Verified
Linde
Experimenting- Industrial gases
- 65,177 staff
- ~300 AI use cases (Oct 2025)
- AI under the CHRO
- No AI-only return reported
Linde deploys AI mainly inside its own plants and supply chain: PowerOptimizer, which times power use at air separation units, machine learning that schedules deliveries from tank telemetry, self-learning plant control and valve diagnostics at more than 40 facilities. Sanjiv Lamba said in Oct 2025 that Linde had about 300 AI use cases, steered by an AI council and tracked like productivity projects, and CHRO Desiree Bacher oversees AI. Linde has published no user count, model provider or return it credits to AI alone.
linde.comTheir approach to AI
- The platform
- Linde builds most of its AI in-house through a global AI team and regional AI & Digital teams, tracks every AI project on the internal platform it uses for productivity projects (Oct 2025), and sells its own AI plant control software, AOPS AIPC, to other plant operators.
- The models
- Linde names no model provider; its disclosed tools are machine learning, deep learning and reinforcement learning models for plants and logistics, and vendor case studies describe an LLM knowledge assistant and a plant model pipeline on Microsoft Azure.
- The reach
- Linde has not said how many of its 65,177 employees use AI tools; its 2025 Sustainable Development Report says employees from all areas of the business have taken part in digitalization training under a digital and AI upskilling program open to all staff.
- The catch
- Linde credits about a third of its productivity gains to AI and digital work together (May 2025) and 36% of 2025 productivity savings to digital initiatives, but has never given AI's own share in dollars, headcount or margin.
Snapshot
- Employees
- 65,177
- Dec 31, 2025, about 28% women (2025 Form 10-K)
- Technology budget
- Not disclosed
- Research and development spending was $147M in 2025, aimed at gas and process technology (2025 Form 10-K)
- AI/ML headcount
- Not disclosed
- A global AI team plus global and regional AI & Digital teams (2025 Sustainable Development Report)
- Who owns AI
- Desiree Bacher, SVP and CHRO
- Oversees AI and digitalization and reports to the CEO; the board's Human Capital Committee oversees AI policy
- Platform
- In-house AI, no named employee platform
- AI projects reviewed on the internal productivity platform (Oct 2025); AOPS sold to other operators
- Models
- Not disclosed
- Machine learning, deep learning and reinforcement learning for plants; an LLM assistant per a vendor
- Reach
- Not disclosed
- Digital and AI upskilling open to all employees (2025 Sustainable Development Report)
- Use cases
- ~300+
- Sales, engineering and design, operations and distribution (Sanjiv Lamba, Oct 2025)
Key findings
Sanjiv Lamba said on Linde's Oct 31, 2025 earnings call that Linde has about 300 AI use cases or more, up from the 105 he said were being deployed on May 1, 2025, and that an AI council steers deployment while each project is reviewed and validated on the internal platform Linde uses for productivity projects.
Linde has not credited a business return to AI alone since May 2025: Sanjiv Lamba put about 31% to 32% of productivity efforts down to digital and AI solutions in May 2025, and the 2025 Sustainable Development Report credits digital productivity initiatives with about 36% of 2025 productivity savings, neither with an AI share or a dollar value.
Linde's 2025 Sustainable Development Report, published Jun 1, 2026, says SVP and CHRO Desiree Bacher oversees AI, that related policies fall under the board's Human Capital Committee, and that Linde has global and regional AI & Digital teams and a cross-functional AI council.
Linde's AI Council approved a Global AI Policy on Jul 31, 2026 that covers every application with an AI component, including vendor tools, and lets Linde confidential information be processed only through AI systems Linde has approved.
Linde's AI runs mainly in its own operations: PowerOptimizer for air separation units, telemetry-driven delivery scheduling, AOPS AI plant control, and ValveWatch valve diagnostics at more than 40 air separation and hydrogen facilities in 2025.
