The press releases came on time when CVS Health won Silver for Customer Experience AI and Gold for Enterprise AI at the 2026 Stevie Awards in late April. The stock had already increased by almost 30% over the previous year, and it was currently trading at about $81. The familiar drumbeat of a healthcare giant announcing to the world that it had figured out how to make artificial intelligence work at scale was accompanied by congratulations from industry analysts and a few LinkedIn posts from company executives. A clear explanation of how all this AI is actually affecting the bottom line was more difficult to find in any of it.
The intriguing questions reside in that gap between financial clarity and recognition. One of the most ambitious AI initiatives in American healthcare has been developed by CVS Health. Its pharmacy workflow tool, which improves medication safety by interpreting prescription instructions using open-source large language models, won a Newsweek AI Impact Award in May.
The company has implemented fraud detection models in its insurance division, robotic automation in its supply chain, and an AI-driven personalization engine that it says has increased medication adherence. In order to teach tens of thousands of employees how to use the technology, HR and technology leadership collaborated to launch an internal AI Learning Academy. It’s quite a bit. It’s another matter entirely whether it’s sufficient to support investor confidence.
A portion of the challenge is structural. CVS runs a vertically integrated business that includes retail pharmacies, pharmacy benefits through Caremark, insurance through Aetna, and an expanding primary care footprint through Oak Street Health and Signify Health. The company has committed to investing $20 billion in technology over the course of the next ten years. While this amount may seem high, it is quickly absorbed by an organization that serves more than 100 million people. In general, investors appear to support the vision. They haven’t received a detailed accounting of the boundaries between measurable margin improvement and AI spending. Efficiency and operational improvements are mentioned in earnings calls, but the wording is sufficiently ambiguous to avoid serious modeling.
Then there is the controversy surrounding digital twins, which arose earlier this year when CVS collaborated with Simile, a startup, to produce artificial intelligence (AI) copies of actual customers for market research. The company’s VP of enterprise customer experience praised the synthetic responders for being “always on,” praising their capacity to ask an endless number of questions without experiencing human fatigue. There was less enthusiasm from critics. One called the 95 percent accuracy figure a measure of data recall rather than true human insight, pointing out that CVS had validated the AI twins against known findings, essentially asking the model to replicate what it had already learned. CVS retorted that the system was intended to comprehend decision-making processes rather than merely replicate historical behavior. Although it’s still unclear who is correct, the discussion showed that CVS is willing to use synthetic data in ways that could subtly change how the company views its own clientele.
It seems like CVS is creating something truly formidable, putting together operational tools and data assets that rivals like Walgreens Boots Alliance or even Amazon would find difficult to swiftly imitate. On paper, the closed-loop system that links clinical interactions, pharmacy fulfillment, insurance claims, and retail behavior is precisely the kind of flywheel that makes AI effective. However, implementing AI at this scale is costly, complex, and fraught with integration risk—especially when you’re trying to stabilize a health benefits segment that has experienced margin pressure while also absorbing multibillion-dollar acquisitions.
You wouldn’t know any of this was going on inside a CVS today if you were walking by. The aisles have the same appearance. There is still a line at the pharmacy counter. However, algorithms are processing patient data through models trained on billions of claims, flagging possible drug interactions, and parsing clinical terminology behind the registers and prescription bins. It’s possible that in five years, this will be seen as the turning point in CVS’s success. It’s also possible that the awards will outlive the returns. For now, investors are waiting for someone to show them the numbers while they watch the trophies pile up.
