Skip to content
Back to Issue 01

Capital flow drill-down

Money in, revenue out: the full breakdown.

Issue 01 · Week 17 of 2026.

The four-category capital scorecard expanded with named transactions, burn-to-revenue context, and per-category trends across recent issues. Each row corresponds to a row in the summary table on the issue page.

Capital flow scorecard

Capital in, revenue out, burn ratio, and movement.

Capital in, revenue out, and direction of travel for this issue.
CategoryCapital inRevenue outBurn to revenueMovement
Frontier LabsOpenAI, Anthropic, Google DeepMind, xAI~$50BPrior ~$30B · Up~$30BPrior ~$25B · Up0.6Anthropic +8.5 GW across NVIDIA, Trainium, TPU — forward-buy that locks in vendor diversity and outruns disclosed demand.
Hyperscaler-HostedAWS Bedrock, Azure Foundry, Vertex, OCI~$48BPrior ~$30B · Up~$10BPrior ~$8B · Up0.21Microsoft absorbs ~1.6 GW from Stargate Abilene — orphaned frontier capacity rerouted into hyperscaler platforms.
NeocloudsCoreWeave, Crusoe, Nebius, Applied Digital~$15BPrior ~$10B · Up$1.6BPrior $1.1B · Up0.4CoreWeave backlog +4x YoY to $66.8B (incl. Meta $21B six-year) — conversion velocity is now the metric, not gross backlog.
On-Prem / HybridEnterprise GPU clusters, sovereign and national programs~$42BPrior ~$25B · UpIndirectPrior Indirect · Flatn/aDeepSeek V4 open crosses frontier; sovereign programs surge — workload migration from rented to owned compute has a frontier-class trigger.

Category

Frontier Labs.

OpenAI, Anthropic, Google DeepMind, xAI

Capital in
~$50Bvs ~$30B · Up
Revenue out
~$30Bvs ~$25B · Up
Burn / rev
0.6Lower means more capital out than in.

The read

Frontier labs are the largest single source of forward compute commitments this week. Anthropic alone disclosed 8.5+ GW across NVIDIA, Trainium, and TPU — distributed across three platforms specifically to reduce single-vendor dependency. The other labs are following: OpenAI's Stargate restructure, Google DeepMind's TPU absorption, and xAI's Memphis expansion. The capital is committed faster than disclosed revenue can absorb it; that gap is the entire reckoning.

This week's transactions

Named transactions and source context.
DateTransactionAmountSource
Apr 7Anthropic-Google / Broadcom 3.5 GW TPU deal$100B / 10 yrTechCrunch, Broadcom 8-K
Apr 20Anthropic-AWS expansion to 5 GWundisclosedanthropic.com, AWS news
Apr 22Stargate Abilene capacity restructure passed back to Microsoft$21B / 1.6 GWDCD, GlobeNewswire

Trend across recent issues

W17
Capital in
$50B
Revenue out
$30B

Category

Hyperscaler-Hosted.

AWS Bedrock, Azure Foundry, Vertex, OCI

Capital in
~$48Bvs ~$30B · Up
Revenue out
~$10Bvs ~$8B · Up
Burn / rev
0.21Lower means more capital out than in.

The read

Hyperscaler capex deployment this week is dominated by Microsoft picking up Stargate's orphaned 1.6 GW capacity. Azure's deceleration on guidance is a wedge against the broader thesis but doesn't yet contradict the aggregate ~$700B 2026 capex trajectory. Q1 prints (Apr 29 - May 6) test whether the platform-pull-through narrative holds under disclosed numbers.

This week's transactions

Named transactions and source context.
DateTransactionAmountSource
Mar 6 / Mar 27Microsoft absorbs cancelled Stargate Abilene capacity$21B / 700 MW + 900 MWDCD, GlobeNewswire
Apr 25Azure guidance commentary heading into Q1 printn/aMicrosoft IR commentary

Trend across recent issues

W17
Capital in
$48B
Revenue out
$10B

Category

Neoclouds.

CoreWeave, Crusoe, Nebius, Applied Digital

Capital in
~$15Bvs ~$10B · Up
Revenue out
$1.6Bvs $1.1B · Up
Burn / rev
0.4Lower means more capital out than in.

The read

Neocloud backlog is up 4x year-over-year, but conversion velocity is the unresolved question. CoreWeave's $66.8B total backlog now includes the new Meta $21B six-year commitment through 2032; Crusoe-MS expansion adds another lever; Nebius and Applied Digital are tracking similar scaling. The capital-in / revenue-out spread is the widest in the stack and depends entirely on customers ramping into committed capacity.

This week's transactions

Named transactions and source context.
DateTransactionAmountSource
Q4 2025CoreWeave revenue backlog (cumulative)$66.8BCoreWeave Q4 25 press
Apr 20Meta-CoreWeave six-year through 2032$21BCoreWeave press
Apr 22Crusoe-MS expansion announcementundisclosedCrusoe press

Trend across recent issues

W17
Capital in
$15B
Revenue out
$1.6B

Category

On-Prem / Hybrid.

Enterprise GPU clusters, sovereign and national programs

Capital in
~$42Bvs ~$25B · Up
Revenue out
Indirectvs Indirect · Flat
Burn / rev
n/aLower means more capital out than in.

The read

Indirect spend through GPU shipments, sovereign programs, and enterprise on-prem deployments. DeepSeek V4 crossing the closed frontier on coding is the demand catalyst this week — sovereign programs aggregate to $80B+ of forward commitments across UAE, Germany, France, Mistral-Sweden, GMI Japan, IndiaAI, and the UK AI Growth Zones. The category has no consolidated revenue line; spend shows up in NVIDIA's data-center revenue, OEM hardware shipments, and government program disbursements.

This week's transactions

Named transactions and source context.
DateTransactionAmountSource
Apr 22-24DeepSeek V4 release (open weights, frontier-class on coding)n/a (catalyst)HuggingFace, deepseek.com
Q1 2026Sovereign AI commitments aggregate across 8 countries$80B+various government announcements
Apr 21UK AI Growth Zones aggregate$38.5BUK government

Trend across recent issues

W17
Capital in
$42B
Revenue out
n/a

Methodology

Capital-in and revenue-out figures are quarterly or annualized as noted in the row, sourced from public earnings disclosures, SEC filings, and lab and vendor announcements. Burn-to-Revenue is revenue divided by committed capital. The trend chart shows up to eight prior issues. Data is updated weekly; revisions in future issues do not retroactively edit this one.

Back to the full issue