A study released by Google Cloud late last year offered some eye-opening insights into where life sciences and healthcare executives are using artificial intelligence (AI), where they are not, and most importantly, where they are finding a return on investment (ROI) from their AI spending.
For years, drug developers have long cited drug development purposes as key reasons for using AI—such as finding challenging drug targets, training large language models for drug discovery, facilitating biomanufacturing, and slashing the time and expense of R&D.
Shweta Maniar, global director, life sciences strategy & solutions for Google Cloud
However, Google Cloud’s report, The ROI of AI in Healthcare and Life Sciences, found “productivity and research” to be only the second most frequent purpose for using AI agents at 39% of executives surveyed. Marketing led the list of purposes with 41% of execs surveyed, tied for the top percentage with tech support. (Respondents could list multiple uses.)
Marketing finished second, though, in generating ROI from agentic AI, with 27% of executives citing that activity compared with 28% citing product innovation and design. Automated document processing was third at 26%.
“This mix of broad and industry-specific use cases sets the life sciences industry apart from other industries—and shows how the most substantial ROI opportunities for the life sciences industry lie in core business functions,” observed Google Cloud, which provides AI as well as cloud computing services.
Adopting agentic AI for core life sciences processes such as quality control (done by 37% of respondents), automated document processing (36%), and supply chain risk identification and process augmentation (33%) “has the potential for truly differentiating impact in the future,” the company added.
Yet users seeking to integrate AI into these use cases will find significant legal and technical hurdles, Google Cloud noted, since those uses involve handling patient health information and proprietary research data, which are subject to additional regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States and the European Union’s General Data Protection Regulation (GDPR).
AI-driven growth
Google parent Alphabet credits AI with enabling sharp growth at Google Cloud, in part through recently-launched products such as Google’s most advanced native multimodal AI model family, Gemini 3, and its most advanced image generation and editing model, Nano Banana Pro.
“Overall, we’re seeing our AI investments and infrastructure drive revenue and growth across the board,” Alphabet CEO Sundar Pichai told analysts on the company’s quarterly earnings call.
AI is also driving a leap in Alphabet’s projected capital expenditures for 2026 to between $175 billion and $185 billion, roughly double the $91.447 billion spent in 2025. Alphabet can make that sort of investment, having finished 2025 with revenue that rose 15%, to $402.836 billion from $350.018 billion, and a 32% year-over-year jump in net income to $132.17 billion from $100.118 billion in 2024.
Alphabet does not break out results for the AI Infrastructure and Generative AI Solutions businesses within Google Cloud. Also, some AI spending—such as specified AI-focused shared R&D activities, including development costs of general AI models—falls outside Google Cloud because it is carried out by the parent company and is thus folded into its “Alphabet-level activities” segment.
During the fourth quarter, Google Cloud operating income nearly tripled, zooming 154% to $5.313 billion from $2.093 billion, on revenue that leaped 48% to $17.664 billion from $11.955 billion. Google Cloud ended 2025 with a staggering $70 billion annual estimated future annual revenue or “run rate,” and a backlog of orders waiting to be processed that grew by 55% quarter over quarter to $240 billion.
Google Cloud and National Research Group based the survey on interviews with 605 leaders of global life sciences and healthcare companies, of which 176 were CEOs or CIOs. All executives surveyed were in companies with 100+ employees and more than $10 million in annual revenues. Nearly 40% of respondents were based in the United States.
Shweta Maniar, global director, life sciences strategy & solutions for Google Cloud, recently discussed the survey and the broader challenges faced by life sciences and healthcare users in applying AI effectively for maximum ROI with GEN Edge.
This interview has been lightly edited for length and clarity.
GEN Edge: Why did Google Cloud commission the survey?
Shweta Maniar: Part of the purpose of us doing this entire survey was this idea that current drug development is a high-stakes, manual marathon that we’ve seen fail 90% of the time. We’re seeing the industry at a breaking point. And we see that technology players have a role to play in this opportunity. We are drowning in information; there is an opportunity for us.
