There are roughly 1,000 companies carrying unicorn valuations in the venture growth market. Six of them have reported a down round in the past 12 months.
Read the funding headlines and private markets look like an unbroken run of up rounds. Christian Munafo's point is that a company which never has to reprice never records a loss, and that about half of those 1,000 have not priced a new round in two or three years.
"So these companies are going to make the runway last as long as they possibly can until they have to reprice themselves."
Munafo has spent 25 years in private markets and has traded secondaries since the early 2000s, when the whole market did five or six billion dollars of deals a year, so the comparison he is drawing is one he has watched from inside.
The full episode is covered here so you can skip it. 33 minutes of audio, 21 minutes of reading.
Here are the 13 takeaways that matter.
👤 Guest: Christian Munafo, Head of Private Growth Strategies at VanEck, who has worked in private markets for 25 years and in secondaries since the early 2000s
🎙️ Hosts: Michael Batnick, Managing Partner at Ritholtz Wealth Management, and Ben Carlson, the firm's Director of Institutional Asset Management
📰 Published: 14 September 2026 on the Animal Spirits podcast feed (The Compound)
🟢 Spotify | 🟣 Apple Podcasts | 🔗 Show notes | ⏱️ 33 min | ✅ Time saved: 12 min
Key Takeaways
Six of about 1,000 unicorns have reported a down round in the past 12 months
Not because the marks are right, but because nobody is forced to set one
About half of those 1,000 have not priced a round in two to three years
Munafo puts the stale value at low trillions of dollars of market cap
A company can now reach $100M of revenue in 12 to 18 months instead of eight to 10 years
Which also means a competitor can, so speed cuts both ways
Valuations are being awarded for product rollouts that have not happened yet
The capital deployed in the zero-rate years has still not come back to investors
Secondaries have gone from $5B to $6B a year to roughly a quarter trillion
Private equity is the hardest of the three private strategies to sell to advisors
Eight to 10 years locked up for two to three points over the S&P
The biggest AI models and the open ones will coexist, splitting revenue from usage
1. The 2021 Hangover
Before the interview, Michael Batnick and Ben Carlson spent several minutes on the period the conversation is about. Carlson opened with the pitch: "Michael, I love a good straight shooter. On today's show, we have one."
Carlson's framing of 2021 and 2022 was a morning-after scene. The pricing of late-stage growth companies, he said, got absolutely insane — "it's like the hangover scene the next day where everyone's walking through the hotel room and there's beer bottles everywhere and there's a tiger in the bathroom and everyone kind of wakes up goes, what happened?"
Batnick placed where he had made the same point before, and how he had put it. "Oh, it was actually on TCAF. A comment was made about valuations back then." His conclusion at the time was not to let investors off the hook entirely, but: "But like, we were all a little bit drunk."
Carlson's practical instruction to anyone marking a private position: if there has not been a priced round in a while, take a haircut. He then made it a joke about the two of them — "Not a Ben Carlson haircut, a Michael Batnick haircut. There we go. All the way to the skin."
Batnick's reason for going into it was the amount of money looking for a home. Everyone was at home on Zoom in 2020 and 2021, Carlson said, and the two of them had repeated conversations with people whose pitch was an idea with no business model and investors throwing money at them anyway.
Batnick's own analogy for what happens next is deliberately undramatic. "But it's not going to be a car crash. It's going to be, what, a flat tire that catches up with you five miles down the road?"
2. Growing Into a 2021 Price
Batnick opened the interview on a company he had been discussing the week before, and used it as the case study for the whole problem.
Airbnb's stock is at about its highest level since it came public, he said, and the story is the starting valuation rather than the business. "It came public in 2021 during a very different environment, trading at whatever, 70 times sales." The company was unprofitable and the market capitalization did not make sense; it took five or six years of operating to grow into it.
Munafo agreed the question was the right one and gave the general version. "I think, as you said back in 21, you had just these absolutely incredible valuation runs without, in many cases, the underlying operating metrics to support them."
The mechanism has changed since then, and it is speed: "I think now you have this environment where you have companies that used to take maybe eight to 10 years to hit 100 million in revenue. They're hitting it in 12 to 18 months, right? In some cases, you have companies hitting a billion revenue inside of a couple years, which is just, it's just incredible."