Where Linde sits
- Stage
- Experimenting
Linde has reported no business return it credits to AI alone since May 2025. The figures on its record share the credit: Sanjiv Lamba said on May 1, 2025 that about 31% to 32% of Linde's productivity efforts come from digital and AI solutions, and the 2025 Sustainable Development Report (Jun 2026) says digital productivity initiatives contributed about 36% of 2025 productivity savings, without an AI share, a dollar value or a headcount figure. The AI plant control results Linde presented in Feb 2026 (2% to 3% more product yield, 1% to 2% less energy) are plant-level figures from a product pitch, not a company return. Linde has not reported a company-wide rollout with a user count either: it gives no count of employees using AI tools, and AI sits with CHRO Desiree Bacher alongside human resources and communications. Linde's AI work since May 2025 is on the public record, from about 300 use cases and an AI council (Oct 2025) to ValveWatch at more than 40 facilities and a Global AI Policy approved on Jul 31, 2026, so Linde sits at Experimenting rather than Unknown. A cost, energy or headcount figure Linde credits to AI in one use case, such as PowerOptimizer, or a stated count of employees using AI tools under a named owner, would move Linde to Piloting.
How the stages are defined - Ownership
- SVP and CHRO Desiree Bacher oversees AI and digitalization; Henning Tomforde leads Global AI & Digital
- Platform
- Hybrid: AI built in-house by a global AI team, with outside builders and Azure
- Models
- Not disclosed; in-house machine learning and reinforcement learning models
- Hosting
- Not disclosed by Linde; vendor case studies put a plant model pipeline and a knowledge assistant on Microsoft Azure
- Reach
- Not disclosed; upskilling program open to all 65,177 employees
How they deploy
AI as part of the productivity machine
Linde treats AI as one source of the productivity projects it runs across the company: Sanjiv Lamba said Linde completed 15,515 productivity projects in 2024 and more than 4,000 in the first quarter of 2025, and that AI projects are tracked the same way, on an internal platform where they are reviewed and validated (Oct 2025). Lamba said Linde's AI team has a stretching goal for the benefit it must deliver and that every use case is judged on its business case.
From single use cases to domains
Sanjiv Lamba said in Oct 2025 that Linde was moving from standalone AI use cases to introducing AI tools across whole domains, and expected the work to scale over the next 2 to 3 years. Lamba said Linde draws on more than 30 years of plant data and 4 to 5 years of machine learning work, much of it already deployed.
Build in-house, with outside help
Linde's AI has been built mostly by its own people: a central AI team that develops solutions and runs AI training, and digital teams inside the business units, according to a 2023 case study by appliedAI, Linde's partner since 2017. Linde has also used outside builders, including STX Next for an employee knowledge assistant and PRODATO for a plant model pipeline, according to those firms' case studies.
Selling plant AI to others
Linde Engineering sells AI plant control to other plant operators through Linde Advanced Operations Solutions: AOPS AIPC, a self-learning control system Linde first ran in closed loop on a small air separation unit in 2021, was presented to an energy industry audience in Feb 2026 with results from steel, chemical and electronics gas plants.
- Deployment model
- Linde runs AI through a global AI team and global and regional AI & Digital teams, under SVP and CHRO Desiree Bacher, who oversees AI and digitalization and reports to the CEO (2025 Sustainable Development Report). Linde's leadership page lists Henning Tomforde as the functional leader for Global AI & Digital, and Handelsblatt described Dexin Luo in May 2026 as Linde's Global Vice President, Digital, AI and IT. A cross-functional AI council sets the strategic frame and oversees deployment, and responsible-use programs are run at regional level. Information technology sits separately: Linde's Global CIO reports to the CFO and the Chief Information Security Officer reports to the CIO (2025 Form 10-K).
- Model providers
- Linde has not named an AI model provider. Linde's disclosed AI uses its own models: PowerOptimizer for air separation units, machine learning models that predict customer tank levels from telemetry (May 2025), reinforcement learning and deep learning models in AOPS AI plant control, and machine learning models in ValveWatch. STX Next says it built Linde a retrieval-augmented generation assistant in which a large language model answers employee questions from Linde's internal documents, without naming the model or the date.
