Only 30% of healthcare organisations say they have scaled advanced or agentic AI across their operations, according to a Google Cloud survey of more than 250 healthcare executives reported on 16 November 20261. Healthcare sits 13 percentage points below the all-industry benchmark on that measure1.
What the survey found
The obstacles are practical rather than a loss of interest. Nearly 60% of the executives said they had run into integration and deployment problems when moving enterprise AI from a pilot to a live system1. Some 40% said unpredictable costs had led them to pause AI projects1.
Cost also shapes how buyers deal with their existing suppliers. More than 80% of respondents named cost as a main barrier to reducing their dependence on current AI vendors1. About 40% singled out the protection of sensitive data as a particularly important consideration, 9 percentage points above the benchmark for other sectors1. Respondents also said they wanted full sight of how an AI tool is governed and the ability to customise it1. The fieldwork was carried out for Google by the consumer insights firm Hypothesis Group1.
Aashima Gupta, global director of healthcare at Google Cloud, framed the problem as one of economics that only shows up at scale. "You don't see that when you're running a 100-person or 200-person pilot," she said1. Gupta argued that health systems need a discipline for measuring and managing the cost of the tokens AI models consume, and went as far as saying "healthcare needs a chief token officer"1. The report cites Accenture research finding that companies using such strategies avoided a projected 44% rise in token costs over two years1.
What it means for the market
For health system buyers, the figures put numbers on a familiar gap between pilot results and enterprise budgets. A tool that looks affordable for a pilot group may not stay affordable across a whole organisation, and hospitals working on thin margins with federal funding cuts ahead have little room for cost surprises1.
For suppliers, the finding that most buyers feel locked in by cost points to competition over switching and model choice. Google is using the results to back an approach built on different models for different tasks rather than a single model1. Its Gemini for healthcare product lets users combine models from Google, OpenAI and Anthropic under one governance layer, and Google says that open approach distinguishes it from Anthropic and OpenAI, which have their own healthcare tools1. "Healthcare is not one workflow," Gupta said1.
Limits of the evidence
The survey was commissioned by Google Cloud, a supplier with a product to sell into this market, and the source gives no breakdown by country or type of organisation1. It reports what executives say about their own progress, not measured adoption. The source does not give the survey's dates or how respondents were selected, so the 30% figure is best read as a self-reported snapshot from a little over 250 leaders1.
Sources
- Healthcare struggles to scale AI past pilot stage: Google report (opens in a new tab)Healthcare Dive · healthcaredive.com
This briefing summarises publicly available research and reporting for information only. It is not medical, investment or legal advice. Follow the sources listed for the primary research and reporting.