40% of executives thought AI could save up to 20%. It didn't deliver. - PitchBook
A Bain & Company survey reveals that despite 40% of executives expecting AI to save up to 20% in costs, 40% of companies saw less than 10% savings and many face rising compute costs and inefficiencies like low-quality AI-generated work, yet 90% plan to increase AI budgets amid growing scrutiny from shareholders and analysts over AI’s actual ROI.
Corporates are increasing their AI budgets but are seeing lackluster productivity gains, according to the latest Bain & Company report.
Bain’s Automation and AI Pathfinder Survey 2026, which included 951 respondents, found that 40% of companies tracking their spending recorded cost savings of less than 10% from their AI initiatives. Despite this, 90% of those surveyed whose AI investments underdelivered plan to increase their AI budgets next year.
The report’s authors noted, “In reality, it is a circular bet with a structural leak.”
As compute and usage costs have risen dramatically, executives and shareholders are scrutinizing the returns on these investments. For example, Amazon recently shut down an internal leaderboard tracking employees’ AI activity after discovering staff were running unnecessary autonomous bots to climb the rankings. Uber reported in May that it had exhausted its entire 2026 AI budget within the first four months.
Wall Street analysts are also becoming more critical. A PitchBook analysis of 186 AI-related questions on B2C earnings calls found that 90% were probing and investigative in tone, with ROI and financial returns now overtaking product roadmaps as the main focus.
Employees’ use of AI tools ranges from coding assistants to AI-generated research and organizing unstructured databases. However, this has also led to a new type of workplace inefficiency, termed “workslop” by research from BetterUp Labs and Stanford’s Social Media Lab: low-quality content that appears polished but lacks substance.
AI tools that underperform on productivity while increasing costs present a dual challenge for many AI application startups. They risk high churn from existing corporate users and may struggle to attract new customers.
For startups using these tools themselves, the rising costs of running agentic tasks are even more pronounced.
Despite some corporations reassessing their spending, there is still significant momentum behind AI investment across industries, fueling a boom in AI application-layer startups. For example, Anysphere, the VC-backed company behind the coding agent Cursor, reached $3 billion in annualized revenue in April, driven by enterprise demand from software development teams.