- Data and infrastructure
- Linde runs more than 1,200 plants, operated from 12 remote operation centers where one operator can run 10 plants, according to a Linde Advanced Operations Solutions presentation from Feb 2026. Sanjiv Lamba said in Oct 2025 that Linde has captured plant data for more than 30 years. 2 vendor case studies place Linde AI on Microsoft Azure: PRODATO moved a Linde Engineering air separation model pipeline into Azure Machine Learning and Azure Data Lake, saying it lets Linde train and deploy models for more than 2,800 plants in over 80 countries, and STX Next hosted the employee knowledge assistant on Azure so data stays in Linde's cloud. Linde says its data center footprint is not significant (2025 Sustainable Development Report).
Platforms and internal tools
PowerOptimizerBy May 2025
PowerOptimizer is an AI optimization tool built by Linde's Global AI and Digital team that plans each hour's production at air separation units with bulk storage from electricity price forecasts, ambient temperature, product demand and plant limits, and feeds the plan into the plant's control system, ramping up when power is cheap. Linde says energy can be 80% to 90% of an air separation unit's production cost. PowerOptimizer was piloted at the Roethenbach plant in Germany, rolled out to plants in Germany, Switzerland and the UK, and is being extended to Southeast Asia, New Zealand and the USA, according to an undated Linde article; Sanjiv Lamba described it on the May 1, 2025 call.
- Users
- Plants in Germany, Switzerland and the UK, with more being added
AOPS AI Plant Control (AIPC)2021, first closed-loop control
AOPS AIPC is Linde Advanced Operations Solutions' self-learning plant control software, trained offline on plant data instead of step tests and retrained within hours when operating modes change. Linde ran it in closed loop on a small air separation unit in 2021, then at plants supplying steelworks, a chemical complex and ultra-high-purity electronic gas production by 2025, with a refinery next; Linde also offers AOPS to other operators.
- Models
- Deep learning and reinforcement learning models built by Linde
ValveWatchExpanded in 2025
ValveWatch is Linde's AI valve health diagnostics platform: machine learning models watch valve behavior in real time, flag drifting, sticking or leaking valves, and point to likely root causes so plant teams can maintain them before they fail. Linde expanded ValveWatch to more than 40 air separation and hydrogen facilities in 2025.
- Users
- 40+ facilities (2025)
AIRGENIOUS
AIRGENIOUS is an AI engine built by Linde's homecare business with Linde's global AI team that predicts which sleep apnea patients will stop using their CPAP devices and recommends an intervention, such as a text message or a call, with clinicians deciding. Linde Médica presented preliminary results from a Spanish study of 679 patients at the European Respiratory Society congress in Sep 2026.
Employee knowledge assistant
STX Next says it built Linde a multilingual retrieval-augmented generation assistant, hosted on Microsoft Azure, that lets employees ask questions in their own language about equipment specifications, safety protocols, maintenance procedures and policies and get answers with citations from Linde documents. Linde has not described the tool, its users or its launch date.
- Models
- Large language model, provider not named
Vendors and partners
Microsoft Azure
Cloud for a plant model pipeline and an employee knowledge assistant, according to vendors · PRODATO and STX Next each say they built Linde AI on Azure; Linde has not named a cloud provider for AI itself.
appliedAI Initiative2017
AI training, workshops and the AIPlantControl reinforcement learning project · appliedAI's 2023 case study says Linde ran its first AI pilot in 2016 and joined appliedAI as a partner in 2017.
Siemens and TU DelftApr 2024
P&ID Co-Pilot, a generative AI tool for drafting piping and instrumentation diagrams · TU Delft's Process Intelligence Research Group announced the project with Siemens and Linde on Apr 18, 2024; no results have been published since.
STX Next
Built Linde's retrieval-augmented generation knowledge assistant · Undated vendor case study.
PRODATO
Moved a Linde Engineering machine learning pipeline for air separation plants into Azure · Undated vendor case study published by BARC.
Universidad Autónoma de Madrid
AIRGENIOUS research through the UAM-Linde chair · The Spanish AIRGENIOUS study presented in Sep 2026 is run within the Cátedra UAM-Linde.
