What is Big Tech AI capex 2026? Big Tech AI capex 2026 is the combined capital expenditure spending by Microsoft, Meta Platforms, Alphabet, and Amazon on AI infrastructure (primarily data centers, GPUs, networking, and power) in calendar year 2026, which Fortune reported on April 30, 2026 will reach approximately $700 billion. That figure is roughly 60 percent higher than 2025 spending and is on pace to be the largest single year capex surge in technology history. For commercial real estate investors, this is the single most important demand signal for the data center sector. For comprehensive coverage of how AI is reshaping commercial real estate, see our complete guide on AI commercial real estate.
Key Takeaways
- Microsoft, Meta, Alphabet, and Amazon will spend roughly $700 billion on AI infrastructure in 2026, a 60 percent year over year increase.
- Microsoft raised 2026 capex to $190 billion on its April 29 earnings call, with $25 billion of the lift driven by higher component pricing.
- Meta lifted 2026 capex guidance to a $125 billion to $145 billion range, citing higher component costs and added data center capacity for future years.
- Alphabet now expects $180 billion to $190 billion in 2026 capex and signaled a significant further increase into 2027.
- Amazon leads the field at roughly $200 billion in 2026 capex, with Morgan Stanley modeling negative free cash flow near $17 billion this year.
- For CRE investors, this is the most important demand pillar for hyperscale data center leasing, powered land, and adjacent industrial assets through at least 2027.
Big Tech AI Capex 2026 Explained
Through April 30, 2026, all four hyperscalers have reported Q1 calendar year results and updated their full-year capex guidance. The numbers are extraordinary. Amazon expects roughly $200 billion in 2026 capex, with the majority earmarked for AWS data center expansion. Microsoft guided to $190 billion, including a $25 billion uplift tied to component price inflation. Alphabet now projects $180 billion to $190 billion, up $5 billion at both ends of its prior range. Meta raised its forecast to $125 billion to $145 billion. Add it up and you get approximately $700 billion in committed 2026 spending.
In context, that single year figure is larger than the entire annual gross domestic product of Switzerland. According to Goldman Sachs Research, consensus capex estimates have under-projected actual spend by more than 50 percent in each of the last two years. Goldman flags that AI companies may collectively invest more than $500 billion in 2026, while McKinsey estimates that global AI capex could reach $6.7 trillion by 2030 just to keep pace with compute demand.
Why This Matters for CRE Data Center Investors
Hyperscaler capex flows directly into commercial real estate in three primary ways. First, build to suit data center campuses, where developers like Digital Realty, Equinix, DataBank, and Applied Digital sign 10 to 15 year leases with investment grade tenants. Second, powered land and shell capacity, where land near substations and fiber routes is acquired and prepared for future data center development. Third, adjacent industrial demand for cooling equipment, switchgear, racks, and last mile logistics. CRE professionals tracking the sector should also see our coverage of the Anthropic-Amazon $25 billion Trainium deal, which committed five gigawatts of AWS Trainium capacity through 2027.
The capex surge has already driven record leasing activity. Digital Realty signed a 200 megawatt AI inference lease in Charlotte, North Carolina in Q1 2026, the largest single lease in company history. Applied Digital signed a $7.5 billion 15 year lease at its 430 megawatt Delta Forge 1 campus in Louisiana. CBRE reported 81 percent profit growth in Q1 2026, primarily driven by data center leasing and capital markets activity in the sector.
The Capex Composition: Short-Lived vs Long-Lived Assets
Microsoft CFO Amy Hood made an important distinction on the Q3 FY2026 earnings call. Roughly two thirds of Microsoft's $190 billion 2026 capex is going into short-lived assets, primarily GPUs and CPUs. The remaining one third goes into long-lived 15 year data center shell and infrastructure assets. The implication for CRE investors is that the 30 percent share of capex flowing into data center physical assets translates into a sustained, multi-year demand pillar for hyperscale and powered land deals.
- Long-lived shell and core infrastructure (roughly 30 percent of $700B): Approximately $210 billion in 2026 alone flowing into data center buildings, power infrastructure, cooling, and land. This is the addressable CRE market.
- Short-lived GPU and CPU spend (roughly 70 percent): Approximately $490 billion flowing into chips and servers, where Nvidia and Broadcom are primary beneficiaries. CRE exposure is indirect through the demand-pull effect on data center capacity.
Five CRE Investment Implications
- Hyperscale lease economics remain favorable. 15 year triple net leases with investment grade tenants at cap rates in the 6.0 to 6.5 percent range continue to clear, with embedded rent escalators. The 200 megawatt Charlotte lease at Digital Realty and the $7.5 billion Delta Forge 1 lease confirm that pricing power remains with landlords in supply constrained markets.
- Powered land repricing is accelerating. Landgate's April 2026 powered land analysis identified $13.1 trillion in repricing potential as rural farmland adjacent to substations gets reclassified as a data center input. CRE investors with land in Ohio, Indiana, Wisconsin, Texas, and the Carolinas should re-underwrite holdings against power availability rather than just zoning.