I’ve been using the metaphor that it’s like being lost at sea, where we’re surrounded by saltwater. There is a mass of data, but none of it is potable. So, we’re trying to boil the ocean and trying to find a few drops of freshwater. 2026 is going to be the year when we’re going to be moving from reactive R&D to proactive research and prediction—R&P—using AI. And AI is going to be the way that we’re going to be able to make biological information usable or drinkable.
The idea is, instead of scientists trying to drink from the fire hose using this fragmented data, agentic AI is actually going to help us make these plans and execute workflows, looking at—really thinking about moving the human role from data entry to high-level strategy. It really is thinking about how we shift the human mindset from being an author to being a reviewer. Again, removing the manual entry work from the high-level strategy. We’re really finally going to be able to, this year, think about how we can replace some of that brute force lab testing.
GEN Edge: How reliant should life sciences companies be on their AI?
Maniar: I have personally been in translational research as well as in bench research. I personally know the hours and hours of manual pipetting labor that go into research. I do want to emphasize that “because AI said so” is not the answer. Trust is going to be everything. This is not going to be about a black box. This year, in 2026, the standard for AI is going to be about traceability. It’s going to be about trust.
And again, I’m using this metaphor: We have to show that the water is pure, or the water is drinkable. That’s actually where we have the opportunity for agentic AI to shine through. With agentic AI, this opportunity of creating an audit trail for every single decision is essentially a recipe because you can follow it, you can replicate it. Every design choice, or every drug candidate, can be linked back; it’s verifiable. So that’s no longer “because this researcher said so,” or “this AI said so.” It’s verifiable, data-backed conclusions.
I think that similarly, that will then allow the idea of simulating drug reactions digitally first, so that if we fail, we can actually really fail fast and fail hopefully less expensively. It’s really ensuring that we have the compounds with the best chance of success to actually move forward into the more time-consuming and more resource-consuming steps. This is going to be, I think, part of how we’re going to see the use of AI and agentic AI really get adopted in these high-stakes environments moving forward.
GEN Edge: What regulatory considerations do life sciences companies need to keep in mind as they partner with tech companies such as Google Cloud to deploy AI?
Maniar: Part of this survey was also that this is not just between the tech companies and the life science organizations. There is a regulatory aspect to this. The most significant bottleneck, I think, is going to be that last-minute scramble for regulatory approval. And there is an opportunity for us, or for AI, that’s poised to turn compliance from a hurdle to something that can be maybe more of a proactive, real-time process.
It is going to be a journey. It’s not a switch that’s going to be flipped on overnight. However, there are these tools that now exist, both regulatory bodies and the life science ecosystem—and we want us trying to work together—to try to understand how these tools can actually inform how we can get the right medicines to the right patients at the right time, the right diagnostics, right care.
The idea is that agents can actually help build and support regulatory submission packages as drugs are being developed. So, right there are some efficiencies to be had. You can use these agents to create and answer complex regulatory queries. At the same time, we’re seeing the use of these tools for powerful quality control. Again, that traceability, that auditability, flagging inconsistencies across thousands of different pages to ensure everything is correct.
But if there is something incorrect, the idea of agents is not just to flag when something is incorrect, but then being able to flag it and then being able to bring it back to where there’s an opportunity for us to create an action of, okay, well, this needs to then go to this department to get the right information and create that connection for that correction. The idea is that AI and AI agents can actually tackle some of the most common sources of delay before a submission actually goes out the door.
GEN Edge: What are some examples of life sciences customers successfully using Google Cloud and Google AI?
Maniar: From the R&D side, we’ve seen Bayer using our Google Gemini [family of AI models] that have been analyzing massive amounts of research and imaging data to accelerate their R&D work, and their time to diagnosis, particularly in their radiology unit. Hackensack Meridian Health is looking at streamlining clinical operations, where they’re automating a lot of their high-volume administrative tasks, with the aim of reducing clinician burnout. You can imagine that there are numerous use cases where we can apply AI to a lot of these administrative tasks.
