Teva's AI Across 30 Plants: Proof of Value Instead of Proof of Concept

On November 4, 2026, IIoT World published a report on a panel from the Industrial AI Summit 2026 about how Teva Pharmaceuticals selects artificial intelligence technologies for its plants. The most visible figure: in one year, through its open innovation program Teva Rise, the company evaluated roughly 300 technologies across its network of 30 plants.

The panelists were Maya Schushan-Orgad of Teva, Adam Procopio of Merck and Kristen Sauter of Adlib Software, with Rick Franzosa of Tech-Clarity moderating. The session was sponsored by Adlib Software. The article notes that AI tools helped summarize the content and that the editorial team reviewed it.

Context: why operational efficiency matters

Operations account for about 70% of Teva's cost structure, so efficiency gains in manufacturing and the supply chain have a direct impact. Meanwhile, the company is shifting from generics to biopharma, a path it calls "Pivot to Growth" and which, according to the article, has delivered 13 consecutive quarters of results.

Start from problems, not technology

The program does not start from a catalog of solutions. Operational experts first define the most relevant problems in manufacturing quality and the supply chain. The examples cited are:

• predicting product quality risk earlier;
• improving asset reliability;
• making better-informed supply chain decisions.

Proof of value: five questions beyond "does it work?"

A proof of concept asks only whether the technology works. Teva's proof of value adds five questions:

• does it work in our environment?
• does it integrate and scale across a network of heterogeneous plants?
• does it deliver measurable value against the problem?
• can it operate in a highly regulated context?
• can users trust the output and act on it?

Teva also checks whether the data actually exists at a given site and whether operators can fit recommendations into current workflows without heavy change management. Each proof of value is designed together with the vendor, with decision-flow mapping and traceability requirements included from the start, not added after release. If a site lacks data or system maturity, even an excellent vendor is put on hold until the site is ready.

The vendor gap

Vendors specialized in pharma build regulatory requirements, explainability and traceability into their products. Those from other sectors often have excellent technology but lack the compliance architecture. In some cases Teva found better technology outside pharma but could not adopt it for exactly this reason.

The data trails generated for traceability serve more than compliance: they feed predictive maintenance models, quality management systems and operational improvements.

What integrators and consultants can take from it

The method is a concrete, large-scale way to get past pilots that never reach production. The criteria are cross-cutting and reusable in smaller settings too:

• data availability at the specific site;
• integration across mixed systems;
• fit with existing workflows;
• traceability of decisions;
• user trust in the output.

Traceability and decision logging must be designed in from day one, especially in regulated industries.

This is the practical takeaway for anyone delivering AI and automation projects: building the decision log together with the solution costs less than adding it later, and it extends its usefulness beyond audits.

Limits of the source

This is a brief panel report, not a case study. The article gives no results from individual proofs of value, no return-on-investment figures and no vendor names. The value lies in the selection method described, not in verifiable results. The topic connects to other cases already covered, such as ALCOA and Stellantis, but here the angle is vendor selection and proof of value.