Chemicals and materials firms are putting AI in the plant and the lab. The bets that move the business are still made the old way.
Predictive maintenance, process optimisation, a materials-discovery pilot: AI is treated as a plant and R&D-tooling project. Meanwhile the capital allocation across the cycle, the portfolio and formulation bets, and the build-versus-buy calls that actually move a materials business get a generic answer or no AI at all.
Request a Strategic Briefing →
Ask most chemicals and materials companies where AI lives and they will point at two places: the plant, where it runs predictive maintenance and process optimisation, and the lab, where it screens formulations and accelerates discovery. Both are real work, and both are worth doing. The trouble is that the decisions that actually move a materials business, where to allocate capital across the commodity cycle, which molecules and product lines to back, the build-versus-buy and footprint calls, still get made the old way or handed to a commodity AI that answers with confidence and no knowledge of your business. The pilots multiply in the plant and the lab. The scaled value stalls before it reaches the bet. That is High-Speed Waste in capital form. Gildoni installs the Havruta Methodology (formerly the Think Partner Methodology) into how leadership reasons with AI, at the altitude where the portfolio and capital judgement is actually made.
AI is everywhere in the plant and the lab, and barely anywhere in the portfolio call
The industry's instinct is to scope AI to the plant and the lab bench. They are the most visible, most measurable places to start, and the vendor market is built to sell there. So that is where the pilots land.
The honest difficulty is that a process or discovery pilot is easy to launch and hard to scale into something the whole business feels. Materials discovery in particular has been transformed: AI can now screen candidate compounds at a rate no laboratory could approach. The capability at the bench is no longer the bottleneck. The reasoning one altitude up, where the capital and portfolio bets are placed, is.
And those bets are heavy. A chemicals and materials business runs on long-lived assets, a brutal commodity cycle, regulatory exposure, and capital commitments that play out over years. The decision to back one molecule, expand one line, or sit out one trough is worth more than any efficiency a plant model will ever return, and it is exactly where AI is least present.
The numbers show the same story from three angles.
new crystal structures discovered by one AI system, the equivalent of nearly 800 years of accumulated knowledge, with about 380,000 predicted stable enough to be worth synthesising. Discovery at the bench is no longer the constraint.
higher equipment uptime, with 5 to 10% lower maintenance cost, from predictive maintenance on plant assets. Real value, and bounded: it optimises an asset you have already decided to run.
of companies are generating real value from AI at scale, even as adoption goes near-universal. The gap between the few and the rest is widening, and it is a leadership-judgement gap, not a tooling one.
The bench has the capability. The portfolio call has the gap.
GNoME shows the potential of using AI to discover and develop new materials at scale.
Why the value stops at the plant and the bench
If the capability is there, why does the value stall? Three reasons.
AI gets scoped as a plant and lab project, so it inherits a plant and lab remit. A predictive-maintenance model shaves downtime on one reactor. A discovery model ranks ten thousand candidate compounds. Neither was ever meant to weigh in on whether to fund the line that makes the winning compound, or to sit out the next trough in the cycle. The remit they were given is the ceiling on the value they can return.
The decisions that actually move a materials business live above the floor. Capital allocation across the commodity cycle, the portfolio bet on which products and molecules to back, build-versus-buy, the footprint and capacity call, the regulatory and decarbonisation trade-offs. These are bets under deep uncertainty, made by people, and they are where AI is least present and most needed.
And where AI does reach those calls, it tends to arrive as a vending machine. Put a portfolio question to a commodity model and it returns a fluent, confident answer that knows nothing about your feedstock positions, your contracts, your regulatory exposure or where you sit in the cycle. Pilots multiply because they are cheap to start. Scaled value does not follow, because nobody installed the reasoning that turns an AI answer into a defensible capital decision.
It is worth being honest about the older truth underneath this. Capital discipline and judgement through a commodity cycle are old skills. AI has not replaced them. It has stress-tested them, and found most materials firms deploying at bench speed in the lab while still deciding at committee speed about the business.
Screening the chemistry is the easy part
Here is what the bench-first instinct leaves out. Every vendor and every roadmap now says the same thing: deploy AI in the plant, screen compounds in the lab, scale the pilot. Almost none of them say how a leadership team actually reasons through the capital and portfolio decisions that sit above the bench, the ones where the real money moves.
That is the real gap, and it is not a tooling gap. A lab full of working discovery models can sit directly beneath a capacity bet still made on a spreadsheet and a feel for the cycle. At the altitude where the bets are largest, that is the Mirror Principle at its most expensive: if the reasoning going into a capital bet was generic, the bet coming out is generic, however much AI ran on the bench below it. A model can rank a thousand candidate compounds. It cannot make the thinking behind a build-versus-buy call any good. That is a different discipline.
