the 10 largest 2026 rounds by yc-like alumni
two independent research runs (the agent on the web, weebo's bots) merged: kalshi $1b, whatnot $545m, supabase $500m at the top; only 5 of 10 rounds overlapped between the runs; 9 disagreements listed, none on round size.
largest 2026 funding rounds by accelerator alumni (merged finding)
window: rounds announced 1 jan – 4 sep 2026. eligibility: the company's founders went through y combinator or a named yc-like program (techstars, a16z speedrun, antler, entrepreneur first, south park commons, sequoia arc, pearx, neo, 500 global) — being an investor in the company is not enough. ranking is by primary equity round size; debt sleeves and credit facilities are noted but not counted. one row = one announced round, so a company can appear twice.
this page merges two independent research runs done on 4 sep 2026: "agent" (claude) and "weebo" (a grok-based bot). neither run found the full list on its own; five of the ten rows come from one run only.
merged top 10
notes on the ranking
- exa and starcloud's extension tie at $250m; either order is defensible. if you prefer one row per company, starcloud collapses to its $250m row and the three $150m rounds below move into a tie for #10.
- the techcrunch url for starcloud's august round says "200-million" in the slug; weebo reports the body says $250m and geekwire and the business wire syndication agree.
- supabase valuation: the company blog says $10b pre-money; pr newswire and cnbc say $10.5b post-money. same round, different framing.
just outside the top 10
| name | round (date) | program | found by | note |
|---|---|---|---|---|
| hightouch | $150m series d at $2.75b (27 apr 2026) | yc s19 | agent | https://www.businesswire.com/news/home/20260427054567/en/Hightouch-Raises-$150-Million-to-Reinvent-How-Marketing-Works-Using-AI |
| happyrobot | $150m series c at ~$1.2b (4 aug 2026) | yc s23 | weebo | https://tech.eu/2026/08/04/happyrobot-lands-150m-series-c-to-scale-agentic-ai-for-enterprise-operations/ |
| preply | $150m series d at $1.2b (21 jan 2026) | techstars berlin 2015 | agent | only non-yc alum near the cutoff; program participation verified by agent via techstars' own profile. https://pulse2.com/preply-150-million-series-d/ |
| emergent | $130m series c at $1.5b (16 jul 2026) | yc s24 | weebo | https://emergent.sh/news/emergent-now-a-unicorn-at-1-5-billion-valuation |
| deepgram | $130m series c at $1.3b (13 jan 2026) | yc w16 | agent | no url given in the agent's run |
rows flagged on the eligibility rules
- regent craft — $240m "series b" (27 aug 2026), yc w21, found by weebo. the company's own release describes it as $120m equity plus $120m debt from erebor bank. on the equity-only basis used here it is $120m and sits below the cutoff; on the headline basis weebo used it would be #9. program evidence is a yc linkedin post rather than the yc directory. https://www.regentcraft.com/news/regent-secures-240-million-series-b-funding-to-scale-seaglider-manufacturing-and-transform-maritime-mobility
- railway — $100m series b (22 jan 2026), found by weebo, program given as "techstars (alumni lists)"; participation not confirmed against a techstars directory or profile. below the cutoff anyway.
- baseten — $1.5b series f at $13b (22 jun 2026) and lovable — $400m series c at $13.3b (12 aug 2026). both found by agent and excluded: south park commons and antler respectively were early investors, and the agent found no evidence the founders went through a program. weebo did not surface either. baseten would be #1 and lovable would tie #5 if the rule were relaxed to "backed by".
- flock safety — ~$200m share sale at $8.4b (apr 2026), yc s17, found by agent. reported by the information only; not company-announced; treated as a single-source secondary sale and not ranked.
- afterquery — series b at $3.2b valuation (early sep 2026), yc w25. both runs exclude it because the round size is undisclosed.
- flutterwave — series e at $3.2b with ripple (16 jun 2026), yc s16, found by agent; size undisclosed.
