
Timeline for productivity gains
Most technology leaders expect only slight to moderate productivity improvements over the next two years. Teams are betting that agentic AI will eventually accelerate software delivery and product development cycles, but the payoff requires solving integration and cost problems first.
For teams planning to adopt these tools, understanding both the technical requirements and the financial commitment is essential. Learning how AI fits into development workflows can help teams make more informed decisions about where and when to deploy agentic systems.
Your membership also unlocks:
