There’s been some epic M&A in the past 12 months.
Nvidia paid $20 billion for Groq. Google closed its $32 billion acquisition of Wiz. Google paid $2.4 billion just to hire Windsurf’s CEO and license their tech. Palo Alto Networks is buying CyberArk for $25 billion. ServiceNow dropped $7.75 billion on Armis—and $20 billion total on M&A in the past 12 months.
The headlines make it feel like the M&A market is roaring back. And in a sense, it is. Really Big Deals in AI are back.
But here’s what the headlines miss: the deals you actually hear the numbers on? Those are increasingly rare. And when we do see the numbers, far fewer are returning real money to investors.
Per SVB’s latest State of the Markets report, there are now 8 M&A deals for every 100 VC deals—the highest ratio in seven years. Back in 2002, it was 1 M&A deal for every 4 VC deals. By 2024, that had cratered to 1 in 20. Now it’s climbing again.
So yes, more deals are getting done.
But of the deals where we actually know what happened, only 7% returned 3x or more to VCs in 2025. That’s the lowest … SVB has on record.
That’s down from 22% in 2021. Down from 15% in 2019. The share of confirmed good outcomes has collapsed by two-thirds.
But 90% of deals in 2025 didn’t disclose terms at all. So while the trend is real, just understand we only have a slice of overall M&A outcomes in this data.
What the Data Actually Shows (And What It Doesn’t)
Let me be precise about what we’re looking at. SVB breaks down M&A outcomes into four buckets based on deal price relative to total VC raised:
Two things jump out:
First, 90% of 2025 deals are undisclosed. That’s up from 69% in 2021. We simply don’t know what happened in nine out of ten deals.
Second, among deals that did disclose, confirmed “good” outcomes have dropped from 22% to 7%. That’s the signal we can actually measure.
Now, is it fair to assume undisclosed deals skew worse? Probably. Nobody hides a win. When Google pays $32 billion for Wiz, they announce it. When a startup gets acquihired for $20M after raising $50M, everyone stays quiet.
But we can’t say “only 7% of M&A deals return real money.” We can say: only 7% of deals confirmed returning 3x+, and the vast majority won’t tell us what happened. Draw your own conclusions about what that silence means.
The Trend Is Clear Even If the Absolute Numbers Aren’t
Here’s what we can say with confidence:
In 2021, roughly 1 in 5 deals confirmed a 3x+ return. Today it’s 1 in 14.
In 2021, 31% of deals disclosed terms. Today it’s 10%.
Whether the undisclosed deals in 2025 are mostly acquihires, soft landings, or just private wins we didn’t hear about—the confirmed good outcomes are way, way down.
That’s not statistical noise. That’s a structural shift.
The Headline Deals: A Different Story
Let’s be clear: there have been some monster deals this year. If you’re one of the lucky ones, M&A in 2025 has been spectacular:
These are real outcomes. Real liquidity. Real wins for investors.
But these deals represent a tiny fraction of total M&A activity. For every Wiz, there are dozens of quiet acquisitions where the founders got jobs and the investors got their money back—maybe. We just don’t hear about them.
The Acquirers Have Changed Too
Here’s what’s actually interesting: 46% of M&A deals in 2025 have a VC-backed buyer.
That’s up from just 10% in 2015. The unicorns are eating each other.
Look at who’s been acquiring since 2023:
Databricks has done 17 acquisitions. GrubMarket has done 17. These aren’t one-off strategic deals—these are systematic roll-up strategies.
When your acquirer is another VC-backed company, not a cash-rich public strategic, the math is different. They’re paying in equity often (at least in part), structured deals, and earnouts. They’re not always writing $500M checks from the balance sheet.
Getting acquired by Databricks or Stripe can end up a big win if they go on to double, triple, 10x in value. But it’s not the same in many cases as being cashed out or getting liquid stock. And those deals are far less likely to be disclosed publicly.
The Acquihire Problem
There’s another trend buried in here: acquihires are replacing real acquisitions.
SVB calls it out directly: “As the frenzy to acquire AI talent heats up, more companies are circumventing conventional M&A through acquihires that bolster their in-house development stack.”
The poster child? Google’s $2.4 billion deal for Windsurf.
Let’s break down what actually happened:
- OpenAI had agreed to buy Windsurf for $3 billion
- The exclusivity period expired on July 11, 2025
- Within hours, Google swooped in with a $2.4 billion “licensing and hiring” deal
- Google got CEO Varun Mohan, co-founder Douglas Chen, and ~40 senior engineers
- The remaining 210 employees? Cognition bought them days later
- Investors got about 4x their money ($243M raised, ~$1.2B to investors)
- But employees who joined in the last year? Many got nothing from the Google deal
For the talent coming over, that’s a great outcome. For some shareholders, it worked. For the employees who didn’t get hired by Google? Many got left holding the bag until Cognition’s rescue.
The Nvidia-Groq deal followed the same playbook. Nvidia paid $20 billion, but structured it as a “non-exclusive licensing agreement” specifically to avoid antitrust scrutiny. Groq’s CEO and president went to Nvidia. The company technically continues to exist. Everyone gets paid—but it’s not really an acquisition in the traditional sense.
Expect more of this. When the hottest asset is the team, not the product or the revenue, traditional M&A structures break down. And founders need to understand that an “acquisition” might mean the acquirer wants your engineers, not your equity holders.
And there are no rules in these deals. So far, VCs have done OK. But founders can always just… walk. And leave the other shareholders in many cases with nothing.
What This Means for Founders
If you’re counting on M&A as your exit path, you need to recalibrate your expectations.
The volume is there. Deals are getting done. The ratio of M&A to VC is the best it’s been in years.
The confirmed wins are down dramatically. Only 7% of deals confirmed returning 3x+ to investors. That’s down from 22% four years ago. And 90% of deals won’t even tell you what they sold for.
The buyers are different. Half of acquirers are now VC-backed themselves. That means less cash, more structure, more risk—and fewer disclosed outcomes.
Acquihires are rising. If your value is your team, not your business, you might get a deal—but your investors might not.
The mega-deals are real but rare. Yes, Wiz got $32 billion. Yes, Groq got $20 billion. But for every one of those, there are hundreds of quiet deals where we never learn what happened.
VCs Have to Swing for the Fences. The Math Got Harder.
M&A isn’t dead. But M&A as a reliable path to venture-scale returns? That’s gotten much harder.
In 2021, you had about a 1-in-5 shot at a deal that confirmed 3x+ returns to your investors. Today it’s 1-in-14. And 90% of deals won’t even tell you what they sold for.
Maybe some of those undisclosed deals were great outcomes. Maybe some unicorn-to-unicorn acquisitions will look genius in five years when the acquirer IPOs. We don’t know.
What we do know: the confirmed wins have collapsed. The silence has increased. And if you’re a founder thinking “worst case, we get acquired,” make sure you understand what that worst case actually looks like now.
It’s probably not a $100M exit with a press release. It’s probably an acquihire, a soft landing, or a deal with so much structure that the VCs get their money back and everyone else gets a job—and nobody talks about the number.
The 2021 M&A market is gone. Plan accordingly.
