AI Funding, Acquisitions, and IPOs, Tracked Live

Scoopkit recorded 333 AI industry funding and deal events between 2026-07-22 and 2026-09-08. Venture rounds account for 159 of them (47.7%); acquisitions account for 63 of them (18.9%); IPOs or secondary sales account for 45 of them (13.5%). The companies appearing in the most deal events are Anthropic (25), Nvidia (24), and OpenAI (15). Every figure on this page comes from a single query against Scoopkit's own /v1/events endpoint filtered to FUNDING_AND_DEALS, not from a hand-kept spreadsheet. You can run the same query on the free tier.

Pulled 2026-09-08 from the live production API. Sample: all 333 FUNDING_AND_DEALS events in the archive, reported between 2026-07-22 and 2026-09-08. Method: one paginated query at 100 events per page, no manual curation, no editorial selection. This page refreshes automatically on a recurring basis rather than being published once.

Deal types over the window

Two of these are worth pausing on. GOVERNMENT_COMMITMENT is larger than most people expect from a category that reads like a footnote, and it captures sovereign and state-level AI spending that a generic funding tracker files under "other" or drops entirely. DOWN_ROUND_OR_DISTRESS is small, but it exists as its own subcategory precisely because those events are the ones a competitive-intelligence query most wants to find and a keyword search for "raises" will never surface.

Companies appearing in the most deal events

This counts any event where the company appears in the orgs array, so it mixes the money going out with the money coming in. Read it as "who is named in AI deal news," not "who raised the most."

Narrowing to events that name an investor gives a much thinner list: Nvidia at 5, Accel at 4, General Catalyst at 4, Index Ventures at 3, a16z at 3. The thinness is the honest finding here, because most deal coverage never names one.

What a merged deal event looks like

The most heavily merged event in the window is ev_b0a7e82781b0410b: Stripe to acquire AI model marketplace OpenRouter for over $7B. That one event merged 9 separate articles.

A raw news API hands you all 9 for that story: 9 rows, 9 near-identical headlines, and the job of working out that they describe one deal. Scoopkit returns one event with an article_count of 9. Across the whole funding set, 65 of 333 events (19.5%) merged two or more articles, which fits how funding news actually behaves. A big round gets picked up everywhere at once.

You can pull that exact event and see the merge for yourself with a free key from the quickstart.

What this data cannot tell you

Three real limits, stated plainly rather than buried.

Dollar totals are still not published, for a narrower reason than before. 190 of 333 funding events (57.1%) carry a money figure, now under 3 canonical field names rather than the 83 the classifier was inventing before the details field got a per-category contract (how that went). The naming problem is fixed; the coverage one is not. 42.9% of deal coverage never states an amount at all, so summing what is here would confidently understate the real total. A partial sum presented as a total is worse than no total, so this page still does not publish one.

Company names are not normalized. The same limitation the model-release tracker documents applies here. "a16z" and "Andreessen Horowitz" count as two separate investors in the tables above, because entity extraction reads them off the article text without a canonical alias map.

The window is short. The free tier's rolling 45-day archive is one funding cycle, not a trend. A quarter with one unusually large round shifts these percentages noticeably.

FAQ

Is there a free API for AI funding news?

Yes. Scoopkit's free tier returns FUNDING_AND_DEALS events at 100 requests per day with a 45-day archive, and it can be used in production. The paid details object, which holds deal-specific fields like round type and valuation, requires the Pro tier at $49/mo. Category and subcategory filtering, headline, summary, organizations, and article counts are all free.

How is this different from a news search for "AI funding"?

A keyword search returns articles, including every separate write-up of the same deal. This returns typed events, deduplicated, with the deal type already classified into one of 7 subcategories. The difference matters most when you are counting things, because counting articles double counts stories.

How often does this page update?

The underlying data updates continuously as the ingestion pipeline runs, and this page is regenerated automatically from a fresh query rather than hand-edited. The numbers above reflect the window that was current on 2026-09-08. If you need live figures, query the API directly instead of scraping this page.

Does it cover deals outside the US?

Yes, though coverage follows English-language source availability. GOVERNMENT_COMMITMENT events in particular skew toward sovereign AI programs that got English coverage.