Will Alphabet's depreciation expense exceed $35B in FY2026?

FY2025 depreciation was $21.1B on a CapEx base that includes prior years' spend plus FY2025's $91.4B. FY2026 adds $175-185B of new CapEx. Even at 6-year useful life, the incremental annual depreciation from FY2026 CapEx alone is ~$29-31B. But assets placed in service throughout FY2026 only generate partial-year depreciation — if evenly deployed, the average FY2026 impact is ~$15B incremental. Add the existing run rate (~$21.1B growing organically to ~$24-25B) and total FY2026 depreciation is approximately $36-40B. This exceeds $35B. The main risk to this estimate is further useful life extension or heavy back-loading of CapEx deployment. Setting at 72%.

The arithmetic supports exceeding $35B, but two factors could keep it below: (1) Management could extend useful lives from 6 to 7 years, reducing annual depreciation by $5-8B — they have precedent and incentive. (2) CapEx composition matters — if a significant portion goes to land and building construction (15-25 year useful lives), the depreciation ramp is slower. Data centers include both servers (3-6 years) and buildings (20+ years). If 40% of CapEx is building/land, the short-life depreciation increment shrinks. Setting at 65% — more likely above $35B than below, but meaningful uncertainty from useful life and asset mix.

Let me work backwards. For FY2026 depreciation to stay at or below $35B: (1) Existing asset base depreciation must not grow beyond ~$25B (requires no acceleration — plausible), (2) FY2026 CapEx incremental depreciation must stay below $10B (requires either heavy back-loading of CapEx or very long useful lives on new assets). $10B incremental means FY2026 CapEx generates only 5.5-5.7% first-year depreciation — below typical patterns for tech infrastructure. Possible if buildings dominate early CapEx, but unlikely given the Cloud infrastructure build imperative. Setting at 68% — the math favors exceeding $35B absent deliberate accounting management.

The math is straightforward. FY2025 depreciation at $21.1B growing organically at ~15-20% = ~$24-25B base. FY2026 CapEx of $180B (midpoint) at ~8% average first-year depreciation rate = ~$14B incremental. Total: ~$38-39B, well above $35B. Only accounting maneuvers (useful life extension, reclassification) keep it below. Setting at 70%.

I give more weight to management's ability to manage depreciation. Google already extended useful lives from 4 to 6 years in 2024 to manage the optics. They could do 6 to 7 years — perfectly defensible given infrastructure improvements. This alone could reduce total by $3-5B. Additionally, CapEx deployment is likely phased: more in H2 than H1 (construction timelines), reducing partial-year depreciation. The company has every incentive to keep depreciation looking manageable while the CapEx ramp continues. Setting at 62% — more likely above than below, but management has tools.

Even with a useful life extension to 7 years, the sheer scale of CapEx nearly doubling makes $35B hard to avoid. The existing PP&E base was already generating $21.1B at accelerating rates (+38% FY2025). The incremental FY2025 CapEx of $91.4B is still depreciating through FY2026 at higher run rates. Plus FY2026's $175-185B. Even aggressive accounting management struggles with these numbers. Setting at 73% — the scale overwhelms the accounting flexibility.

$21.1B base growing organically plus massive CapEx ramp. Math supports exceeding $35B. Main risk is useful life extension. Setting at 70%.

Arithmetic favors exceeding $35B but CapEx deployment timing and useful life management create uncertainty. Likely above $35B but not certain. Setting at 66%.

The $35B threshold is conservative given the CapEx trajectory. Existing depreciation at $21.1B and growing fast, plus massive new CapEx. Even with phased deployment and possible useful life extension, $35B+ is more likely than not. Setting at 74%.