And then a third example that is different than the first two is how we’re working with Eversana, which is leveraging the agentic AI ecosystem to help a lot of our ecosystem partners to reinvent the way that they’re helping pharmaceutical companies do their commercialization and their patient service models, so that these pharma companies can drive their market efficiency and their patient access programs.
The point is, 2026 is not just the use of agents to find efficiencies in the R&D space, but really across the value chain for the healthcare and life sciences ecosystem, that entire ecosystem. The focus and our mission have really been, how do we provide this secure agentic infrastructure that actually turns the oceans of data into streams of discovery, so that we can allow every researcher, every employee to see patterns that are faster than what has been previously humanly possible?
GEN Edge: You noted that users cannot simply rely on AI saying so, but trust needs to be established and verified. How will life sciences users address the issue of trust and the need to verify what is coming out from Google Cloud?
Maniar: I think this is actually not just about life sciences. I think this is just generally how Google Cloud builds and supplies its tools overall.
Our Google Cloud and AI services are built to be HIPAA-compliant. Our platform undergoes rigorous independent third-party audits. We have airtight security, and we have a large security team.
What you’re also asking about is the transparency and the explainability aspect of it. The black box of AI is a non-starter from our perspective. Every person, whether you are a tech person by background or not, is now today a tech person. It is our responsibility to make sure that it is not a black box, but rather a glass box.
We have a transparent audit trail so that it can be traced back with verifiable scientific data. But then also, we have provided tools like explainable AI and model cards so that we can help the users, or in this case the scientists, interpret the model’s behavior so that we can then give the scientists the insights that they need so that they can trust the results and be able to explain them to the regulators or to those who are questioning.
GEN Edge: What has changed to allow this evolution from reactive to proactive?
Maniar: Overall, what has changed is that these capabilities with agentic AI are now allowing for more data types and the multi-modality to talk to each other. This interoperability is now allowing more data and more insights to be drawn.
During a McKinsey-hosted discussion at the J.P. Morgan 44th Annual Healthcare Conference, it was noted that the world created more than 11,000 exabytes of healthcare and life sciences data. That includes electronic health records, whole genome sequencing, wellness information, everything. Of the 11,000 exabytes of data that were created in 2025, the world only created insights or used 3% of that; 97% of that information is just floating around and completely unused.
So, there’s an opportunity. It’s not about creating more information. It’s also about using these AI tools to create previously unknown insights with the data that these organizations all have. This is what we mean by being proactive, because the organizations have this information. They might even have previously unknown insights that are sitting within their proverbial lab notebooks. And now with these agentic tools, they can elicit that. And that allows them to be much more proactive.
GEN Edge: How much of a grasp on AI in the cloud do biopharmas have? And what gaps in their knowledge do they look to fill when they reach out to Google Cloud?
Maniar: Life science organizations are on varying AI journeys. There are some that are far more advanced that have been piloting for a few years. Where I would say most are right now is somewhere in the middle. I think that 2025 was the year of experimenting, the year of piloting.
2026 is going to be the year when we see organizations that are scaling. And that was certainly evidenced by a lot of the conversations that we had at J.P. Morgan, where many organizations are really coming and trying to understand, how do we scale in different parts of our organization, whether that’s in R&D, whether that’s in their manufacturing supply chain and logistics, or in the commercial parts of their organizations. It’s not about experimenting anymore, but now organizations are trying to understand.
And it’s not just about the low-hanging fruit, but how do we now go to scale? How do we find that one use case that we tried in a lower-stakes setting in our company? And now, how do we scale that for multiple parts of our business? I believe you’re going to see this year a lot more life science is going toward scale. And scale can mean different things depending on how large your organization is.
GEN Edge: We hear companies talk about using AI to run clinical trials, streamline those trials, and make them faster or less costly. Bristol-Myers Squibb last year discussed using AI to share information about its products with employees. Where in their ecosystems are life sciences customers using AI the most, and where are they going to expand that use in 2026?
