How manufacturers, airlines and logistics companies deploy AI
Manufacturers, airlines, logistics companies and consulting firms make money from AI that reads operational data. FedEx's models predict failures in sorting equipment and save $10 million or more a year, and Caterpillar's AI turns machine data into dealer service leads that closed $1.1 billion of sales in a year. AI coding tools cost Uber more than it planned: Uber used up its 2026 budget for them in about 4 months, then capped spending at $1,500 a month per employee for each tool.
Verified See methodology
How manufacturers, airlines and logistics companies have deployed AI
- Piloting
Caterpillar
Construction and mining equipment, engines and turbines
Helios, the cloud data platform Caterpillar built in-house about 2020 and runs mostly on AWS services, holds more than 16 petabytes and carries Cat AI Assistant, a group of AI agents Caterpillar launched for customers, dealers and technicians in early 2026.
$1.1B
Sales closed from AI-generated service leads
- Experimenting
Boeing
Aerospace and defense
Boeing Conversational AI, unveiled in 2023 as Boeing's first large-scale generative AI platform, lets employees query company data inside Boeing's network; Boeing ran more than 70 generative AI applications by Dec 2025.
70+
Generative AI applications
- Experimenting
GE Aerospace
Aircraft engines, services and systems
AI Wingmate, an internal generative AI chat app GE Aerospace built with Microsoft in spring 2024 and launched in Jun 2024, runs inside the company's virtual private cloud; Palantir's AIP runs in select supply chain functions (Mar 2026).
5-7 days
Shop-visit turnaround cut by the AI Material Assistant
- Experimenting
Honeywell
Building, process and industrial automation
Forge, the software layer Honeywell built and sells, connected more than 324,000 customer sites in Mar 2026 and gained Forge Cognition, an agentic AI and digital twin platform, in Oct 2026.
~$1B
Forge software recurring revenue
- Experimenting
John Deere
Agricultural, construction and forestry machinery
John Deere runs AI through a central technology stack it built in-house, from cameras and compute on the machine to the cloud-based John Deere Operations Center, where the JD generative AI assistant entered early access in Sep 2026.
5M+
Acres sprayed with See & Spray (2025)
- Experimenting
Siemens
Industrial technology: automation, industrial software, infrastructure and rail
SiemensGPT, an internal platform Siemens built on AWS with implementation partner FNTIO, gives employees vendor-agnostic access to large language models and a no-code builder for bots and agents (AWS case study, Oct 2026).
80,000
Monthly SiemensGPT users
- Scaling
FedEx
Transportation and logistics
Atlas, the enterprise data platform FedEx built, held a substantial portion of company data and supported more than 200 AI use cases in Feb 2026, and FedEx plans to move all of its data onto Atlas by the end of 2027.
$10M+ a year
Predictive maintenance savings
- Piloting
Uber
Ride-hailing, delivery and freight
Michelangelo, Uber's in-house machine learning platform since 2016, plus an AI Gateway to outside and self-hosted models, an agent platform built in early 2025 and an MCP Gateway with more than 800 servers (Oct 2026).
2x
Code output per engineer with AI coding tools
- Experimenting
A.P. Moller - Maersk
Container shipping and integrated logistics
A.P. Moller - Maersk gives employees Microsoft Copilot and, according to vendor Dify, replaced its in-house LLM app platform with Dify Enterprise, which ran about 90 production apps by late 2025.
~2,900
Technology professionals in India
- Experimenting
American Airlines
Airlines
American Airlines builds its operational AI in-house, such as the Smart Gating machine learning tool, runs key workloads on Microsoft Azure, its preferred cloud since May 2022, and says some of its AI is licensed from third parties (2025 Form 10-K).
5 of 9
Hubs running Smart Gating
- Experimenting
Delta Air Lines
Airlines
Delta Air Lines builds operational AI in-house, such as Baggage AI in Atlanta, runs most systems on AWS, and buys specialist AI: Fetcherr for fare recommendations and Airbus's Skywise platform for predictive maintenance.
