OpenAI’s M&A Machine: Six Acquisitions and Counting

OpenAI killed Sora on Tuesday. The same day, it tacked $10 billion onto its record $110 billion raise. That’s the week in miniature: the company that just raised more private capital than any startup in history decided a video app burning $1M a day wasn’t worth the compute.

OpenAI shut down Sora, six months after launching the standalone video app to mass fanfare and an App Store chart-topper. The app goes dark on April 26; the API follows on September 24. According to a WSJ investigation, users peaked at one million and collapsed to under 500,000, while the app burned roughly $1 million in compute daily. Sam Altman made the call: kill it, free the chips, refocus on enterprise. The casualty was a $1 billion Disney deal — Disney found out less than an hour before the public announcement. OpenAI is targeting $10 billion in annualized revenue before a likely IPO, and video generation doesn’t get you there. Consumer experiments that look like a capability flex don’t impress the public market or investors. Enterprise contracts do. Read more…

Shield AI raised $2 billion at a $12.7 billion valuation — $1.5 billion in Series G led by Advent International and JPMorgan Chase, plus $500 million in preferred equity from Blackstone. The San Diego company makes autonomous AI systems for military aircraft and now plans to acquire Aechelon Technology, a defense software company that trains pilots and tests autonomous systems. JPMorgan and Blackstone writing checks into a Series G defense startup signals something: this is no longer a niche bet. Defense tech is now institutional capital’s territory. At $12.7 billion, the bet is that government contracts are more durable than commercial SaaS revenue — and in the current geopolitical climate, it’s hard to argue otherwise. Read more…

Harvey raised $200 million at an $11 billion valuation, led by GIC and Sequoia. That’s the fourth fundraise in 14 months, and the valuation jumped $3 billion since the last round. Revenue is reportedly $190 million annualized — meaning investors are pricing this at roughly 58x ARR. The bear case writes itself. But the bull case is more interesting: Harvey has locked up most of the 100 largest U.S. law firms, 500+ in-house legal teams, and 50 asset management firms across 60 countries. That’s not a feature getting killed by the next model update. That’s distribution. Sequoia has backed this company three times. When a firm writes a third check at a valuation 15x higher than the first, they’re not hedging. They’re declaring a category winner. Read more…

OpenAI has already made six acquisitions in 2026. All of last year, it made eight. At this pace, it’ll double that by December.

The two most recent deals — Astral (open-source developer tools) and Promptfoo (AI application testing), both announced in March — tell you exactly what OpenAI is buying and why. These aren’t product acquisitions. They’re capability grabs dressed in open-source credibility. Astral built the Ruff linter and uv package manager, tools used by millions of Python developers daily. Promptfoo lets teams test and evaluate AI applications before they ship. Neither is a consumer product. Both are things software engineers love and use without thinking about it.

That’s the thesis. OpenAI wants to become the default infrastructure layer for the developer ecosystem — not just the model provider, but the toolchain. Every open-source tool it absorbs is a distribution channel. Every developer who uses an OpenAI-owned tool before they know it’s OpenAI-owned is a potential enterprise customer later.

The M&A pace makes more sense in context. OpenAI closed $110 billion in February at an $840 billion post-money valuation. It’s reportedly targeting $10 billion in annualized revenue ahead of a potential IPO. That’s roughly a 70x revenue multiple that needs to be justified to public market investors who will demand more than model performance benchmarks. They’ll want moats. Distribution. Lock-in. Open-source developer tooling is a moat.

But the pace also creates risk. OpenAI has now acquired 17 companies in three years, with the acceleration happening almost entirely in the last 18 months. Integration is hard. Culture is hard. The Windsurf acquisition — a $3 billion deal that collapsed last year — is a reminder that not all of this goes cleanly. And according to a Fortune/HSBC analysis, even with $120 billion in the bank, OpenAI’s cumulative free cash flow remains deeply negative through 2030, with a projected $207 billion funding shortfall still ahead.

The Sora shutdown is part of this story. OpenAI is rationalizing. Kill the expensive consumer experiments. Focus compute on revenue-generating products. Acquire the tools developers already rely on. Build the enterprise distribution that justifies the valuation. It’s not the behavior of a research lab anymore. It’s the behavior of a company preparing to go public — and doing so with urgency.

The question is whether the M&A machine can integrate fast enough to create real compound leverage, or whether it’s accumulating complexity at the same pace it’s accumulating portfolio companies. The answer will matter a lot when the S-1 lands.