Maniar: Right now, we have seen that the fastest and the most measurable ROI has been in internal workflows, where you find the most inefficiencies. For example, at the top of the funnel in R&D, which I alluded to at the beginning of our conversation, I think there’s the biggest opportunity there.
When I’m talking about lower risk opportunities, these are maybe administrative tasks that are just internal in your own organization. Where we’re seeing these opportunities that are starting to grow from there is still in R&D. We’re also seeing quite a heavy movement into manufacturing and supply chain tasks. I believe that is also due to the macroeconomic headwinds that are also impacting the way organizations are looking at their long-term goals.
GEN Edge: We’ve heard for years that biopharmas would be most attracted to AI because of its potential to cut the long timeline of drug development and thus the cost of developing new treatments, thus bringing new medicines to patients faster. Yet one of the biggest fears is its potential for large-scale expense cutting, including jobs. How much do customers look for that, and how much of the broader potential for savings have companies tapped into?
Maniar: That’s a question that I think is not just for this industry. It’s a question that is larger than just this industry. I think that we care a lot about how technology affects different roles and different jobs. And I think there’s no doubt that jobs are going to shift, as they always have with any evolution and revolution of technology. This is not the first time this has happened in history.
But there will also be many jobs that are going to be complemented by these advances in technology, where entirely new jobs are going to be created, because the example that we often give is, who could have imagined flight attendants even existing before air travel was created? This is a challenge, I think, that’s larger than any one company or one government can solve or one industry can solve.
GEN Edge: How can life science organizations work with government and other partners to address this challenge?
Maniar: I think we have to work together on programs that can help people make a living and find meaning in work. But when it comes particularly to the way we work with the life sciences industry, this is a question I often receive, as leaders are trying to understand how they can help their employees understand that we need to advance our employees to understand AI.
And one of the things I also say often is that the people who are willing to learn about AI are going to be the ones who are going to move ahead. The companies that are going to learn about AI are the ones that are going to be moving ahead, because it’s no longer where you can just have a stiff arm and try to bury your head. This is affecting the entire ecosystem.
This is not about experimenting with AI at this point. This is how the entire life science industry is being affected by how these technologies are going to play a role in the future of how this entire industry works. Technology is not a peripheral, adjacent part of this industry. Technology is now a foundational part of how the life science industry works. It’s just how you adapt to it.
GEN Edge: How much is Google Cloud intended to be integrated with or supplant the legacy tech that a biopharma or a healthcare provider has?
Maniar: Our intention is that we want our technology to be an enabler for these industries, not for us to supplant their technology. That is not the intention. It is meant to be the underlying infrastructure and platform on which these organizations run.
GEN Edge: Has Google quantified the potential savings of time and money from AI?
While AI agents represent the new frontier, the foundational value of gen AI continues to deliver compounding returns: 73% of healthcare and life sciences organizations are already seeing ROI on at least one AI use case, such as productivity, customer experience, and business growth. Among respondents, 83% saw revenue gains of 6% or more, 72% saw improved productivity, and 45% reported moving from an initial idea to a production use case in just 3–6 months.
GEN Edge: A year ago, there was talk about Google’s AI effort falling behind ChatGPT. But Google launched several new products in the fall, such as Gemini 3 and Nano Banana Pro, plus there have been major new investments that Google has talked about in AI infrastructure. How much did all that enable, frankly, a different narrative, including from several prominent news outlets?
Maniar: All of this is impacting the way that we are perceived in the market, I would say, but this is what we have always been working on. This is something that we’ve always been focused on internally, and this is now how people are seeing this externally as well. I think that I would probably leave it at that.
We’ve been focused on what we’re doing, what’s right for the customer. There’s always a little bit of movement in the market: Somebody’s moving in this direction or that direction. Our focus is, really, how do we support the industry? And how do we enable the industry by continuing to use the new innovations that are coming within our organization to support and find the best and the brightest, and the newest capabilities to support our customers and our partners?
The post Google Cloud Survey: Life Sciences Leaders Find ROI in Agentic AI appeared first on GEN – Genetic Engineering and Biotechnology News.