Leadership
AI, data and platform leaders
Executives
AI org map
AI and digital under the CHRO, with global and regional AI & Digital teams and an AI council; IT under the CFO
Linde places AI and digitalization under SVP and CHRO Desiree Bacher, who reports to CEO Sanjiv Lamba, and the board's Human Capital Committee oversees AI policies and programs (2025 Sustainable Development Report). A cross-functional AI council acts as Linde's AI oversight board and approved the Global AI Policy on Jul 31, 2026. Information technology is a separate function: the Global CIO reports to CFO Matt White, and cybersecurity falls under the board's Audit Committee (2025 Form 10-K).
Linde's AI workforce sits in a global AI team, which builds AI tools and runs AI training, and in global and regional AI & Digital teams, with digital teams also embedded in business units (2025 Sustainable Development Report; appliedAI, 2023). Linde has not published how many data scientists or AI engineers it employs. Linde Engineering's Advanced Operations Solutions unit develops and sells the AOPS plant automation and AI control software.
Reporting lines
CEO and Chairman
Sanjiv Lamba
Human Resources, Communications, AI & Digitalization
Desiree Bacher, SVP and CHRO
Global AI & Digital
Line inferred from the CHRO's remit for AI and digitalization; Linde has not published it
Digital, AI and IT
Dexin Luo, Global Vice President
Title from Handelsblatt (May 2026); reporting line and relation to Global AI & Digital not published
AI Council
Cross-functional AI oversight board with working groups; members not published; line to the CEO inferred
Finance
Matt White, EVP and CFO
Information Technology
Linde's Global CIO reports to the CFO and the CISO to the CIO (2025 Form 10-K); Skibo's title as CIO is not stated in a Linde source
Engineering
Stefano Innocenzi, SVP
Includes Linde Advanced Operations Solutions, which sells AOPS AI plant control; line to the CEO inferred
Operations Excellence & Productivity
Shannon Lenze
Reporting line not published
Linde's leadership page, checked Nov 5, 2026, lists Henning Tomforde for Global AI & Digital and Mike Skibo for Information Technology under functional leadership, while Handelsblatt called Dexin Luo Linde's Global Vice President, Digital, AI and IT in May 2026; Linde has not published how these roles relate or whom Tomforde and Luo report to.
Where the AI workforce sits
Global AI team and global AI & Digital team
Linde has published no AI headcount; the central team builds AI tools and runs AI training (appliedAI, 2023).
Not disclosedRegional AI & Digital teams and digital teams in business units
Responsible-use programs are managed regionally (2025 Sustainable Development Report).
Not disclosedLinde Advanced Operations Solutions (Engineering)
Builds the AOPS automation and AI control software sold to Linde plants and outside operators.
Not disclosed
Named people
Use cases
Linde had about 300 AI use cases or more in Oct 2025, Sanjiv Lamba said, from the front end of sales to engineering and design, and the named ones run mostly in operations: PowerOptimizer for air separation units, telemetry-driven delivery scheduling, AI plant control, ValveWatch valve diagnostics and smart plant inspections in China. Linde has published no return it credits to AI alone for any use case. Its firm-level figures credit AI and digital work together, about 31% to 32% of productivity efforts in May 2025, or digital initiatives alone, about 36% of 2025 productivity savings, with no AI share.
