Sixteen Years of Watching the Needle Move: A Long View of Pharmaceutical R&D Productivity
When we published the first iteration of what would become Deloitte's flagship pharmaceutical R&D productivity series back in 2010, nobody on the team imagined we would still be writing it sixteen years later. The topic remains consistently compelling, and so consistently fascinating for the industry as well as leaders, analysts, and commentators. As I handover leadership of this work, I wanted to reflect on my own views of what the data has told us, what it means, and what might concern us as we look ahead.
This is not a report, but it is a look back across sixteen editions. It is my attempt to make sense of a dataset that has grown into one of the most-cited barometers of biopharma's capacity to create value from scientific and clinical innovation.
The Arc of the Data: A Story in Three Acts
Act One: The Slow Fall (2010-2019)
When we began our analysis in 2010, the average internal rate of return for the world's largest biopharma companies stood at 10.1%. That number looks almost nostalgic today. It reflected an industry earning the rewards of the blockbuster era - when a single small molecule could generate billions in annual revenue and fund the next generation of innovation for a company. It was not, as we now understand, a sustainable model.
Over the following nine years, the IRR fell in almost every single year. By 2019, it had reached 1.5% - a number so low that it prompted our tenth-anniversary report to ask, bluntly, whether the current R&D model was viable at all. The causes were structural and well-documented across our reports: escalating clinical development costs (from $1.19bn per asset in 2010 to nearly $2bn by 2019), shrinking peak sales forecasts as markets fragmented into smaller and more specialised patient populations, and lengthening cycle times that pushed the economics of drug development ever further into the red. This “scissors dynamic” is the central story of the decade.
What made this decline particularly striking was its relentlessness. There were brief intermissions - most notably in 2014, when a cluster of high-value late-stage assets briefly lifted the average IRR back to 7.2% - but these proved temporary. The structural forces outweighed the episodic ones.
Act Two: The Pandemic bump (2020-2022)
COVID-19 arrived, and for a brief period, the story changed completely. The speed at which the industry mobilised to develop vaccines and treatments was as extraordinary as it was inspirational and it showed up dramatically in our numbers. In 2021, the combined cohort's IRR had surged to 6.8%, the highest level since 2014. Average cycle times fell. Forecast peak sales jumped back above $500 million per asset. It looked, for a moment, like a turning point in a world looking to the industry for solutions.
It was not. Or at least, not yet. In 2022, as the most valuable COVID-19 assets were approved and moved into the commercial portfolio - and therefore out of our late-stage pipeline model - the IRR collapsed to 1.2%, the lowest figure in the history of the series. It was a statistical artefact as much as a genuine productivity collapse, but it illustrated something important: the industry had not yet internalised the lessons of COVID-19 development speed. The pandemic had shown what was possible when regulatory flexibility, pre-agreed protocols, and unprecedented investment alignment came together. But as the emergency receded, so did the urgency and the willingness to do extraordinary things together every day as an industry and regulator.
The 2022 result was a sobering reminder that external disruption is not a substitute for systemic change.
Act Three: A Recovery? (2023-2025)
The last two years have brought genuine, if fragile, cause for optimism. In 2023, the IRR recovered to 4.3%, and in 2025 it reached 7.0% - driven by a wave of high-value late-stage assets in obesity, diabetes, and Alzheimer's disease. In particular, GLP-1 therapies have reshaped the industry's financial landscape, with two companies in our 2025 cohort projecting average peak sales exceeding $2 billion per asset. The number of projected blockbusters (assets forecast to achieve peak sales above $1 billion) rose to 29 new entrants in 2024 alone.
These are real improvements. But it is worth saying clearly: they are concentration-driven with significant risks. If GLP-1 therapies are excluded from the 2025 analysis, the IRR falls from 7.0% to 2.9%. The recovery is real, but it is not yet systemic or the result of a new model. And with R&D costs continuing to climb - reaching $2.67 billion per asset in 2025, an increase across 12 of the top 20 companies - the underlying economics of drug development remain challenging. When combined with high competition in a specific therapy area, costs are likely to rise in clinical but with diminishing commercial returns. What the Numbers Actually Tell Us
The cost problem is structural, not cyclical
Across sixteen years, the average cost to bring a late-stage asset from discovery to launch has nearly doubled - from $1.19 billion in 2010 to $2.67 billion in 2025. This is not primarily inflation. It reflects the genuine increase in protocol complexity, complex study endpoints, companion diagnostics, the shift toward specialist and rare disease indications requiring smaller and harder-to-recruit trial populations, and the competitive intensity in therapeutic areas like oncology where multiple programmes are chasing the same patient cohorts. Until the industry finds a durable way to compress Phase III cycle times - which increased a further 12% in 2024 - cost management will remain a structural headwind.