- corgi's other 2026 rounds — $108m series a (jan; weebo labels it seed + series a combined), $106m series b1 at $2.6b (28 may), undisclosed series b2 at $4b (jul). all below the cutoff or undisclosed.
where the two runs disagree
| item | agent says | weebo says |
|---|---|---|
| whatnot revenue | "revenue not public; $8b gmv in 2025" (tubefilter/quartz/cnbc) | "on track to surpass $1b annual revenue in 2026" citing inc. (https://www.inc.com/jennifer-conrad/whatnot-just-clinched-a-20-billion-valuation/91386777) |
| supabase date and valuation | jun 4-5, $10.5b post-money | 4 jun; notes company blog says $10b pre-money vs cnbc/pr newswire $10.5b post-money |
| afterquery status | series b "closed" on sep 1-2, "confirmed by yc's gustaf alstromer" | "round not confirmed closed (forbes/tc 'reportedly')" |
| corgi yc directory url | https://www.ycombinator.com/companies/corgi | https://www.ycombinator.com/companies/corgi-insurance (weebo also notes techcrunch wrote "spring 2024" vs directory s24) |
| corgi jan 2026 round | "$108m series a" | "$108m seed + series a (combined)" |
| slash | "yc alum (batch not re-verified)", $100m series c "apr 2026 per alleywatch" | yc s21, $100m series c, 16 apr 2026 |
| starcloud | only the $170m march round; august extension absent | both rounds; ranks the $250m extension above the $170m round |
| ranking basis | primary equity only; debt and credit facilities excluded (astranis counted at $300m not $455m) | announced headline size (regent counted at $240m including $120m debt) |
| supabase arr | "not public" | "not public"; adds a sacra third-party estimate of ~$170m arr, marked estimate only |
no row has different round sizes between the two runs for the same round; the disagreements are on dates by a day, framing, and eligibility handling.
what one run found and the other missed
found by weebo, missed by agent: helion energy $465m (#4), starcloud $250m extension (#7 tie), regent craft $240m/$120m equity, happyrobot $150m, emergent $130m (and its $70m series b), railway $100m, plus the exclusion checks on ramp, cuspai, modal labs, cognition, instinct.
found by agent, missed by weebo: replit $400m (#5), astranis $300m (#6), exa $250m (#7 tie), hightouch $150m, preply $150m (the only non-yc program alum near the cutoff), deepgram $130m, avoca $125m+, zepto ~$115m pre-ipo placement, render $100m, flutterwave (undisclosed), flock safety (secondary), eight sleep $50m, plus the program-participation exclusions of baseten and lovable and the "no 2026 round" checks on rippling, deel, brex, gusto, faire, razorpay, scale ai, monzo, rappi, opensea, datarobot, talkdesk, airalo, cleo.
counts
- candidates checked: agent 45, weebo 42 (self-reported in each run).
- overlap of the two top 10s: 5 rounds (kalshi, whatnot, supabase, starcloud $170m, corgi). merged top 10: 5 rounds from both, 3 from agent only, 2 from weebo only.
- rows in the merged top 10 with two independent sources cited across the six files: 10 of 10 (astranis' $2.8b valuation is single-source; the round size is not).
- disagreements logged: 9, none on round size.
methods comparison
agent seeded from an internal accelerator cohort file, then swept techcrunch's mid-year unicorn roundup, alleywatch's monthly "largest us rounds" lists line by line, businesswire/prnewswire, and the yc company directory (a 404 there was treated as "not yc"). it was stricter on eligibility — it separated "went through the program" from "backed by" (baseten, lovable), verified preply's techstars class through techstars' own profile and 2015 press, and excluded debt — and it covered the most non-yc programs. it ran out of search budget before finishing (tandem, radar and slash's batch left unverified) and missed helion, the largest miss in either run, along with starcloud's august extension, happyrobot and emergent.
weebo worked from company newsrooms and press releases first, then techcrunch/cnbc/geekwire/tech.eu, then the yc directory, and documented each row's verification chain (announcement → independent press → directory). that chain found helion, regent, happyrobot and emergent. it missed replit, astranis, exa, hightouch and preply, covered non-yc programs only nominally (railway via "alumni lists"), ranked a half-debt round at its headline figure, and cited one source without a url (the business wire correction for starcloud's march round). its exclusion table (ramp, modal, cognition) is useful hygiene the agent did not do.
net: the runs are complementary rather than redundant. neither top 10 was right on its own; the merged list changes four of ten positions relative to either run.