Resource Header Hubspot (18)
PRESS RELEASE

Enthought Survey Finds R&D Leaders Expect Agentic AI to Change How Research Is Done, but Most Organizations Are Early in Putting It Into Practice


The Ambition Gap report shows 65% of R&D organizations already use agentic AI and 55% plan to raise AI budgets by 10% or more, while only 12% have a defined roadmap

AUSTIN, Texas — Nov. 19, 2026 — Nearly two-thirds of R&D leaders expect agentic AI to change how research is done, not only speed it up, but most of their organizations are still early in putting it into practice, according to a new survey from Enthought, which specializes in science-first AI for enterprise R&D. The report, The Ambition Gap: How R&D Leaders Are Approaching Agentic AI in 2026, draws on responses from R&D executives, directors, managers and scientists in advanced materials, chemicals, semiconductors and other science-driven industries.

Agentic AI refers to AI systems that can plan multi-step work, use software tools and act on the results. These systems are moving from demonstrations into everyday use, and in industries where being first to a new material, formulation or chip design often decides who wins the market, R&D leaders are weighing how quickly to adopt them. The survey shows where R&D organizations stand today and what sets apart the organizations moving fastest.

Where the industry stands
Agentic AI is already part of R&D work, and leaders expect its role to grow well beyond productivity.

  • Adoption is broad but mostly individual. 65% of R&D organizations already use agentic AI in some form, most often with individuals running single agents for their own work. Only 8% have connected agents or agentic systems working across a team, lab or organization.

  • Leaders see agents as research partners. 37% of R&D leaders see agents as research partners working alongside scientists, and 26% call agentic AI a transformative force for R&D. Only 22% see it mainly as a productivity tool.

  • Leaders are ready to let agents act. 49% are comfortable with agents running multi-step tasks on their own within guardrails, and only 13% want AI that only makes suggestions.

  • Few organizations have a roadmap. Only 12% of R&D organizations have a defined roadmap for agentic AI, and 20% have no defined strategy at all.

  • Investment is rising. 55% plan to increase AI budgets by 10% or more in the next 12 months, including 65% of organizations still forming their approach to agentic AI.

What separates the organizations moving fastest
The survey groups respondents by their budget plans. Accelerators are increasing AI budgets by 10% or more, and Late Adopters have no dedicated AI budget yet. The differences between them point to where R&D leaders can focus next.

  • Organizations investing now are pulling ahead. 75% of Accelerators use agentic AI today, compared with 32% of Late Adopters. 53% of Late Adopters also have no defined strategy for agentic AI.
  • Organizations with AI capability inside R&D use agents more widely. 47% of organizations have dedicated R&D AI teams or scientists leading AI work. These organizations are nearly three times as likely to use agents across teams rather than one person at a time (36%, compared with 13% of other organizations).
  • The barriers are mostly managerial, not scientific. Lack of in-house technical expertise is the most common barrier (35%), followed by scientific validity and budget (33% each). Two-thirds of the barriers respondents named are managerial or operational, such as budget, the business case and leadership priority.
  • Concerns about validity do not stop adoption. 68% of those who name scientific validity as a barrier are still comfortable letting agents act with human review.
  • Measuring value is a practical first step. 32% of organizations have no framework yet for measuring AI's return. Half of them also have no dedicated AI budget, compared with 6% of organizations that have a framework.
  • Agents are moving into experimental work. Literature review is the most common use (75% of current users), and 47% of current users also apply agents to experimental design, simulation and digital twins, or property, formulation and molecular prediction.

“We surveyed R&D leaders across our network to see where their organizations stand with agentic AI in 2026, and the clearest pattern was a gap between ambition and practice,” said Wendy Gonzales, Vice President of Marketing and Strategy at Enthought, who led the survey. “Most leaders recognize that agentic AI is now transforming how enterprise research is done, but their organizations are still early in adoption. The organizations moving fastest tend to have someone who owns AI strategy, a way to measure value and real budget behind the work. We expect the gap between them and everyone else to grow rapidly in 2027.”

“In 25 years of working with scientists and engineers, I haven’t seen a technology as consequential as agentic AI. It’s unlike the technologies that came before it, because it changes how the work itself gets done,” said Eric Jones, CEO of Enthought. “We’re in a transition period right now, and the R&D organizations that start building agentic systems in the next year or two will set the pace in their industries for years.”

The full report covers 12 key takeaways with 15 supporting charts, including who owns AI strategy and budget in R&D, how organizations measure the return on agentic AI, where agents are already in use, what stands in the way of wider adoption, and the traits shared by the organizations investing most in agentic AI. It is available at https://www.enthought.com/agentic-ai-in-rd-2026-survey-of-rd-leaders.

About the survey
The Agentic AI Leadership Pulse 2026 is a survey of R&D leaders in Enthought’s global network, fielded from July to October 2026. It is meant as a snapshot of how science-driven R&D organizations are approaching agentic AI. Results are based on 101 qualified respondents, mainly in chemicals, advanced materials, semiconductors, aerospace, automotive and energy, and include executives, directors, managers and scientists.

Note to editors: Charts from the report are available for publication with attribution to Enthought, Inc.

 

About Enthought
Enthought specializes in science-first AI for enterprise R&D. The company engineers innovative solutions that scale discovery and product development across advanced materials, specialty chemicals and semiconductors. As pioneers of the SciPy ecosystem, Enthought has been at the forefront of scientific computing since 2001. Today, it helps companies leverage the unprecedented opportunities of AI. Learn more at https://www.enthought.com.

Media Contact
Enthought
media@enthought.com
512-536-1057



Enthought
helps companies all over the world solve complex data challenges unique to enterprise scientific R&D by leveraging advanced technologies that accelerate innovation and transform the business. We bring an unparalleled blend of expertise and experience in advanced AI/ML techniques, scientific research and data, and business strategy. Enthought is headquartered in Austin, Texas, USA with additional offices in Tokyo, Japan; Cambridge, United Kingdom; and Zürich, Switzerland.

Learn more about us here.