Microsoft, Google, and Meta's $100B AI Spending: Who Benefits?

When Microsoft, Google or Meta announces $100 billion in artificial intelligence spending, the money does not simply go to Nvidia. It flows through an enormous industrial supply chain involving semiconductor manufacturers, data-center developers, electricity providers and equipment suppliers.

Some companies receive payments as infrastructure is built, while others collect recurring revenue after facilities begin operating. Understanding that distinction helps explain who actually benefits from the AI investment boom.

The scale is unprecedented. According to the International Energy Agency's latest research, major technology companies are expected to spend more than $700 billion on capital expenditure in 2026, largely supporting data-center expansion.

Advanced computing hardware represents a major destination for AI investment.

Nvidia sells the GPUs and related systems used to train and operate AI models. AMD supplies competing accelerators, while Broadcom and Marvell help technology companies develop specialized chips.

However, designing a processor is not the same as manufacturing it. Taiwan Semiconductor Manufacturing Company (TSMC) produces advanced chips for major technology customers, while suppliers of high-bandwidth memory and semiconductor equipment support additional production stages.

The supply chain becomes even more complicated when companies develop custom AI chips. Google, Amazon and Microsoft increasingly design their own accelerators, reducing dependence on particular suppliers without eliminating manufacturing costs.

Microsoft offered an unusually clear breakdown in its fiscal third-quarter 2026 results.

Of its $31.9 billion in capital expenditure, approximately two-thirds was allocated to shorter-lived assets, primarily GPUs and CPUs. The remaining investment supported longer-lived infrastructure.

Once companies acquire computing hardware, they need somewhere to operate it.

Data-center construction generates business for engineering contractors, building-material suppliers, electrical-equipment manufacturers and infrastructure developers.

Companies such as Vertiv supply cooling and power-management equipment, while Schneider Electric and Eaton provide electrical-distribution technology. Arista Networks and other networking specialists connect enormous clusters of processors.

Electricity introduces another layer of spending.

AI facilities need grid connections, transformers and long-term power supplies. These requirements can create opportunities for utilities and equipment suppliers even when demand for additional GPUs begins slowing.

The distinction is important because electricity bottlenecks can delay otherwise completed projects.

An AI company might already own its servers but still need to pay contractors, utilities and infrastructure providers before putting them into service.

The economics change once the infrastructure becomes operational.

Electricity providers collect payments for power consumption, while maintenance contractors and equipment vendors can earn ongoing service revenue. Data-center landlords may receive rent under long-term agreements.

Cloud providers such as Microsoft Azure, Amazon Web Services and Google Cloud then attempt to recover their investments by selling computing capacity to customers.

AI developers and businesses using those services ultimately need enough revenue to justify the infrastructure costs.

This explains why the companies financing AI facilities are not necessarily the businesses receiving the largest initial payments.

Importantly, $100 billion in announced capital expenditure is not necessarily $100 billion immediately transferred to suppliers. Budgets can include equipment deliveries, construction projects and finance leases recognized at different times.