100%
SkyMiles members with Delta Concierge
- Experimenting
Southwest Airlines
Airlines
Southwest Airlines builds AI on Amazon Web Services, its preferred cloud since Mar 2023, and plans to move from a largely on-premises environment to an AI- and agent-enabled architecture on AWS by 2028 (Jun 2026).
2,700+
Developers using Kiro
- Experimenting
Union Pacific
Freight rail transportation
UP Chat, an internally developed large language model platform, gives Union Pacific staff generative AI inside the company network (2025 Building America Report, Sep 2026), and Union Pacific's AI transportation planning runs on a platform the railroad built (Jan 2025).
8-10
Major AI projects in use (Apr 2026)
- Experimenting
United Airlines
Airlines
United Airlines builds AI on AWS, its preferred cloud since Nov 2021: CIO.com reported in Dec 2024 that United uses its own United Data Hub and a Bedrock-based ML platform called Mars, after United spent several hundred million dollars from about 2020 to leave legacy mainframes.
3.3M+
Connections saved by ConnectionSaver
- Experimenting
UPS
Package delivery and logistics
UPS builds its own AI tools, among them MeRA for customer email, Deal Manager for pricing, ORION for delivery routes and Next Gen Brokerage for customs; its primary IT operations sit in an owned New Jersey facility, and it has worked with Google Cloud since 2019.
~90%
Cross-border transactions processed digitally
- Piloting
McKinsey & Company
Management consulting
Lilli, which McKinsey & Company launched in 2023, searches and synthesizes more than 100,000 of the firm's documents and interviews, and is the only AI tool where staff may enter confidential client data (Jun 2025).
25,000
AI agents
- Piloting
PwC
Audit, tax and consulting
ChatPwC, an assistant PwC built on Microsoft's Azure OpenAI Service in 2023, sits beside agent OS, the agent orchestration platform PwC launched in Mar 2025, and PwC One, an AI platform PwC launched for its teams and clients in Mar 2026.
285,000+
Microsoft Copilot users
- Experimenting
Deloitte
Professional services: audit, consulting, tax and advisory
Deloitte builds AI tools per member firm and business, such as Solaria in Europe, MyAssist in Australia, Omnia for audit and Ascend for consulting delivery, buys enterprise assistants from Anthropic and Google, and runs its own NVIDIA AI factory (Mar 2026).
470,000+
People with access to Claude
Findings
What the industrials profiles show
FedEx is the only Industrials company at Scaling, on returns it credits to AI in 2 functions: proprietary models that predict failures in sorting equipment save $10 million or more a year (Feb 2026), and AI that catches non-standard packages missing a surcharge recovers tens of millions of dollars of revenue a year (Sep 2026). Uber, at Piloting, reports one such return, code output per engineer doubled on its current measure with AI coding tools, which let Uber moderate engineering hiring (Aug 2026). Neither FedEx nor Uber credits a firm-level figure to AI, and FedEx does not say how much of its $4 billion of DRIVE savings came from AI.
Caterpillar is the only manufacturer with a return it credits to AI: $1.1 billion of closed-won sales in 2024 from service leads its AI engine builds from 30 data sources in Helios and sends to dealers, reported at its Nov 2025 Investor Day. Boeing, GE Aerospace, Honeywell, John Deere and Siemens report operating gains or customer results instead, such as shop turnaround cut by more than 5 days where GE Aerospace's AI Material Assistant runs (Jan 2026), more than 17 inspection hours saved per Boeing airplane by a part-validation tool (Dec 2025) and 50% to 60% herbicide savings for John Deere's See & Spray customers (May 2026). Honeywell's roughly $1 billion of Forge software recurring revenue (Jul and Oct 2026) is sales to customers, not a return on Honeywell's own use of AI.
