Signal Pulse Report: We asked 441 Endpoints News readers how AI is actually showing up in their jobs. What stuck out first: 48% of respondents now describe themselves as heavy AI users in the workplace. The split between those who use AI regularly and those who don't turns out to matter a lot for how people see the future.
Artificial intelligence has moved from a side project to a fixture on biopharma agendas in just a few years. In 2025, companies signed billion-dollar AI partnerships, GPU suppliers found drug discovery to be a new growth engine, and it seemed every second conference panel promised an imminent shake-up in R&D.
That’s why Endpoints Signal, the new intelligence platform from Endpoints News, decided to sanity-check the hype. We surveyed our readers to see how AI is actually showing up in their work, and where they think it’s all heading. We asked where AI delivers the most value, what’s holding it back, and how opinions differ among scientists, executives and the investors who bankroll them.
The upshot: AI’s impact today is patchy and problematic — but expectations for the next five years are about as high as a three-espresso morning in Silicon Valley.
Artificial intelligence today
Ask biopharma people what AI is doing for them right now and you might get a shrug. More than half of respondents said the current impact was "modest" or "inconsequential." Only 15% said AI is doing transformative work in R&D today, even after years of excitement and money.
In their own words:
“AI is driving up pre-money valuations for no good reason.”
“It takes oxygen from other companies more likely to deliver therapeutics”.
“ brought an AI assistant online, but it has no access to our internal systems — this is a good example of where AI fails. If it cannot be trusted on internal systems then ultimately it does not provide the functionality needed to be useful.”
“A lot of noisy distraction”
Beneath the lukewarm assessment, however, there’s a big split. Half of respondents say they’re using AI tools heavily in their work, including predictive models for hit-to-lead, to LLMs that crank out regulatory documents. The heavy users are reporting real workflow change, and this experience significantly shapes their perceptions of AI’s current impact.
Five-year transformation
But those mixed opinions converge when we look forward a few years. Three quarters of respondents expect AI to transform drug R&D by 2030 — and 12% think we’ll get there even sooner, in just three years.
That optimism is backed by significant action in the industry this year, including Eli Lilly building an AI supercomputer with Nvidia using more than a 1,000 B300 GPUs — high-end, interconnected gear that would be brag-worthy in any industry. Then there’s the OpenFold3 initiative to pool proprietary protein data from labs around the world, including Johnson & Johnson, Bristol Myers Squibb, Takeda, AbbVie and others. Not to mention billions of dollars flowing into a half-dozen AI funding and licensing deals.
Experience is the great accelerant. Among those who already use AI extensively in their work, 90% expect drug discovery and development to be practically unrecognizable by 2030.
Artificial intelligence is nothing new in biopharma R&D; it’s been around for years in the form of machine learning and predictive models with names like QSAR (quantitative structure-activity relationship) — long before the current wave of large language models. That experience gives biopharma a more nuanced perspective.
And job role also shapes conviction in AI — though not nearly as much as firsthand experience with the tools. Executives and scientists are the most bullish on AI’s promise, while the finance crowd remains the least convinced.
Chips aren't the bottleneck
Given all the headlines about GPUs, “compute cost” might be expected to dominate any list of AI obstacles. But in Signal’s Pulse Poll, it barely registers. When we asked what’s actually holding back AI in biopharma, two answers towered over the rest: translating models to real-world biology (43%) and getting access to high-quality biological data (31%). Workflow integration came in third (13%). Leadership buy-in and compute cost were afterthoughts.
The answers reflect a classic “last mile” problem, except the last mile is a swamp of messy, siloed data and decades-old processes. Respondents described AI that can spot early-stage lung cancer on a chest X-ray, predict protein structures, and optimize trial protocols — but only if the right data are cleaned, labeled, and wired into the lab and clinic.
Who stands to win the AI race?
The AI race in biopharma is increasingly a contest between incumbency and insurgency. On one side sit the giants — companies with vast clinical datasets, sprawling trial networks and the capital to snatch up AI platforms as fast as they emerge. Those advantages matter. If AI ends up rewarding whoever can train the best models on the most biologically rich data, big pharma starts several laps ahead.
But the surprise in this AI Pulse is how many respondents see room for a very different kind of winner — tech companies newly marching into healthcare. In dozens of other industries, the pattern has been the same: Entrenched players underestimate outsiders who bring different incentives, better tooling, and an allergy to traditional operations.
To many respondents, the threat isn’t that big tech will build the next blockbuster molecule; it’s that they’ll capture the software, infrastructure and model layers that everyone else eventually depends on.
Heavy users live in a different future.
The cleanest dividing line in the Pulse Poll isn’t job title or company size — it’s AI usage intensity. Heavy users, who say they use AI "extensively" or "frequently" are more likely to say AI is already having a significant impact today, are more bullish on a 2030 transformation, and are noticeably less worried about integration into workflows. Light users, who use it "occasionally," "rarely" or "never" see more hype, more friction, and more risk that AI turns into yet another forgotten slide from a corporate strategy powerpoint.
That gap echoes what David Reese, the chief technology officer at Amgen, told us onstage at the Endpoints AI Day in October. As Reese put it, there won’t ever be a “big bang” in AI — just a time when you look back and realize the work is being done in a qualitatively different way. For the heavy AI users in this poll, that moment may already be here.
“There’s this staggering resource sitting in front of us," Reese said, describing the amount of untouched biopharma data that could be tapped by AI. "It’s like the mid-19th century: Oil hasn’t quite been discovered yet, but there’s this immense resource. That is what’s sitting right in front of us now.”
How AI is useful today, in their own words:
“Helps to slightly accelerate drug discovery by optimizing hit-to-lead generation. Helps a lot to generate documents for clinical development”
“Finding a biomarker”
“Drug molecule optimization process”
“Biologics drug design”
“AI supported identification of early stage lung cancer from Chest X-rays in real life clinical practice”
“Finding optimal protocols for pre clinical and clinical trials”
“Drug development, specifically drug identification and phasing out animal testing — replaced by in silico simulations and hybrid AI”
Methodology
Signal’s AI Pulse Poll surveyed 441 verified Endpoints News subscribers to understand how artificial intelligence is showing up in biopharma workflows and how practitioners expect it to shape drug R&D over the next five years. The poll was fielded online over a one-week window in October 2025. Responses were drawn from professionals across biopharma, life sciences, investment, healthcare services and adjacent sectors.
Respondents self-identified their AI usage intensity (“extensive,” “frequent,” “occasional,” “rare,” or “never”). Job roles were identified via Endpoints login (executives, investors, scientists, staff, consultants and others). These two variables serve as the primary segmentation lenses in the report. Aggregate results are unweighted. Open-ended answers were lightly edited for clarity.
↧