Harvey raised $200 million Series E at an $11 billion valuation, co-led by GIC and Sequoia. Founded in 2022 by a former lawyer and a Google DeepMind researcher, Harvey builds AI agents that handle legal work — contracts, due diligence, compliance, litigation prep — for law firms and in-house teams. It has locked up most of the top 100 U.S. law firms and 500+ corporate legal departments. The company is now deploying embedded engineering teams directly inside client organizations, a services layer that deepens switching costs far beyond standard SaaS. At 58x ARR, this is a category-creation bet, not a revenue multiple. Read more…

eMed raised $200 million Series A at a $2 billion-plus valuation, led by Aon Consulting. eMed is a Miami-based digital health platform that manages GLP-1 weight-loss programs end-to-end for employers — covering remote diagnostics, clinician-guided prescribing, and ongoing adherence coaching. GLP-1 medications are the most requested workplace benefit today, yet only one in five companies provides coverage. eMed is betting that employer willingness to pay for this is structural, not cyclical; it reports 90%+ member adherence, more than double the industry norm. Tom Brady is both investor and chief wellness officer — make of that what you will. Read more…

Xona closed a $170 million Series C led by Mohari Ventures Natural Capital. Xona is building Pulsar, a commercial navigation satellite constellation in low Earth orbit designed to replace — or at minimum significantly augment — GPS. Its satellites fly 20x closer to Earth than GPS and broadcast signals 100x stronger, providing centimeter-level accuracy that works indoors, under tree cover, and in GPS-denied or jammed environments. With geopolitical GPS spoofing on the rise and autonomous systems demanding precision GPS can’t reliably deliver, the pitch has gone from interesting to urgent. The company plans to deploy 258 satellites; the first production-class satellite launched last June. Read more…

Dash0 closed $110 million Series B led by Balderton Capital. Founded in 2023 by the team behind Instana (acquired by IBM), Dash0 is an observability platform built natively on OpenTelemetry — the open standard for collecting logs, traces, and metrics from cloud systems. What sets it apart from incumbents like Datadog is its agentic layer: AI agents that don’t just surface problems but autonomously resolve them, creating dashboards, validating deployments, optimizing costs, and detecting security anomalies without waiting for a human to act. Six hundred paying customers in under two years, including Zalando and Taco Bell. Valued at $1 billion. Read more…

Steno raised $49 million Series C led by Savano Capital Partners. Founded in 2018 and based in Los Angeles, Steno is a tech-enabled court reporting and litigation support company that provides deposition services to law firms while layering in AI tools built directly from its access to real case data. Its flagship AI product, Transcript Genius, lets attorneys semantically search, query, and summarize deposition transcripts in minutes rather than hours. Thousands of firms use it monthly. The hybrid model — service provider plus software company — gives it training data and distribution that pure SaaS legal AI competitors can’t replicate.

Thesis Care closed $45 million Series A led by Oak HC/FT. Thesis Care is a New York-based platform that uses AI to increase clinical capacity — specifically, enabling care teams to manage more patients without proportionally increasing headcount. The company targets the staffing crisis hitting healthcare providers, where nursing shortages and rising patient volumes are forcing hospitals and health systems to find ways to do more with less. Oak HC/FT is one of the more operationally rigorous healthcare-focused VCs; when they lead a Series A in this space, it’s generally because the unit economics have been stress-tested.

The Sora shutdown story isn’t over. The WSJ investigation dropped Sunday revealing the $1M/day burn and the Disney blindside — and the implications for OpenAI’s product strategy will keep getting picked apart. Watch for how OpenAI frames the narrative heading into IPO preparation. “Disciplined capital allocation” is the preferred framing; “product graveyard” is the less flattering version. How Wall Street analysts process this will matter when the S-1 lands.

On the fund side, whispers are circulating that General Catalyst is moving toward a close on its next fund, reportedly targeting $10 billion — which would make it one of the largest VC fund closes in history. No announcement yet, but LP conversations are reportedly in late stages. Expect something in the next two to three weeks.

In defense tech, sources suggest at least two more autonomous systems companies are in active raise processes at $500 million or above. Shield AI’s $2 billion round validated the asset class for institutional capital in a way that smaller checks hadn’t. If you’re building AI for the military and haven’t started your Series C conversations, you’re late.

Finally, watch Harvey’s competitors. An $11 billion valuation for a four-year-old legal AI company is a green light for every law firm technology vendor to reprice. Expect Ironclad, Relativity, and several European legal AI startups to announce raises in the next 30 days using Harvey’s round as the comparable.