Collection coverage — 17 September 2026
This snapshot contains 17,277 company–program records, not 17,277 distinct startups and not a complete census of investments. Companies may occur in multiple portfolios. Of these, 13,567 rows have investment/studio evidence within the selected investor groups, including firm-level and historical relationships; the count is not the number of confirmed accelerator admissions. Another 1,728 rows are broader 500 Global portfolio entries, kept for context.
All 19 entries from the post are accounted for below. SOSV is a parent rollup, so its count must not be added to HAX and IndieBio. Public names cannot recover undisclosed or stealth investments. Directory completeness means the visible source was exhausted on this date, not that the fund has disclosed its complete history.
| Program |
Records |
Relationship |
Coverage and limitations |
| Y Combinator |
6,225 |
Equity investor |
6,225 entries from the community-maintained yc-oss mirror; full mirror download, not independently verified against the official directory. Stealth and omitted companies are not covered. |
| a16z Speedrun |
278 |
Equity investor |
278 names; exhausted public API pagination. Homepage says over 300 startups funded, so this is not full investment history. |
| Techstars |
5,660 |
Equity investor |
5,660 directory records: 5,174 flagged accelerator companies; 486 other/network entries retained but excluded from investment matching. Public index fully paginated. |
| Founders, Inc. |
192 |
Equity investor |
192 names extracted from the official portfolio page; complete page snapshot, historical omissions unknown. |
| Sequoia Arc |
30 |
Equity investor |
30 explicitly named Arc participants from selected 2022–2023 cohort announcements and current testimonials. Incomplete; stealth companies omitted. Do not substitute all Sequoia investments. |
| South Park Commons |
202 |
Equity investor |
202 official portfolio entries; Founder Fellowship membership is not established for every company. |
| HF0 |
3 |
Equity investor |
Only two official residency alumni plus one company-confirmed HF0 investment; highly incomplete. |
| PearX |
39 |
Equity investor |
39 explicitly identified PearX/predecessor alumni from official announcements and showcases; incomplete historical cohort coverage. |
| 500 Global Flagship |
510 |
Equity investor |
510 records with GA/SF batch labels mapped to Flagship. Another 1,728 firm records remain unassigned/other. Official 2024 history says 900+ Flagship alumni, demonstrating incomplete tagging. |
| AI Grant |
92 |
Equity investor |
92 names across all four published batches; complete published batch lists, not necessarily all undisclosed investments. |
| AI Fund |
43 |
Venture studio / investor |
43 named public portfolio companies; five unnamed stealth themes excluded. Listing does not establish identical studio involvement in every company. |
| NVIDIA Inception |
6 |
Support, not equity investment |
6 explicit member spotlights only; no complete public member roster obtained. |
| Microsoft for Startups |
3 |
Support, not equity investment |
3 explicit program beneficiary examples; no complete public roster obtained. Generic Azure customer logos were not treated as confirmed program members. |
| Google for Startups Accelerator |
488 |
Support, not equity investment |
Regional pages across 10 geographies/themes; includes historical general accelerator alumni. Full global directory endpoint unavailable; AI-specific membership not confirmed for every record. |
| SOSV |
779 |
Equity investor / parent |
785 public directory listings consolidated to 779 company-program records after six duplicate aliases. Includes HAX, IndieBio and legacy/other programs. Do not add parent and children. |
| SOSV SF & NY (IndieBio) |
294 |
Equity investor / SOSV program |
294 directory entries with explicit historical IndieBio cohort labels. Current program naming and locations may differ from those labels. |
| SOSV HAX |
210 |
Equity investor / SOSV program |
210 company-program records after merging duplicate aliases; explicit HAX historical cohorts. |
| Greentown Labs |
731 |
Support / incubator membership |
731 startup member records, including alumni; complete exported member page snapshot, not full historical financing data. |
| Activate |
268 |
Non-dilutive fellowship |
268 companies in the fully expanded public directory. Includes US and Singapore entries; official aggregate historical totals exceed this directory. |
Provenance and collection
Official public API/directory extraction: Speedrun (all API pages), Techstars (all search-index pages), 500 Global (full returned public portfolio payload), Greentown (public rendered-page data), SPC (embedded directory data), Founders Inc. and AI Fund (official portfolio pages), AI Grant (all published batches). AI Fund company profiles supplied short product descriptions.
Browser extraction: SOSV (all 785 listings after repeated Load More; six duplicate aliases merged) and Activate (all 268 visible companies after full expansion). Individual SOSV profile requests returned 403, so the directory is the evidence; profile links are navigation references. Arc, PearX and HF0 use explicitly named primary announcements/testimonials rather than inferred membership from their parent-fund portfolios.
Google: ten regional/theme pages, preserving repeated source evidence when a company appears in more than one. The global alumni endpoint returned 503. Historical cards do not reliably identify AI-only cohorts. NVIDIA and Microsoft are explicit selected examples; no complete public member inventory was obtained. These are the largest remaining scope gaps alongside HF0, Arc, PearX and Flagship historical batch attribution.
YC: all 6,225 entries from the third-party yc-oss mirror, with official profile URLs. Three well-known company profiles (Airbnb, Stripe, Ginkgo Bioworks) were spot-checked against YC; this is not a representative accuracy audit or validation of all mirror records.
Snapshots and browser exports are saved under data/raw/. Every normalized record includes retrieval date, source URL, extraction method, scope, relationship, cohort when available, and accumulated evidence. data/deduplication_log.json records consolidation. Failed exploratory scripts are retained for audit, not advertised as a working refresh pipeline.
Rules that prevent inflated results
- Preserve stable source IDs where provided. Same company names can refer to different entities; do not merge YC entries just because their names match.
- Merge only the reviewed duplicate SOSV aliases within a program; retain all source evidence. No global claim of unique startup count is made.
- Exclude 486 Techstars non-accelerator/network records from investment matching. Their investment relationship is unverified, not proven absent.
- Keep 1,728 unassigned/other 500 Global companies outside the Flagship match corpus.
- Keep all 202 SPC rows marked as firm portfolio with Fellowship membership unverified.
- Mark credits, support, membership and fellowship separately from equity investment.
- Keep 238 same-domain cross-program overlap candidates for review; domain equality is not definitive company identity.
Not available consistently
Original investment amount/date, ownership, entry valuation, admission-time revenue, founder backgrounds, rejection decisions, return multiples and full follow-on history. A cohort or company founding year is not substituted for an investment date. Current descriptions may reflect pivots. Portfolio-status fields are source-reported, can be stale, and are not normalized into an outcome model.