The Future of Industrial AI with Honeywell Technologies Chairman and CEO Vimal Kapur

Summary

Honeywell Technologies Chairman and CEO Vimal Kapur and John Waldron frame Honeywell’s transformation as a response to complexity, with the company separating into three public entities to sharpen focus and unlock growth. The discussion centers on industrial automation as mission-critical infrastructure serving always-on environments like hospitals, data centers, refineries, semiconductor fabs, and LNG plants, where Honeywell sees a path from deterministic automation toward autonomy by using data already generated across tens of thousands of sites and millions of assets. Kapur defines physical AI as industrial AI for continuous-process systems, emphasizing human-in-the-loop decision-making rather than full automation. He also highlights increased R&D intensity, the Forge software platform, labor shortages and replacement rates, and quantum computing as a complementary opportunity, especially for drug discovery and cryptography.

Notes

• Honeywell is being simplified into three public companies to reduce complexity and sharpen strategic focus.
• The breakup is framed as a response to complexity becoming a barrier to the company’s next phase of growth.
• Honeywell’s transformation included two major external shifts: ChatGPT and the aerospace cycle.
• The company’s automation business is positioned as infrastructure that cannot fail and must run continuously.
• Honeywell says roughly 60% of automation revenue comes from capital spend and 40% from OPEX, aftermarket services, and software.
• That mix is presented as structurally secular and capable of mid-single-digit growth.
• Honeywell sees industrial automation as a three-part growth engine: capital cycles, OPEX/service cycles, and AI-driven compounding.
• The company argues the biggest untapped value is in operational data it already generates but does not fully use.
• The shift from automation to autonomy is expected over the next five to ten years.
• Honeywell is increasing R&D because new products are the main growth driver.
• 2025 R&D spend is 4.8% of total revenue.
• R&D intensity was raised by roughly 40 to 50 basis points.
• Honeywell’s Forge platform collects operational data and layers AI algorithms on top of it.
• Forge became breakeven last year and is now economically self-sustaining.
• The company says it now makes money from Forge rather than subsidizing it.
• Physical AI is defined as industrial AI for continuous-process systems, not humanoid robotics or discrete widget manufacturing.
• Honeywell’s target environment is always-on systems running 24/7.
• The company distinguishes its industrial AI thesis from mainstream robotics narratives.
• The core value proposition is augmentation, not replacing humans in every workflow.
• AI is framed as an agentic copilot that presents options while humans retain final authority.
• One example suggests AI could save four hours a day of repetitive data hunting.
• Industrial AI applications are expected to remain human-in-the-loop rather than fully autonomous.
• Kapur says the industrial AI stack will mostly coexist with people rather than displace them entirely.
• Honeywell’s industrial opportunity depends on proprietary data that is not publicly available for model training.
• Quantum computing is viewed as complementary to classical computing, not a replacement.
• The near-term quantum focus is on use cases and economics.
• Drug discovery is identified as the biggest unsolved quantum opportunity.
• Cryptography is also highlighted as a potential quantum use case.
• Kapur says the quantum market is still in the phase of proving where value creation will come from.
• The company argues leadership should simplify metrics before simplifying operations.
• Corporate functions should narrow their role and let businesses own more decisions.
• Leadership discipline is presented as time management and ruthless prioritization.
• The conversation links industrial AI adoption to labor scarcity, retirements, and low replacement rates.
• Large industrial assets require skilled operators even as AI improves workflows.
• Examples of asset classes in scope include semiconductor fabs, refineries, and data centers.