| Use case | Area | Status | ROI |
|---|---|---|---|
| PowerOptimizer for air separation units PowerOptimizer uses AI to schedule air separation unit production hour by hour against electricity prices, demand and tank levels, and runs at plants in Germany, Switzerland and the UK, with rollout under way in Southeast Asia, New Zealand and the USA; Linde says it brings significant cost savings without giving a figure, and Sanjiv Lamba said optimizing power by even a small percentage has a big impact (May 2025). | Plant operations | In production | |
| Telemetry and predictive delivery scheduling Linde deploys telemetry in customer tanks and machine learning models that track each customer's usage and schedule bulk deliveries so trucks and tanks are used efficiently and customers are refilled at the right level (May 2025). | Distribution | In production | |
| Smart plant operations in China Linde uses drones and robots to take routine readings at air separation units in China and feeds them into an AI program that raises alerts for human intervention; Sanjiv Lamba said in May 2025 that Linde had won approval in a couple of Chinese provinces to reduce manning at such sites, without giving a number. | Plant operations | In production | |
| AOPS AI Plant Control AOPS AIPC controls air separation and other process plants with self-learning models that adjust valves and compressor loads; Linde's Feb 2026 presentation cites 2% to 3% more product yield, 1% to 2% less energy and 3 to 12 month paybacks over 3 years of operation at selected plants. | Engineering and plant operations | In production | |
| ValveWatch valve diagnostics ValveWatch uses machine learning to spot and classify valve faults in real time at more than 40 air separation and hydrogen facilities, which Linde says improved control stability and cut avoidable energy loss at early sites (2025). | Plant maintenance | In production | |
| AIRGENIOUS sleep apnea adherence AIRGENIOUS predicts which CPAP patients are at risk of stopping treatment and recommends interventions; Linde Médica's preliminary Spanish results (Sep 2026) showed about 0.13 to 0.15 hours more nightly use than standard follow-up, and the study's final analysis is still to come. | Healthcare | Pilot | |
| AI chatbots for customers Linde lists online AI-based chatbots among the channels it uses to engage customers on safety, product knowledge, reliability and quality (2025 Sustainable Development Report). | Customer service | In production | |
| Employee knowledge assistant A multilingual retrieval-augmented generation assistant on Microsoft Azure answers employee questions from Linde's technical, policy and operational documents with source citations, according to its builder STX Next; Linde has not described the tool. | Firmwide | In production | |
| P&ID Co-Pilot Linde, Siemens and TU Delft began building a generative AI tool in Apr 2024 that translates process flow diagrams into draft piping and instrumentation diagrams using transformer language models. | Engineering | Pilot |
Adoption
Linde spread AI from a first pilot in 2016 (appliedAI) to 105 AI use cases being deployed in May 2025 and about 300 or more by Oct 2025, Sanjiv Lamba said, most of them machine learning on plant, logistics and customer data rather than employee chat tools. Linde has published no count of employees using AI tools. Linde's 2025 Sustainable Development Report says a digital upskilling program, open to all employees, runs 1 to 3 month sessions with coaching, boot camps, no-code and low-code tools and AI basics, and that employees from all areas of the business have taken part; employees also take annual Code of Business Integrity and cybersecurity training that covers responsible AI use. Linde's AI plant control has moved from one small air separation unit in 2021 to plants supplying steel, chemical and electronics customers by 2025, and ValveWatch reached more than 40 facilities in 2025.
By the numbers
Rollout, first pilot to today
2016
Linde ran its first AI pilot project, according to appliedAI's case study on Linde's AI plant control.
2017
Linde joined the appliedAI initiative as a partner for AI training, workshops and the AIPlantControl reinforcement learning project.
2021
Linde ran its AI plant control software in closed loop for the first time, on a small air separation unit.
Apr 18, 2024
TU Delft announced the P&ID Co-Pilot, a generative AI tool for piping and instrumentation diagrams built with Siemens and Linde.
2025
Linde expanded ValveWatch, its AI valve diagnostics platform, to more than 40 air separation and hydrogen facilities.
May 1, 2025
Sanjiv Lamba said Linde had 105 AI use cases being deployed, including an AI power optimizer and telemetry-driven distribution, and that about 31% to 32% of its productivity efforts come from digital and AI solutions.
Sep 1, 2025
Desiree Bacher, previously Senior Vice President of Communications, AI and Corporate Procurement, became CHRO and kept oversight of AI and digitalization.
Oct 31, 2025
Sanjiv Lamba said Linde had about 300 AI use cases or more, an AI council overseeing deployment, and a shift from standalone use cases to AI across whole domains.
Jul 31, 2026
Linde's AI Council approved a Global AI Policy covering all Linde entities and every application with an AI component.
Economics & controls
Spend and ROI
Technology budget
Linde does not disclose a technology or AI budget. Linde's research and development spending was $147 million in 2025, against $150 million in 2024, and is aimed at gas processing, separation, liquefaction and clean energy technology rather than AI (2025 Form 10-K). Linde says its data center environmental footprint is not significant (2025 Sustainable Development Report).
