AI

Caterpillar Became an AI Company

CAT's Q1 2026 earnings showed AI data centres drove 41% power generation growth. Here's how this became one of the biggest AI infrastructure plays on earth.

30 April 2026·8 min read·AI & Tech
Caterpillar Became an AI Company

On 30 April 2026, Caterpillar reported Q1 2026 results that stopped analysts cold. Revenue came in at $17.42 billion, up 22% year on year. Its Power and Energy segment posted $7.03 billion in sales, also up 22%. Power generation revenue within that segment surged 41%. The company's order backlog reached $62.7 billion, up $27.7 billion versus the same quarter a year earlier. Customers are placing orders for Caterpillar equipment extending well into 2028. For a breakdown of CAT’s core business and SA mining exposure, see our full Caterpillar stock profile.

The driver for most AI data centres.

Data centre operators are not waiting for grid connections that take years to permit, approve, fund, and build. They are buying Caterpillar natural gas engines and generator sets to build on-site power plants and generate electricity immediately. One developer described having 600 Caterpillar natural gas reciprocating engines on order for a single 4,000-acre campus in Utah that will not connect to the grid at all. The AI infrastructure build-out has turned Caterpillar, a company most people associate with yellow bulldozers and mine trucks, into one of the clearest AI beneficiaries on any major exchange.

CEO Joe Creed confirmed on the earnings call that the backlog is record-level and that customers are committing to longer-term orders than Caterpillar has historically seen. The company is projecting low double-digit revenue growth for full-year 2026, with Power and Energy leading. Caterpillar raised its adjusted operating margin guidance above its previous forecast. Google and Oracle data-centre deals were specifically highlighted. Caterpillar shares are up more than 160% in the last year and 41% in 2026.

What this reveals about AI infrastructure economics

The Caterpillar story illustrates a principle that is easy to miss when AI discourse focuses entirely on model benchmarks and chip specifications. Every inference request that returns a result consumed electricity. Every AI training run consumed significant electricity. Every data centre housing the servers that run these models requires power infrastructure that cannot come from a grid constrained by permitting timelines and transmission capacity limits.

Caterpillar's fastest-growing segment is not construction equipment or mining machinery. It is power generation for AI compute infrastructure. The company that built its reputation digging holes and moving earth has become critical infrastructure for the intelligence economy.

This is what the picks-and-shovels framing looks like in practice. The companies that supply the physical infrastructure supporting AI do not need to make a single AI model or write a single line of code to capture enormous value from the AI buildout.

The SA angle: what is happening in our mining sector

Anglo American has been deploying digital twin technology, predictive analytics, and AI-driven water management across its operations under its FutureSmart Mining programme. In 2022, Anglo American unveiled the world's largest hydrogen-powered mine haul truck at its Mogalakwena platinum mine in Limpopo Province.

Anglo American Platinum was demerged from Anglo American as an independent company in May 2025 and rebranded as Valterra Platinum, with a primary JSE listing and a secondary London listing. The platinum AI and automation story now belongs to a separately listed entity.

Kumba Iron Ore, majority-owned by Anglo American, has been expanding semi-autonomous and fully autonomous truck operations at its Sishen mine in the Northern Cape. Autonomous haul trucks operate 24 hours a day, require no shift changes, and eliminate certain categories of human error risk in open-pit environments. Kumba trades on the JSE under the ticker KIO.

Glencore operates significant coal assets in South Africa through its local subsidiary and is deploying predictive maintenance and sensor-based monitoring across its operations. Gold Fields has been deploying machine learning tools to improve ore body modelling at its South Deep mine in Gauteng. The global AI in mining market was valued at $6.39 billion in 2025 and is projected to reach $41.13 billion by 2035, growing at a compound annual rate of 20.6%.

The bigger investment question: which JSE-listed company benefits most from AI infrastructure?

Construction materials is a credible angle. Afrimat, which supplies aggregates and industrial minerals, is a direct input supplier to any significant infrastructure build in South Africa. PPC and Sephaku Cement are concrete inputs for any data centre foundation or building. If the SA data centre market grows from $2.16 billion in 2024 to $3.40 billion by 2030 as projected, the physical construction alone generates demand for local materials suppliers.

Logistics is more indirect but real. The large listed trucking and freight businesses all benefit from any increase in large-scale construction and infrastructure activity.

However, the more direct JSE play on global AI infrastructure is probably not construction at all. It is power. Eskom is not listed. But the companies supplying renewable energy into the SA grid as independent power producers, the cable and transformer manufacturers supporting grid upgrades, and the industrial equipment companies servicing both mining and energy customers carry exposure to the same underlying theme that has made Caterpillar so valuable: AI needs power, and building that power infrastructure requires heavy capital goods.

Wilson Bayly Holmes-Ovcon (WBHO) and Aveng are the listed construction businesses with the most direct exposure to large capital projects. They carry significant execution risk but also stand to benefit if the SA data centre pipeline accelerates.

Investor Takeaway: The Caterpillar principle transfers directly to the JSE context. AI infrastructure builds demand for power equipment, construction materials, logistics services, and specialised engineering. On the JSE, watch the listed construction and industrial services sector alongside the listed materials suppliers for exposure to SA data centre investment. The picks-and-shovels principle is as valid in Johannesburg as it is in Dallas.