External innovation has become the norm, not the exception
One of the most significant shifts documented across the series has been the dramatic rise of externally sourced innovation. In 2021, 66% of the combined cohort's late-stage pipeline volume was projected to come from externally sourced assets, up from roughly 50% a decade earlier. This was an outlier representing the acquisition of COVID assets by the large Pharma in the report. In 2025, nearly half of all forecast revenues were expected to be generated through externally sourced innovation. The large-cap biopharma companies have, in effect, become sophisticated integrators and developers of innovations originating elsewhere. This changes the economics, the risk profile, and the strategic calculus of R&D investment in fundamental ways. It also creates a challenging question for every R&D leader - what is the value of internal Discovery and Research investment? This is a financial question as much as a scientific one.
Novel mechanisms of action are the engine of value creation
Our most recent analysis found that novel mechanisms of action - representing just 28% of the late-stage pipeline on average - account for 53% of projected revenue. The premium attached to genuine therapeutic innovation is not just scientific, it is financial. Companies that have been willing to pursue uncharted territory, whether in GLP-1 biology or ADC engineering in oncology are generating disproportionate returns. The lesson is consistent across every era of this data: differentiation at the science level is the most durable source of long-term productivity.
The pendulum of optimism about technology moves faster than the technology itself
Every edition of this report has identified a technology - genomics, personalised medicine, digital biomarkers, platform technologies, AI - as the potential solution to the productivity challenge. And in each case, the technology has taken longer to translate into measurable R&D efficiency gains than the optimists anticipated. This is not a reason for cynicism about AI; the capabilities now available are genuinely transformative in ways that earlier technologies were not. But it is a reason for discipline. The most productive companies in our cohort are not those who have made the most announcements about AI adoption. They are those who have embedded data-driven decision-making into their development processes, shortened their internal cycle times, and invested in the early-stage pipeline with long-term conviction.
What I Would Say to the Industry
Sixteen years of authoring this report has, inevitably, given me perspectives. I offer them in the spirit of constructive challenge, not criticism from an outsider.
First: the patent cliff of 2025-2030 is real, and M&A alone will not solve it. The pipeline replenishment strategies that have worked historically - large-scale acquisitions of late-stage assets - are becoming less available and less differentiated as competition for high-quality external innovation intensifies. The companies that will thrive are those investing now in differentiated early-stage biology, in platforms with platform economics, and in therapeutic areas where they have genuine scientific edge.
Our most consistent finding across sixteen years of this analysis is simple: the companies that invest boldly in genuine innovation, tolerate the near-term cost of doing so, and build the internal capability to translate scientific insight into clinical proof, generate the highest long-term returns. The IRR follows the science not the market.
Second: AI's contribution to R&D productivity will ultimately be measured in cycle time and cost, not in announcements and increasing spend. The critical question for the next five years is whether the industry can use AI-enabled insights to reduce Phase III attrition - the single biggest driver of the cost escalation we have documented across this series. If it can, the economics of drug development change materially. If it cannot, the headline IRR figures will remain structurally challenged regardless of which blockbuster therapeutic class happens to be entering late-stage development in a given year.
Third: the productivity of the industry is not just a financial question. It determines how many patients access how many medicines how quickly. The 1.2% IRR of 2022 did not just threaten shareholder returns; it signalled a system under stress. The recovery of 2023 to 2025 is welcome, but it has been led largely by a single therapeutic area - obesity - that represents perhaps the largest unmet medical need in the developed world. Sustaining that recovery across a broader portfolio including neurodegenerative disease will require sustained courage in R&D investment, not just optimisation of the assets already in development.
A Personal Note
I have been fortunate to work on this report with a cohort of Deloitte colleagues across a period of extraordinary change in the pharmaceutical industry: the genomics revolution, the rise of biologics, the emergence of cell and gene therapy, the COVID-19 mobilisation, and now the early stages of what may become a genuine AI-driven transformation of drug discovery. The data has been humbling in its complexity and instructive in its consistency.
The story of pharmaceutical R&D productivity is not, ultimately, a story about cost curves or IRR models. It is a story about the difficulty of turning scientific uncertainty into reliable commercial value and about the remarkable persistence of an industry that keeps trying. The 2025 results suggest that trying is beginning, again, to pay off. I leave this work with genuine optimism about what comes next, and with deep respect for the scientists, clinicians, and strategists working every day to move the needle.