What AI itself costs
Linde has not published what it spends on AI, on models or on cloud capacity for AI. Linde Advanced Operations Solutions says its AI plant control pays back its initial investment in 3 to 12 months at the plants where it has run, a figure for the product rather than for Linde's AI program (Feb 2026).
Reported ROI
Linde has reported no return it credits to AI alone. Sanjiv Lamba said on May 1, 2025 that about 31% to 32% of all Linde's productivity efforts come from digital and AI solutions, a share of productivity work credited to the 2 together with no dollar value. Linde's 2025 Sustainable Development Report, published Jun 2026, says digital productivity initiatives contributed about 36% of overall productivity savings in 2025 and that Linde tracks AI-related productivity initiatives, which often bring financial and resource savings, but gives no AI share. Dexin Luo discussed how Linde calculates the return on AI investments on Handelsblatt's Disrupt podcast in May 2026; no figure from that conversation has been published in writing.
Productivity markers
Linde completed 15,515 productivity projects in 2024 and more than 4,000 in the first quarter of 2025, and tracks AI projects on the same internal platform (Sanjiv Lamba, May and Oct 2025). Linde's Feb 2026 presentation on AOPS AI plant control cites, for selected plants, 35% faster load changes, 2% to 3% more product yield and 1% to 2% lower energy use compared with classical advanced process control, and one operator running 10 plants from a remote operation center.
Headcount
Linde had 65,177 employees at Dec 31, 2025 (2025 Form 10-K). Linde booked $273 million of cost reduction program and other charges in 2025, and CFO Matt White said in Feb 2026 that those restructuring charges relate mainly to headcount, most of them in Engineering. Linde recorded a further $425 million restructuring charge in the third quarter of 2026, about half of it severance in lower-growth regions. Linde has not credited any of these reductions to AI; the one AI-linked staffing change on record is Sanjiv Lamba's May 2025 remark that drones, robots and AI let Linde reduce manning at some sites in a couple of Chinese provinces, without a number.
Governance and risk
Ownership and board oversight
Linde's SVP and CHRO Desiree Bacher oversees AI, and AI policies and programs fall under the board's Human Capital Committee (2025 Sustainable Development Report). Linde's Audit Committee oversees information and cybersecurity risk, with a Global CIO reporting to the CFO and a Chief Information Security Officer reporting to the CIO (2025 Form 10-K).
Global AI Policy and AI Council
Linde's Global AI Policy, approved by the AI Council on Jul 31, 2026, makes the AI Council Linde's AI oversight board, sets 10 AI principles, among them safety, human-centered use, accuracy, IP compliance, transparency and explainability, fairness, security, data protection and accountable use, and applies to every application with an AI component, vendor and third-party tools included. Linde regions and segments may add their own procedures to meet local law.
Data boundary
Linde's Global AI Policy allows Linde intellectual property, confidential information and know-how to be processed only through AI systems and platforms Linde has approved, and the 2025 Sustainable Development Report says employees may use only Linde IT-approved AI applications. STX Next says the employee knowledge assistant it built keeps Linde data inside Linde's Azure environment.
Controls on deployed AI
Linde's responsible-use programs, managed at regional level and still being implemented when the 2025 report came out in Jun 2026, restrict sensitive AI capabilities through role-based controls and approvals, label AI-generated content and AI-supported decisions, let stakeholders contest AI-supported outcomes, monitor deployed models for drift and retrain them, and assess fairness and bias (2025 Sustainable Development Report). AIRGENIOUS keeps clinicians in the loop: the AI ranks patients and suggests interventions, and care teams decide and act.
Training and regulation
Linde requires every employee to complete annual Code of Business Integrity and cybersecurity training that includes responsible AI use, and the Global AI Policy adds AI training for employees exposed to AI and for staff who build, buy or run AI. Linde's 2025 Form 10-K lists data protection, including artificial intelligence, among the regulated areas whose changes could raise compliance costs.
Recent signals
Oct 29, 2026
Linde's third-quarter 2026 call reported electronics sales up 15% and about $2 billion of electronics projects under construction, alongside a $425 million restructuring charge; Linde's executives did not discuss AI in Linde's own operations on the call.