No airline has reported a business return it credits to AI since May 2025, and American Airlines, Delta Air Lines, Southwest Airlines and United Airlines are all at Experimenting. Airlines report operating and customer measures instead, such as a nearly 30% improvement in bag transfers from Delta Air Lines' Baggage AI in Atlanta (Dec 2025) and more than 3.3 million connections saved by United Airlines' ConnectionSaver (Jun 2025). Delta Air Lines CEO Ed Bastian said in Jul 2026 that AI's meaningful imprint may take another year or 2, and the only savings figure for Southwest Airlines, $6 million a year for Agentforce customer chat (Sep 2026), is a Salesforce projection.
The sector's largest cost programs mix AI with other levers and give no AI share: UPS credits its $3.5 billion of 2025 savings and 48,000 fewer operational positions to network reconfiguration and automation, American Airlines credits the nearly $1 billion of annual run-rate savings it expects from reengineering by the end of 2026 to technology, procurement and process work combined, United Airlines tied a 4% drop in management headcount to process changes and AI together (Oct 2025), and A.P. Moller - Maersk named AI in the productivity push behind closing about 1,000 corporate roles to save $180 million a year (Feb 2026) without saying how much comes from AI.
Professional services firms report AI use and hours, not returns: PwC counts more than 285,000 Copilot users and 54% of its global workforce using AI tools weekly (Apr 2026 figures), and McKinsey & Company more than 75% of employees using Lilli monthly (Jun 2025), which puts both firms at Piloting. McKinsey & Company said AI saved 1.5 million hours of search and synthesis work in 2025 (Jan 2026), and PwC counted more than 1 million hours of Copilot capacity (Apr 2026), both without a dollar value. Deloitte, at Experimenting, gives access figures such as Claude made available to more than 470,000 people (Oct 2025) but no firmwide count of employees using its AI tools.
Stages
Where manufacturers, airlines and logistics companies stand
Scaling
Scaling means the company reports returns it credits to AI in several functions, or a firm-level AI figure, and runs a company-wide AI platform with a named owner.
Piloting
Caterpillar, McKinsey & Company, PwC, Uber
Piloting means the company reports a business return it credits to AI in at least one use case, or a company-wide AI rollout with a user count and a named owner.
Experimenting
A.P. Moller - Maersk, American Airlines, Boeing, Deloitte, Delta Air Lines, GE Aerospace, Honeywell, John Deere, Siemens, Southwest Airlines, Union Pacific, United Airlines, UPS
Experimenting means the company has reported neither a business return it credits to AI nor a company-wide AI rollout with a user count. Pilots, partnerships, plans and spending alone keep a company at Experimenting.
Returns
The returns manufacturers, airlines and logistics companies credit to AI
Only realized returns the company reported and credited to AI itself, with their dates: revenue, cost, headcount, losses avoided or output per employee.
| Company | Use case | Return |
|---|---|---|
| FedEx | Predictive maintenance (MOBIUS)Network operations | Maintenance savings of $10 million or more a year, credited to proprietary AI models (Feb 2026)Source |
| FedEx | Surcharge capture on non-standard packagesPricing | Tens of millions of dollars a year in recovered surcharge revenue, credited to AI (Sep 2026) |
| Caterpillar | Condition monitoring prioritized service eventsServices and dealers | $1.1B closed-won sales in 2024 from leads its AI engine generates (Nov 2025) |
| Uber | AI coding agents (Software Factory)Engineering | Code output per engineer doubled on Uber's current measure, slowing engineering hiring (Aug 2026)Source |
Compare
Industrials side by side
| Company | Ownership | Platform | Models | Hosting | Reach | Stage |
|---|---|---|---|---|---|---|
