In a video interview with Pharma Commerce, Kevin Dondarski, Deloitte’s life sciences R&D strategy leader, describes how his company’s 15th annual Deloitte report on the return on investment (ROI) in pharmaceutical R&D reveals a continued upward trend, with a projected ROI of 5.9% for 2024. This marks a notable reversal from the consistent decline seen throughout much of the 2010s. During that period, ROI dropped largely due to a more challenging commercial environment, which reduced the projected value of drug pipelines. This was influenced by both business dynamics and a shift toward developing treatments for more specific, nuanced patient populations, often with smaller market sizes.
However, the past two years have shown a positive turnaround, with year-over-year increases in projected ROI. Two key drivers are behind this recent momentum. First, the overall projected value of late-stage drug pipelines has grown, especially when adjusted for risk, signaling stronger confidence in the commercial viability and clinical success of these therapies.
Second, a small number of exceptionally high-value drug programs have had an outsized impact. Among these, treatments in the GLP-1/obesity space stand out, significantly boosting the total value of the late-stage pipeline. These high-potential programs are reshaping expectations and driving much of the improvement in projected R&D returns.
The renewed optimism in pharma R&D ROI is largely attributed to an increase in risk-adjusted value across the pipeline and the emergence of breakthrough therapies in lucrative areas such as obesity treatment, highlighting a shift toward more valuable and promising innovation.
Dondarski also comments on the strategies companies can adopt to manage or mitigate escalating drug development costs; current best practices for balancing long-term pipeline sustainability with short-term financial returns; how GLP-1’s success influence future investment strategies in high unmet-need areas; how AI and automation realistically reduce clinical development timelines; and much more.
A transcript of his conversation with PC can be found below.
PC: How can AI and automation realistically reduce clinical development timelines, and what barriers are currently limiting their widespread adoption in R&D?
Dondarski: If I think about what the barriers or limitations are, I think there’s two that come to mind. The first is the decision of, do we want to be first, or do we want to [focus on] the velocity that companies are taking to adopt it? There’re some companies who’ve decided to immediately jump in the pool and go all in. Then, there’s other organizations that are kind of waiting to see how the technology matures before they make that hurrah.
The second piece is the fragmentation of decision autonomy, and how those technologies can benefit different parts of the organization. There’s so much utility in a lot of the things that we see in deployment—it crosses functional boundaries. Even if there’s the recognition to move forward with something, you need alignment from multiple functions within R& D. Within R&D IT, there’s usually security or cyber concerns, so it’s really just an alignment exercise.
I feel like that’s a major barrier that, on one hand, seems simplistic compared to what the R&D part of the industry is actually doing, but it can’t be understated. I think that’s been a barrier for different organizations.
I think we’re at an interesting inflection point. To your first question, when GenAI really became a thing or became more of a part of the public vernacular, there was a lot of emphasis on POCs and MVPs and individual use cases, etc. I think there’s a lot of success associated with that. I think what we see now is a broader migration to stop thinking about, let’s say, individual point-in-time or point-in-the-value-chain applications, and more of how can this enable a more transformative way for us to think about the end-to-end value chain.
The easiest examples that come to mind are within content generation or document authoring, where the same capabilities have utility across all aspects of the lifecycle, and I think we’re really starting to see a push in that direction. I think there’s still some uncertainty with respect to the more innovative or research side of the equation, where I think there’s an appreciation that leveraging data and AI can accelerate different processes and ways of driving innovation, but more is to come in that world, or it’s still too early to tell, in terms of how disruptive it can be in research.