Sep 7, 2026
Linde Médica presented preliminary AIRGENIOUS results at the European Respiratory Society congress in Barcelona: 679 patients in Spain, with AI-guided follow-up showing slightly higher CPAP use than standard follow-up over 90 days.
Jul 31, 2026
Linde's AI Council approved a Global AI Policy that sets 10 AI principles and allows Linde confidential information only in AI systems Linde has approved.
Jul 31, 2026
Sanjiv Lamba said on the second-quarter 2026 call that electronics was Linde's fastest-growing end market, with demand tied to hardware for AI, and that Linde added $1 billion of new electronics wins to its backlog.
Jun 1, 2026
Linde's 2025 Sustainable Development Report said the CHRO oversees AI, described its AI council and responsible-use controls, reported ValveWatch at more than 40 facilities, and credited digital productivity initiatives with about 36% of 2025 productivity savings.
May 8, 2026
Dexin Luo, introduced as Linde's Global Vice President, Digital, AI and IT, discussed on Handelsblatt's Disrupt podcast how Linde calculates the return on its AI investments.
May 1, 2026
Linde said on its first-quarter 2026 call that electronics sales grew 10%, driven by investment in advanced chips to support AI.
Feb 25, 2026
Linde's 2025 Form 10-K reported 65,177 employees and listed data protection, including artificial intelligence, among the regulations that could raise its compliance costs.
Feb 19, 2026
Stephan Hazzam of Linde Advanced Operations Solutions presented results from 3 years of AOPS AI plant control at Go Digital Energy in London, including 2% to 3% higher product yield at selected plants.
Feb 5, 2026
Sanjiv Lamba said on the fourth-quarter 2025 call that investment in AI and digital infrastructure drove unprecedented activity in 2025 while traditional industrial markets retrenched, and that Linde had started more restructuring for 2026.
Quick answers
- How does Linde deploy AI?
- Linde deploys AI mainly inside its own plants and supply chain: PowerOptimizer, which times power use at air separation units, machine learning that schedules deliveries from tank telemetry, self-learning plant control and valve diagnostics at more than 40 facilities. Sanjiv Lamba said in Oct 2025 that Linde had about 300 AI use cases, steered by an AI council and tracked like productivity projects, and CHRO Desiree Bacher oversees AI. Linde has published no user count, model provider or return it credits to AI alone.
- What does Linde use AI for?
- Linde had about 300 AI use cases or more in Oct 2025, Sanjiv Lamba said, from the front end of sales to engineering and design, and the named ones run mostly in operations: PowerOptimizer for air separation units, telemetry-driven delivery scheduling, AI plant control, ValveWatch valve diagnostics and smart plant inspections in China. Linde has published no return it credits to AI alone for any use case. Its firm-level figures credit AI and digital work together, about 31% to 32% of productivity efforts in May 2025, or digital initiatives alone, about 36% of 2025 productivity savings, with no AI share.
- Who leads AI at Linde?
- AI at Linde is led by Desiree Bacher (Senior Vice President and Chief Human Resources Officer; oversees AI and digitalization); Henning Tomforde (Functional leader, Global AI & Digital); Dexin Luo (Global Vice President, Digital, AI and IT). Ownership of AI on the org chart: SVP and CHRO Desiree Bacher oversees AI and digitalization; Henning Tomforde leads Global AI & Digital.
- What AI platform does Linde use?
- Linde builds most of its AI in-house through a global AI team and regional AI & Digital teams, tracks every AI project on the internal platform it uses for productivity projects (Oct 2025), and sells its own AI plant control software, AOPS AIPC, to other plant operators.
- Which AI models does Linde use?
- Linde names no model provider; its disclosed tools are machine learning, deep learning and reinforcement learning models for plants and logistics, and vendor case studies describe an LLM knowledge assistant and a plant model pipeline on Microsoft Azure.
- How many Linde employees use AI?
- Linde has not said how many of its 65,177 employees use AI tools; its 2025 Sustainable Development Report says employees from all areas of the business have taken part in digitalization training under a digital and AI upskilling program open to all staff.