The shape of the market has changed too. In 2021 the run-up was everywhere, across every innovation theme. Now, he said, there is more concentration of value and more concentration of capital in fewer and fewer names.
His verdict splits: "And in some cases, I think, frankly, the valuations have gotten ahead of themselves. And in other cases, it's actually warranted."
3. Paid Before They Can Ship
Batnick asked whether discipline is harder now, given how much money is chasing a small number of deals on take-it-or-leave-it terms. Munafo said yes, and named exactly what he is refusing to pay for.
"I think in this time around, there's more money but going to fewer companies." For the companies getting it, he said, some of the valuations are really hard to swallow.
The two themes he sorts everything into are AI — touching software, hardware and physical products — and the reindustrialization of the economy: space, defense, advanced manufacturing, energy.
The specific thing that stops him is being asked to pay for manufacturing that has not been proved: "So you have companies in like the defense space, for instance, in certain areas that are generating, all sorts of capabilities, autonomous systems, and they're getting valuations that essentially give them credit for executing on a rollout of multiple product lines."
"I think for us, like that's the biggest thing that we're getting concerned about and while we're pausing, other investors are not." He can see the argument on the other side — the addressable market is so large, and the founders are often seasoned operators who have done it before. "Frankly, we're trying to avoid that."
The alternative tactic is to get the same exposure at a different price: buying secondaries at better pricing to offset the valuation froth. "But you're absolutely right, it is hard to be disciplined."
4. The Money Never Came Back
Munafo then widened out to the decade that produced all of this, and his argument is that the bill has not arrived rather than that it will not.
The years after the financial crisis and before COVID were, he said, a nice glide path: a lot of money subsidizing a lot of interesting companies that transformed the world and returned a great deal to investors at every stage. "And that environment is now over."
The new problem is that speed runs in both directions. "Investing during disruptive times and disruptive technologies, who the hell knows where this is going" — a company that reaches a billion dollars of revenue overnight can be disrupted overnight, because there are a million copycats.
His firm's shorthand for that: "we have a saying if it goes up like a rocket it could fall like a bomb." In many cases there is no moat at all; in hardware, capital itself becomes one. "So your ability to outraise the cohort that you're competing against, in some instances becomes a bit of a moat."
The part he says is not being reported is what happened to the money put to work during the zero-rate years: "But if you actually look at the data, there's still a tremendous amount of unrealized value locked up from those ZERP-oriented capital deployment vintages that investors have not received, right?" He is describing capital committed while interest rates were near zero, which has still not been returned.
"There's one of the lowest kind of distribution cycles over the past kind of five, six, seven years that we've seen in a very long time." A handful of large listings will not fix it: more and more of the exit value anyone reads about is in fewer and fewer names.
"The story that's not being told is there is a tremendous amount of unlocked value that's continuing to sit out there." And some of it is not value at all: "But there's going to be a lot of capital destruction from that era, from companies that have been rendered obsolete and that frankly have not pivoted enough and don't have an ability to raise more capital to sustain themselves."
He was careful not to overclaim on timing. "The ending of the story is not yet written."
5. The PE Zombie Numbers
Batnick brought the show's own research into it: a Pitchbook report he had read that morning on what it called the private equity zombie problem. This is the host's material, not the guest's.
By Pitchbook's count, as Batnick read it, there are 13,509 private-equity-backed companies currently sitting in US sponsor portfolios. Of that universe, 33.8% have been held for more than five years, and 2,536 have exceeded the traditional exit window.
The dollar figure he read out: "So GPs are sitting on more than $860 billion in buyout nav across funds that are more than seven years old." NAV is net asset value — the general partners' own carrying value for those businesses.
Munafo's addition is that buyout assets carry a second problem the headline number does not show: "The other aspect of that on the buyout side, Michael, is those companies are all levered, right? So you've got a leverage problem on top of the duration issue for the equity investors." And on top of the leverage, an interest rate problem.
The venture growth companies he invests in are the opposite case, and he is blunt about why: "These companies typically don't have leverage. They don't deserve it." They are not printing cash or paying dividends; they are spending on growth.
6. Six Down Rounds in 1,000
Munafo turned the second Pitchbook statistic into a quiz for the hosts, which is how the episode's headline number arrived.
The universe he set: roughly 1,000 reported unicorns in the venture growth landscape. Some publications say more; he told them to work with 1,000.