| FedEx | Vishal Talwar, EVP and Chief Digital and Information Officer, on the Executive Committee | Hybrid: Atlas data platform built in-house; robotics and SaaS bought | Proprietary predictive models; generative AI provider not named | Cloud-first, mainframes largely retired; Azure behind Surround at its 2020 launch; Google Cloud for the Army project (2026); Atlas cloud not named | AI fluency program for about 300,000 employees (Feb 2026) | Scaling |
| Caterpillar | Group President Bob De Lange, over Chief Digital Officer Ogi Redzic and CTO Jaime Mineart; no dedicated AI chief | Built: Helios data platform and Cat AI Assistant, built in-house | Qwen3 4B and NVIDIA speech models in the in-cab demo; NVIDIA-based physical AI models; off-board models not disclosed | Helios mostly on AWS, with Snowflake for enterprise data and Microsoft as cloud partner; physical AI training on CoreWeave; in-cab models on NVIDIA Jetson Thor | Customers, dealers and technicians; employee use not disclosed | Piloting |
| McKinsey & Company | Kate Smaje, Global Leader of Technology & AI; QuantumBlack leads AI work for clients | Hybrid: Lilli built in-house on bought models; ChatGPT bought | OpenAI and Cohere inside Lilli; ChatGPT for work without client data | Multiple public clouds, none named, since its last data center closed in 2024; Lilli calls OpenAI and Cohere models, and CodeWall saw Amazon S3 storage paths (Mar 2026) | More than 75% of employees use Lilli monthly (Jun 2025) | Piloting |
| PwC | Global Chief AI Officer Joe Atkinson, with Chief AI Officers in PwC US and PwC UK | Hybrid: ChatPwC, agent OS and PwC One built in-house; Copilot, ChatGPT Enterprise and Claude bought | OpenAI, Anthropic and Microsoft Copilot, chosen by tool | ChatPwC and Microsoft 365 Copilot on Microsoft's cloud; agent OS deployable on Azure, AWS, Google Cloud, Oracle Cloud, Salesforce or on premises | 285,000+ Copilot users; 54% use AI weekly (Apr 2026 figures); 330,000+ trained in AI | Piloting |
| Uber | CTO Praveen Neppalli Naga, who leads engineering and science | Hybrid: Michelangelo and AI gateways built in-house; ChatGPT Work bought | OpenAI and Anthropic behind an AI Gateway, plus open-weight models | Co-located data centers, Oracle Cloud Infrastructure, Google Cloud and AWS (Trainium pilot); OpenAI and Anthropic APIs, personal data redacted first | Near 100% of engineers (Aug 2026); ChatGPT Work open to 30,000+ staff (Oct 2026) | Piloting |
| A.P. Moller - Maersk | CTIO Navneet Kapoor, owner of the technology and AI risk; CDO Krishnan Srinivasan | Hybrid: Microsoft Copilot bought; LLM apps built on Dify Enterprise | Not disclosed; Microsoft Copilot for staff, Azure OpenAI in job postings | Microsoft Azure for the data platform, Microsoft's cloud for Copilot, edge computers on vessels for Star Connect; Dify hosting not disclosed | About 25,000 active LLM app users (late 2025, per Dify) | Experimenting |
| American Airlines | Ganesh Jayaram, Chief Digital and Information Officer; GenAI center of excellence in IT | Hybrid: In-house tools on Microsoft Azure; some AI licensed from third parties | Not disclosed; third-party AI licensed (2025 Form 10-K) | Microsoft Azure as preferred cloud (May 2022), Smart Gating built on Azure; reservation system hosted by a third party; own data centers for AI not disclosed | Staff use not disclosed; generative AI trip planner on aa.com (Oct 2025) | Experimenting |
| Boeing | CIDO Dana Deasy, reporting to the CEO; chief AI and autonomy officer under the CTO | Hybrid: Boeing Conversational AI built in-house; Palantir Foundry and Azure bought | Not disclosed for the internal platform; LLMs trained on Boeing data; open models in prototypes | Internal AI inside Boeing's secure network (host not named; AWS, Google Cloud and Microsoft are its clouds since 2022); Dataliner on Microsoft Azure | 8,000 trained (Dec 2025); 22,000 active users at last count (Nov 2024) | Experimenting |