AI pilots cluster low, in the plant and the lab, which is where most materials firms stop. The decisions that move the business, capital across the cycle and the portfolio and molecule bets, sit at the leadership altitude above. The red is the distance between where AI is piloted and where the value actually lives: optimised at the bench, ungoverned in the portfolio.
What the Havruta Methodology installs at portfolio-leadership level
The Havruta Methodology is that discipline. It changes the default behaviour of the machine from agreeing with you to reasoning with you, which is exactly what a capital or portfolio bet needs.
The Flip
The Flip puts the machine on the other side of the question. Instead of confirming the expansion case, it argues against it: where is this demand read soft, what is the cycle about to do that the plan is not pricing in, what would have to be true for this molecule bet to be wrong. The leadership team gets challenged before the market does the challenging.
Ground Truth
Ground Truth keeps the reasoning anchored in the real business, its actual feedstock positions, contracts, regulatory exposure and where it sits in the cycle, rather than the generic industry language an AI produces by default. A capital decision built on a plausible average is worse than no AI at all.
Decision Velocity
And Decision Velocity lets the team decide at the speed the cycle moves, compressing the path from question to defensible capital plan without handing the judgement to the machine.
The fuller account of how all of this works is on the methodology page. If you own the AI transformation itself across the business, see AI for the Head of AI Transformation.
What this is not
This is not process automation and it is not a piece of plant or lab software. It is not predictive-maintenance tooling, a process-control system, a materials-informatics platform, a LIMS, or an MES and historian. It is not AI training or general AI literacy. The tooling and the platforms are a separate market. This is the thinking underneath the decisions they never reach.
It changes how leadership reasons about the decisions it already owns: the capital allocation across the cycle, the portfolio and molecule bet, the build-versus-buy, the footprint and decarbonisation call.
Where a materials leadership team starts
The methodology is installed along a ladder, and a leadership team enters at the rung that fits.
Most begin with the Eye-Opener Workshop, a half-day in which the team sees the shift on its own real work.
A leadership group embeds the practice through the Havruta programme, taking the discipline across the team.
A single high-stakes question, a capacity bet, a molecule or portfolio call, a build-versus-buy decision, can be worked through Advisory Havruta.
How a division leader allocates capital across the portfolio with AI
For the leader who owns the allocation across a portfolio of businesses that all argue for more, the role page takes the same discipline to the seat where the capital call is actually made. A Strategic Briefing is how to decide where to begin.
Go to the Division MD pageLeadership questions about AI in chemicals and materials
Is AI in chemicals and materials a plant project or a leadership decision?
It is both, but most firms only treat it as the first. Predictive maintenance, process optimisation and materials discovery are genuine plant and lab work and worth doing. The decisions that actually move the business, though, capital allocation across the cycle, the portfolio and molecule bet, build-versus-buy, the footprint call, live above the bench. Scoping AI only to the plant and lab leaves those calls to be made the old way or handed to a generic answer. The value you are missing sits at the leadership altitude, not at the bench.
Why do chemicals AI pilots multiply but rarely scale?
Because a plant or lab pilot is cheap to launch and was never scoped to reach the decisions that carry the real money. BCG's data shows only about five percent of companies generating real value from AI at scale, even as adoption goes near-universal. A discovery model can rank candidate compounds, but it cannot improve a capital bet sitting one altitude up. Scaled value follows only when the reasoning at the portfolio and capital level changes too, and that is a discipline, not another tool.
Isn't materials discovery the highest-value place to use AI?
It has reshaped the bench: AI has screened millions of candidate structures that no laboratory could approach. But discovery is bounded. It tells you which compounds are promising; it cannot tell you whether to fund the line that makes the winner, how to time the capital against the cycle, or which product to back when several compete for the same budget. Those calls move far more money than any screen, and they are exactly where AI is least present today.
Is this process-control or a materials-informatics platform?
No. It is not process-control software, a materials-informatics platform, a LIMS, an MES or a historian, and it does not touch your plant or lab stack. Those address the bench. This addresses the thinking above it: how a leadership team reasons through a capital or portfolio decision so the answer is truly theirs, anchored in their real feedstock, contracts and position in the cycle, and stress-tested before the market tests it for them.
How should leadership use AI for capital and portfolio decisions in chemicals?
Start by treating AI as a thinking partner for the decision, not a vending machine for an answer. Make the model argue against the expansion or molecule case rather than confirm it. Anchor it in your real feedstock positions, contracts, regulatory exposure and where you sit in the cycle, rather than generic industry language. Then decide at the speed the cycle moves, without handing the judgement to the machine. The bench pilots tell you what the chemistry can do. This is how you reason about what to do with the business.
Where should we start?
With a Strategic Briefing, or with the Eye-Opener Workshop, where a materials leadership team sees the difference between instructing AI and reasoning with it on its own real work. From there the path depends on whether you are embedding the practice across the leadership team or working a single high-stakes question, such as a capacity bet, a molecule or portfolio call, or a build-versus-buy decision.