AI Capex Flow MapHyperscaler → Supplier Attribution
All suppliers
Compute SiliconSEC filings

Broadcom (AVGO)

Custom AI ASICs (Google TPU, Meta MTIA), Tomahawk networking switches

Modeled AI capex flow
$39.0B

from Big 5 forward AI-attributable spend

Hyperscaler exposure
40.0%

of revenue (latest 10-K)

Top hyperscaler customer
GOOGL

modeled $17.9B flow

AI revenue run-rate
$22.0B

Q4 FY25 earnings (Dec 2025)

Forward visibility: AI revenue run-rate

The leading indicator AM analysts read as capex commitments convert to supplier revenue.

Management-disclosed AI semiconductor revenue annualized; not a contractual backlog. Track against named 3 + 'pipeline' commentary.

As of: Q4 FY25 earnings (Dec 2025)Source filing
Flow attribution by hyperscaler

Modeled dollars from each hyperscaler's AI-attributable forward capex.

CustomerModeled flow% of total
Alphabet (GOOGL)$17.9B45.8%
Microsoft (MSFT)$7.9B20.2%
Meta Platforms (META)$7.6B19.4%
Amazon (AMZN)$4.9B12.6%
Oracle (ORCL)$0.8B1.9%
Computed from disclosed customer concentration and per-hyperscaler accelerator preferences. Per-edge values can be reconstructed from the Methodology page.
Notes & source

Three named AI ASIC customers as of FY25 (Google, Meta, ByteDance) plus pipeline. AI semiconductor revenue ~$15B FY25 trending toward $25B+ FY26 per company commentary.

View SEC filings page