Alphabet AI commitments 2026 are being cited as evidence of a nearly $2 trillion race for hardware and memory. The filings support a narrower, messier conclusion: Alphabet reported about $811 billion in total purchase commitments by the second quarter of 2026, according to Tom’s Hardware’s reading of company disclosures. That is not an $811 billion memory order, nor is the four-company total a clean AI-hardware budget.
The difference matters because these figures are promises and obligations extending into the future, not a single year’s cash spending. They can cover chipmaking capacity, servers and other technical equipment, but also energy, buildings, contract manufacturing and leases. Treating the pile as a memory tab is the sort of accounting shortcut that produces a tidy headline and a bad comparison.
What does Alphabet’s $811 billion commitment figure include?
Tom’s Hardware reported that Alphabet’s total purchase commitments reached roughly $811 billion in Q2 2026, up from about $140 billion to $150 billion in Q3 2025. The report explicitly says the figure covers Alphabet’s total commitments, not memory alone. Its analyst-derived four-company tally put Alphabet, Microsoft, Meta and Amazon at about $1.968 trillion in combined commitments at that point.
On that same basis, Microsoft stood at approximately $678 billion, Meta at $349.3 billion and Amazon at $130 billion, Tom’s Hardware reported. Apple was listed at about $57 billion, including $56.2 billion due within 12 months. Those figures describe total obligations or commitments and should not be read as directly comparable allocations to AI chips or RAM.
There is no good basis in the available disclosures to declare Alphabet the leader in a like-for-like hardware contest. Companies categorize and report these obligations differently, and their reporting dates differ. A later Bloomberg report, published July 31, described almost $2.4 trillion in commitments among Alphabet, Meta, Microsoft and Amazon, including leases, buildings, energy and equipment.
That report said Alphabet had disclosed $902 billion in purchase commitments, contractual obligations and leases that had not begun, including technical equipment, energy and leases. It put Meta’s future spending at almost $700 billion, with about half tied to data-center leases not yet started, some payable over as long as 30 years. Bloomberg also noted that Meta includes some Reality Labs consumer-hardware spending, while Alphabet and Amazon identify content licenses among planned commitments.
Why AI still matters to these commitments
AI is plainly part of the backdrop. Compute requires processors, memory, storage, energy and the facilities that house them. A data center is where those pieces are assembled into a working service, with power and cooling constraints alongside the chips. But the disclosures do not isolate an AI-only share for the reported totals.
McKinsey projected in 2025 that worldwide data-center capital expenditure could reach $6.7 trillion through 2030, including $5.2 trillion for AI-capable facilities. It also estimated that AI workloads could account for about 70% of global data-center capacity demand by 2030. Those are forecasts of capital expenditure, a different measure from contractual and purchasing commitments, and McKinsey cautioned that major uncertainties remain.
The defensible takeaway is large and rising future obligations by the biggest cloud companies. The stronger claim, that they have committed nearly $2 trillion specifically to secure AI hardware and memory, is not established by the reported figures.
This story draws on original reporting from Tom's Hardware.