| Deloitte | Global AI leader Nitin Mittal; member-firm AI chiefs such as Deloitte UK's Chief AI Officer | Hybrid: Member-firm tools built in-house; Claude and Gemini Enterprise bought | Anthropic Claude, Google Gemini, OpenAI GPT via Azure, NVIDIA Nemotron | Deloitte's own NVIDIA AI factory (with HPE and Dell) for Zora AI and an internal assistant; Google Cloud, Azure OpenAI, HPE Private Cloud AI and Oracle Cloud; DARTbot and PairD in private environments | Claude available to 470,000+ people (Oct 2025) | Experimenting |
| Delta Air Lines | Amala Duggirala, Chief Digital & Technology Officer, reporting to the CEO | Hybrid: In-house tools on AWS; Fetcherr pricing and Airbus Skywise bought | Not disclosed; Amazon Bedrock used to test models (2023) | AWS as preferred cloud plus Delta's own data centers; vendor platforms for pricing (Fetcherr) and maintenance (Airbus Skywise) | All SkyMiles members for Delta Concierge (Aug 2026); staff use not disclosed | Experimenting |
| GE Aerospace | Chief Data & AI Officer Anton Groom (Aug 2026) and CIO David Burns; a cross-functional AI board | Hybrid: AI Wingmate built with Microsoft; Palantir AIP bought for supply chain | OpenAI GPT-4 via Azure OpenAI Service (2024), Palantir AIP, in-house vision and predictive models | Microsoft Azure (Azure OpenAI Service) inside GE Aerospace's virtual private cloud; GE Aerospace's secure cloud; Palantir AIP (host not disclosed); Waygate's InspectionWorks Insight cloud | Every employee has AI Wingmate (2025); last measured count 14,000 active (Nov 2024) | Experimenting |
| Honeywell | CTO Suresh Venkatarayalu sets AI strategy; CDO Peter Kropik runs the AI Center of Expertise | Hybrid: Forge built in-house on NVIDIA, Dell and Google Cloud technology | NVIDIA Nemotron and Cosmos, fine-tuned open-source models, Google Gemini under a 2024 deal, physics-based models, and LLMs from Azure, Google and AWS for internal use (Oct 2025) | Dell AI Factory servers with NVIDIA GPUs at customer sites and the edge, Google Cloud Vertex AI, Microsoft Azure for Forge apps (2020 deal), and Azure, Google and AWS model services for internal tools (Oct 2025) | All teams trained or being trained (Oct 2026) | Experimenting |
| John Deere | SVP and CTO Jahmy Hindman, who runs the central technology stack | Hybrid: Tech stack and Operations Center built in-house; JD's language model bought | In-house machine learning perception models; JD on an undisclosed frontier LLM | Cameras and GPUs on the machines for vision and autonomy; Operations Center data at the edge and in cloud environments John Deere does not name (AWS lists Deere as a customer); JD's model host undisclosed | ~1.2M connected machines, 450,000+ monthly digital users (Aug 2026); employee use not disclosed | Experimenting |
| Siemens | Peter Koerte, Managing Board member for Data & AI; EVP Vasi Philomin heads the Data & AI unit | Hybrid: SiemensGPT built in-house on AWS; coding and sales agents bought | Anthropic, OpenAI, Meta and others in SiemensGPT; Azure OpenAI in the Industrial Copilot | AWS for SiemensGPT and the Siemens Data Cloud; Microsoft Azure OpenAI for the Industrial Copilot; Salesforce for sales agents; own data centers not disclosed | 80,000 monthly SiemensGPT users (AWS case study, Oct 2026) | Experimenting |
| Southwest Airlines | Lauren Woods, EVP & CIO; enterprise AI led by VP Justin Bundick | Hybrid: AWS cloud and AI tools bought; AI solutions built by an in-house team | Not disclosed; AWS Kiro, Amazon Quick, Salesforce Agentforce and a plugin in OpenAI's ChatGPT | Largely Southwest's own on-premises systems plus AWS, its preferred cloud, with an all-AWS target in 2028; Agentforce runs in Salesforce's cloud and the ChatGPT plugin in OpenAI's | 2,700+ developers on Kiro (Jun 2026); staff use not disclosed | Experimenting |
| Union Pacific | EVP and CIO Rahul Jalali, reporting to the CEO, with business co-owners | Hybrid: UP Chat and AI planning built in-house; AI cranes acquired | Undisclosed large language models in UP Chat; in-house planning models | UP Chat runs on Union Pacific's secured data network behind its firewall; the cloud or data center hosting the models is not disclosed | UP Chat on company computers; user count not disclosed | Experimenting |
