Verifiable Compute — the sector that matters most and gets talked about least

The single sentence

If you read one sentence about verifiable compute as a venture sector this year, read this one: median 14-day commit-velocity acceleration across the top 25 verifiable-compute orgs is 2.4× the cross-sector median for the trailing 11 of 12 weeks, with contributor influx running 1.7× the cross-sector median over the same window — yet aggregate venture mindshare for the sector (measured by partner posts on LinkedIn, podcast minutes, and pitch-deck mentions captured by the panel) sits at 18% of comparable AI-native devtools mindshare.

Every other line in this drop is a footnote to that sentence.

Why this sector, why this month

The Stadium Pitch this month named verifiable compute as the most consequential structural shift of 2026. This drop is the operationalisation of that claim. Where the Stadium Pitch frames the thesis, the drop names the 25 orgs and the three breakouts that turn the thesis into a watchlist.

The methodology is the same engine that ships the Acceleration Watch every Sunday: 14-day commit velocity vs the org's own trailing 90-day baseline, two-period confirmation to filter spike noise, contributor-quality filter (no bot accounts, no green-square farming, no batch-imported translations).

What's different this month: we're publishing the full top 25 with anchored 14-day deltas, the contributor-influx maps for the top 5, and the three pre-Crunchbase breakouts as a single PDF + CSV bundle that sits in your Dropbox until a partner asks you the verifiable-compute question.

The top 25 — what's in the bundle

The PDF lists each org with: name, GitHub handle, sector sub-classification (zero-knowledge proof systems, attestation infrastructure, confidential compute, replicated state, cryptographic accumulators, or hybrid), 14-day acceleration percentile, 30-day contributor delta, stage inference (Pre-Seed / Seed / A / B inferred from team size + investor signal where available), and a single-line engineering thesis written by the founder.

The CSV gives you the same fields plus the raw weekly commit counts for the trailing 12 weeks so you can re-run any contributor-quality filter you prefer in your own notebook.

Three orgs in the top 10 are what we call 'velocity-without-mindshare' — ranked top-decile on the engine, mentioned by zero of the 47 sector-relevant LinkedIn-active partners we tracked last month. Those are the rows worth a Tuesday-morning cold email.

The three pre-Crunchbase breakouts

Each breakout comes with: GitHub org, the engineering signal that surfaced it (contributor influx, infra-repo creation pattern, or commit-velocity regime change), a 1-paragraph thesis, and 5 diligence questions worth asking the founder if you take the meeting.

Breakout #1 is a four-person team that spun up three new attestation-related repos in the trailing 6 weeks while contributor count went 2 → 9. Breakout #2 is a single-founder ZK-prover acceleration project whose 14-day commit velocity went 4× over a clean baseline window. Breakout #3 is a confidential-compute toolchain whose maintainer base shifted from one company's engineers to a six-company contributor mix in the past 90 days — the open-source-as-go-to-market pattern.

Names + handles are in the PDF. We don't surface them in the public abstract because the lead-time premium is 21–47 days and we're protecting your read.

What the data does NOT say

Every drop carries this section. The honest read of the calibration boundary keeps the methodology trustable.

The verifiable-compute panel is acceleration-positive but sample-thin — n=47 active orgs vs n=219 across the full backtest. Confidence on the 21–47-day lead band is wider for this sector than for AI-native devtools or developer-platforms (where n is 3–5× larger).

Two of the top 5 ranked orgs are research-stage and may never raise a priced round (academic spinouts that absorb back into university IP). The PDF flags which orgs carry that risk.

We don't yet have a clean signal for what we'd call 'infrastructure substrate' projects (e.g. EigenLayer-style restaking) where the velocity is in extension surface, not core repo. That's a methodology gap we're explicitly working on for the July methodology drop.

How members use this

Three patterns we expect to see across the Insider cohort over the next 30 days:

(1) Sector-Sweep cross-reference. Insiders who already bought the verifiable-compute Sector Sweep last quarter use this drop as the velocity update on the same panel. The Sweep gave you the deep diligence; this gives you the next 60-day velocity read on the same orgs.

(2) Cold-outreach calendar. The three breakouts are calendar-able — block 30 minutes Tuesday morning, draft the cold email per breakout, send before the close of week. Lead-time math says you have a 21–47 day window before the deck circulates.

(3) Partner-meeting brief. The single-sentence framing + the top-25 chart are partner-meeting material. Drop into a slide, attribute to 'The Data Nerd panel n=47, May 2026', cite signals.gitdealflow.com/continuity/2026-05-verifiable-compute. We've already done the hard part of the calibration — your job is to translate to the room.

What members get

The full 25-page PDF, the raw CSV, the contributor-influx maps for the top 5, and the three breakout briefs are gated to paid members behind /api/v1/insider/drops/2026-05-verifiable-compute. Use your Insider API key. Endpoint returns a signed download URL valid for 24 hours.

If you're on the Free Acceleration Watch, you can read this abstract and the public essay above (which goes deeper than what made it into the public Sunday digest). The artefact bundle requires Insider Circle (€97/mo) or higher.

If you upgrade today, you get retroactive access to every prior drop — though as the inaugural drop, that's just this one. Upgrade compounds from here.