- What return has Linde reported from AI?
- Linde has reported no business return it credits to AI alone since May 2025. The figures on its record share the credit: Sanjiv Lamba said on May 1, 2025 that about 31% to 32% of Linde's productivity efforts come from digital and AI solutions, and the 2025 Sustainable Development Report (Jun 2026) says digital productivity initiatives contributed about 36% of 2025 productivity savings, without an AI share, a dollar value or a headcount figure. The AI plant control results Linde presented in Feb 2026 (2% to 3% more product yield, 1% to 2% less energy) are plant-level figures from a product pitch, not a company return. Linde has not reported a company-wide rollout with a user count either: it gives no count of employees using AI tools, and AI sits with CHRO Desiree Bacher alongside human resources and communications. Linde's AI work since May 2025 is on the public record, from about 300 use cases and an AI council (Oct 2025) to ValveWatch at more than 40 facilities and a Global AI Policy approved on Jul 31, 2026, so Linde sits at Experimenting rather than Unknown. A cost, energy or headcount figure Linde credits to AI in one use case, such as PowerOptimizer, or a stated count of employees using AI tools under a named owner, would move Linde to Piloting.
Open questions
- What share of Linde's productivity savings comes from AI alone, in dollars, given that Linde credits about 31% to 32% of productivity efforts to digital and AI together (May 2025) and about 36% of 2025 savings to digital initiatives?
- What does PowerOptimizer save Linde in power costs each year, given that Linde calls the savings significant without a figure, and how much of the AOPS AI plant control yield and energy gains accrue to Linde's own plants rather than customers' plants?
- How did Dexin Luo describe Linde's method for calculating AI returns on Handelsblatt's May 2026 podcast, and will Linde publish the result?
- How many Linde employees use generative AI tools such as the knowledge assistant STX Next built, and which model providers sit behind them?
- How do Henning Tomforde's Global AI & Digital function and Dexin Luo's Digital, AI and IT role relate, and whom does each report to?
- Who sits on Linde's AI Council, and does it report to the CEO, the CHRO or the board's Human Capital Committee?
- How many sites and staff were affected by the AI-assisted reduction in manning in China that Sanjiv Lamba mentioned in May 2025?
- How many data scientists and AI engineers does Linde employ, and where are they based?
- Which clouds run Linde's AI models beyond the Microsoft Azure work its vendors describe, and how much runs on site in plant control systems?
- What did the P&ID Co-Pilot with Siemens and TU Delft produce after its Apr 2024 launch?
Sources
Linde
- Linde leadership, executive and functional leaders (checked Nov 5, 2026)
- Desiree Bacher, SVP and CHRO, biography
- Sanjiv Lamba, CEO and Chairman, biography
- Linde appoints Sanjiv Lamba as Chairman and CEO (Sep 29, 2025)
- Linde Global Artificial Intelligence Policy (approved Jul 31, 2026)
- Linde publishes 2025 Sustainable Development Report (Jun 1, 2026)
- AIRGENIOUS: enabling better patient care with AI (Linde)
- How Linde is transforming energy management with AI, PowerOptimizer (Linde)
- Stephan Hazzam, Linde Advanced Operations Solutions: realised benefits from AI plant controls (Feb 2026)
Filings, reports and earnings calls
- Linde 2025 Form 10-K (filed Feb 25, 2026)
- Linde 2025 Sustainable Development Report (Jun 2026)
- Q3 2026 earnings call transcript (Oct 29, 2026)
- Q2 2026 earnings call transcript (Jul 31, 2026)
- Q1 2026 earnings call transcript (May 1, 2026)
- Q4 2025 earnings call transcript (Feb 5, 2026)
- Q3 2025 earnings call transcript (Oct 31, 2025)
- Q2 2025 earnings call transcript (Aug 1, 2025)
- Q1 2025 earnings call transcript (May 1, 2025)
Partners and vendors
Data confidence
- Confirmed
- Stated in a primary or reputable public source.
- Estimated
- Derived or approximated from public reporting.
- Unconfirmed
- Reported, but not yet corroborated.
Linde profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 5, 2026