| United Airlines | Jason Birnbaum, CIO; a Responsible AI Executive Board sets AI governance | Hybrid: Mars ML platform and United Data Hub, built in-house on AWS | Through Amazon Bedrock; production LLMs not named | AWS, preferred cloud since 2021; 70-90% of workloads in the cloud (Apr 2024), the rest on premises | Customers through ConnectionSaver (3.3M+ connections saved, Jun 2025) and delay messages; 90%+ of managers on United GPT (Apr 2025, no newer count) | Experimenting |
| UPS | Bala Subramanian, EVP and Chief Digital and Technology Officer, reporting to the CEO | Hybrid: In-house tools (MeRA, Deal Manager, ORION) on bought models and cloud | OpenAI GPT-3.5 Turbo and GPT-4 behind MeRA (2024); others not disclosed | OpenAI models through Microsoft for MeRA; Google Cloud partner since 2019; primary IT operations in an owned New Jersey facility | Employee use not disclosed; AI assistants for customers in 20+ countries (Jun 2026) | Experimenting |
Questions
AI in industrials, answered
- Which manufacturers, airlines and logistics companies are furthest along with AI?
- In the Enterprise AI diffusion report, FedEx is at Scaling, Caterpillar, McKinsey & Company, PwC, and Uber are at Piloting, and A.P. Moller - Maersk, American Airlines, Boeing, Deloitte, Delta Air Lines, GE Aerospace, Honeywell, John Deere, Siemens, Southwest Airlines, Union Pacific, United Airlines, and UPS are at Experimenting. A stage follows from the returns and rollouts a company has publicly reported in the last 18 months, not from its spending or plans.
- How do airlines use AI?
- Airlines use AI in airport operations, customer service, pricing and software development. American Airlines assigns gates with Smart Gating, its in-house machine learning tool; Delta Air Lines runs Baggage AI in Atlanta, buys fare recommendations from Fetcherr and opened its Delta Concierge assistant to all SkyMiles members on Aug 10, 2026; United Airlines' ConnectionSaver had saved more than 3.3 million connections by Jun 2025; and Southwest Airlines has more than 2,700 developers on AWS's Kiro (Jun 2026). No airline has reported a business return it credits to AI since May 2025, and all 4 are at Experimenting.
- How do manufacturers use AI?
- Manufacturers use AI in their products, factories, dealer service and employee assistants. Caterpillar's AI engine builds service leads from 30 data sources in Helios and sends them to dealers, which produced $1.1 billion of closed-won sales in 2024 (reported Nov 2025), the only return a manufacturer in the sector credits to AI. GE Aerospace gives every employee AI Wingmate and cut shop turnaround by more than 5 days where its AI Material Assistant runs (Jan 2026), Siemens gives employees SiemensGPT on AWS, John Deere reports 50% to 60% herbicide savings for See & Spray customers (May 2026), and Honeywell builds AI into Forge, Experion and TrackWise. Caterpillar is at Piloting, and Boeing, GE Aerospace, Honeywell, John Deere and Siemens are at Experimenting.
- How do logistics companies use AI?
- Logistics companies use AI in network operations, pricing, customs, routing and trade compliance. FedEx predicts failures in sorting equipment with proprietary models, saving $10 million or more a year (Feb 2026), and recovers tens of millions of dollars of surcharge revenue a year with AI that catches non-standard packages (Sep 2026). UPS runs ORION for delivery routes and Next Gen Brokerage for customs, and says AI helped it absorb a more than 300% rise in customs brokerage filings in 2025 (Mar 2026). A.P. Moller - Maersk launched Trade & Tariff Studio (Jun 2025) and partnered with Altana on trade compliance (Apr 2026), and Union Pacific says AI fuel planning lowers costs, without a figure (Sep 2026). FedEx is at Scaling, and UPS, Maersk and Union Pacific are at Experimenting.