Why R&D Waste Persists Despite Widespread AI Adoption
Wiley Science and Engineering Content Hub Patsnap
A new survey of 200+ R&D leaders finds companies still burn huge chunks of budget on projects that flop before hitting market. AI adoption is up everywhere, but most teams still use it for grunt work, not for deciding what's worth building.
Based on reporting by Wiley Science and Engineering Content Hub, Patsnap — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Ask any senior R&D professional where the money disappears and you'll get a familiar answer: late-stage failure. A new benchmark survey of more than 200 R&D leaders across North America, the UK and Europe found that over a third of organizations sink somewhere between a quarter and 40 percent of their R&D budget into projects that never actually reach the market. That's not a rounding error. That's a structural leak.
The pain gets sharper the later a project dies. Nearly half of the respondents said killing a project at a late stage costs their organization more than a million dollars each time. Think about what that means in practice: teams keep funding, staffing, and testing something for months or years before anyone admits it isn't going to work, and by then the bill has already stacked up.
Here's the part that should sting a little more than it does. AI adoption inside these R&D organizations has surged, according to the report. But most of that AI firepower is aimed at execution — running experiments, crunching data, speeding up tasks that were already going to happen. It's not being pointed at the harder question of which projects deserve the investment in the first place. The tools got smarter. The decisions about where to point them didn't.
The report, based on responses from over 200 senior leaders spanning nine industries, argues the fix isn't more AI for AI's sake — it's earlier intelligence. Competitive data, market signals, and patent information matter most during ideation and feasibility, before an organization has committed serious resources. Wait until a project is deep into late-stage development, and even the best intelligence just tells you what you already suspected: this one isn't going to make it.
The report — produced with IEEE Spectrum and Wiley, and sponsored by Patsnap — frames this as a gap between AI adoption and AI application. Plenty of companies can now say they use AI. Fewer can say they've pointed it at the moment where it would actually save them money: the decision of whether to start down a path at all.
My take — AI-written commentary, not fact-checked reporting
None of this should be surprising, and yet somehow it always is. Organizations keep buying AI tools and pointing them at the parts of R&D that were never the expensive problem — execution — while the actual money pit, deciding which projects deserve funding in the first place, stays run on gut feel and momentum. A million-dollar late-stage cancellation isn't a technology failure; it's a sequencing failure, and no amount of AI adoption fixes a process that asks the right question too late.
Read more about this at: Wiley Science and Engineering Content Hub