- Power and grid constraints become the binding constraint. Arizona Governor Katie Hobbs released a strategic energy plan in April 2026 with 31 recommendations and signaled an end to data center sales tax incentives. Northern Virginia, Phoenix, and Atlanta face multi year power interconnection queues. Sites with secured interconnection agreements command material premiums.
- Capital markets risk is real. Bank of America estimates Amazon will run a $28 billion free cash flow deficit in 2026. Meta's free cash flow is projected to drop nearly 90 percent. If a recession hits or AI revenue underdelivers, capex schedules can be pushed out, which would soften pre-leasing momentum on speculative builds.
- Concentration risk on the tenant side. A small number of tenants drive most new hyperscale leasing: Microsoft, Amazon, Alphabet, Meta, Oracle, and OpenAI. Diversification at the asset level is illusory if all your leases ultimately credit back to the same handful of names.
What the Bear Case Looks Like
The $700 billion question, as Fortune phrased it on April 30, is whether the pace of investment is calibrated correctly or whether Big Tech is building ahead of demand. Anthropic just hit a $30 billion annualized revenue run rate in April 2026 and OpenAI surpassed $25 billion. But OpenAI's CFO Sarah Friar reportedly raised internal concerns about paying future compute contracts, as covered in our analysis of the OpenAI compute crunch. If AI revenue plateaus while $700 billion per year of capex continues, the sector will face overcapacity and cap rate decompression on speculative builds. CRE investors looking for hands-on AI implementation support can reach out to Avi Hacker, J.D. at The AI Consulting Network for guidance on stress testing data center underwriting models against compute demand scenarios.
How CRE Investors Should Use AI Tools to Underwrite This Cycle
The data center sector is now too complex to underwrite with spreadsheets alone. Investors should use AI tools like Claude, ChatGPT, Gemini, and Perplexity to monitor hyperscaler earnings calls, parse 10-Q disclosures for capex commitments, track power interconnection queues by ISO region, and stress test pro formas against multiple AI demand scenarios. Industry research projects AI in real estate to reach a $1.3 trillion market by 2030 at a 33.9 percent CAGR. The 92 percent of corporate occupiers that have initiated AI programs are already operating in this paradigm. The 5 percent who report achieving most AI program goals are the firms that have integrated AI underwriting workflows into deal teams. For sector context, see CBRE's coverage of global real estate trends, which has consistently flagged hyperscaler data center demand as the dominant CRE growth driver. If you are ready to transform your underwriting process with AI, The AI Consulting Network specializes in exactly this.
Frequently Asked Questions
Q: How much will Big Tech spend on AI capex in 2026?
A: Microsoft, Meta, Alphabet, and Amazon are projected to spend approximately $700 billion combined on AI infrastructure in 2026, according to Fortune reporting on April 30, 2026. Individual guidance is roughly $200 billion at Amazon, $190 billion at Microsoft, $180 to $190 billion at Alphabet, and $125 to $145 billion at Meta.
Q: What share of Big Tech AI capex flows into data center real estate?
A: Microsoft CFO Amy Hood disclosed on the Q3 FY2026 call that roughly one third of capex goes into long-lived 15 year data center shell and infrastructure assets, with two thirds going into short-lived GPUs and CPUs. Applying that ratio to $700 billion implies roughly $210 billion of 2026 hyperscaler capex flowing directly into data center buildings, power infrastructure, and land.
Q: Why did Meta and Microsoft stocks fall after their earnings beats?
A: Both companies beat earnings expectations but raised 2026 capex guidance higher than investors expected. Meta lifted 2026 capex to $125 to $145 billion, and Microsoft increased its forecast to $190 billion, including a $25 billion impact from component pricing. Higher capex implies lower near-term free cash flow and higher future depreciation, compressing margins until AI revenue catches up.
Q: Should CRE investors still buy data center exposure at these levels?
A: The demand signal from $700 billion of 2026 capex is real, but pricing has already moved. Hyperscale data center cap rates compressed materially in 2025 and early 2026. Investors should focus on powered land, secondary markets with interconnection capacity, and value add conversions of legacy industrial sites near power substations rather than buying stabilized hyperscale assets at peak pricing. For personalized guidance on implementing these strategies, connect with The AI Consulting Network.
Q: What is the biggest risk to the Big Tech AI capex thesis?
A: The biggest risk is a mismatch between capex pace and AI revenue growth. Anthropic and OpenAI have both crossed $25 to $30 billion in annualized revenue, but those figures are still small relative to $700 billion in annual infrastructure spending. If enterprise AI adoption slows, the sector faces overcapacity, lease renegotiation pressure, and cap rate decompression on speculative builds. CRE investors should stress test underwriting models against scenarios where capex growth pauses for 12 to 24 months.