His question: "How many of those do you think have reported down rounds in the past 12 months?" Everyone hears about up rounds, he said, so how many down?
Both hosts guessed low, then Batnick reversed and went high. "No, I was going to guess 30%. 300."
The answer, and the reason: "Six companies. Yeah, because they don't want to yet, right?"
The consequence is that the clock keeps running rather than the number changing. Some of these companies may be profitable, he allowed — "Now, you can on one hand, say some of them may be profitable. Okay, great. If you're profitable, you don't need to raise money. God bless you."
"The reality is most of these companies, shocker, are not yet profitable. So they're using the runway and extending it as long as they can until they have to reprice, until they get bought on the cheap, or until they go away."
7. Half Have Not Repriced
Batnick asked a follow-up quiz question that puts a size on the first answer.
His question: "So how many of these companies, the 1,000 companies, right? What percentage of those would you say have not priced a new round in the last two to three years?" The guess in the room was 80%.
Munafo's answer was about half — and he attached a number to it: "So about half of those companies, you're talking about probably low trillions in market cap of arguably stale market cap value."
The behavior change is the tell. "These are the companies that were raising every six months. They were. They're not now, right?" They raised enough that they have the runway not to.
Carlson asked how many of the 1,000 are software companies. Munafo did not have the figure to hand and offered to come back with it: "A lot. I can come back to you with the stat. The answer is a lot."
He kept insisting this is arithmetic rather than pessimism: "This is not me being like Dr. Doom. This is just saying that you're looking at these numbers, right?" There will be incredible wins and incredible success stories — companies raising three or four rounds in 12 to 18 months, each one larger than the last. "But the frequency of that across the book is smaller."
"The question is, what happens to the rest of them?"
8. Why 2029 Is Not Far Enough
Batnick put the cynical version of the industry's position to him: that the reckoning is far enough away to be someone else's problem.
His framing: "I think investors mentally understand that the 20, 21 vintages are terrible, and maybe they see AI as their get-out-jail-free card, and they think that the 24, 25, 26 vintages will be better." By the time the earlier funds actually show up as cash returned or not returned, he said, the vintage is nine years old. "So these numbers will show themselves actually they will mark to market in 2029. Who cares? It's so old."
Munafo said the argument used to work and now does not, because the buyer is out of room: "However, if you factor in that the exit activity for the past 10 years has been pretty low, you have a situation where you have investors that are already quite overweight illiquid assets." Without a path to turning that into cash through more mergers or more listings, a lot of institutional investors are tapped out.
That, he said, is why every manager is now courting registered investment advisors and the wealth channel: a new pool of capital without the legacy overhang of what he called zombie capital. He thinks that is good for the ecosystem and needs handling carefully for all the reasons above.
The fundraising conversation has changed accordingly: "Hey, we did fund three a couple years ago. It's time for fund four, either pony up or you're not going to be included anymore." Managers, he said, will have a much harder time raising.
9. Secondaries Are the Valve
Batnick asked what actually happens if allocators simply refuse the next fund. Munafo answered in one word and then gave the size of it.
"So one word, secondaries."
The scale change over his own career: "So I've been in secondary since the early 2000s. And back then there was probably like five, six billion of annual deal volume." Back then it was all musical chairs with limited partner stakes — an endowment selling its interest in a fund.
"And the last 20 years, you've seen this evolution where now annual deal volume this year will probably be a quarter trillion, at least."
The composition has changed as much as the size: "And at least half of that, it's going to involve everything but LP stakes." He listed the transactions that make up the other half — fund recapitalizations, winding down funds, and strip transactions that manufacture liquidity for the manager ahead of exits so capital can go back to investors and be recycled. His firm was doing them 20 years ago, he said, before they had names.
"Secondaries is going to be a major release valve for all of this unlocked, unrealized NAV that's creating problems." That covers fund-level solutions and asset-level ones: tender offers, liquidity for employees, cashing out early investors.
10. What Private Equity Sells
Asked what discounts look like, Munafo gave the ranges, and then explained why he thinks buyout funds will struggle to sell themselves to financial advisors.
The normal-market ranges, by asset class: "Usually, like, in a normal environment, you'll see, like, I don't know, 10 to 30 percent discounts in venture, maybe, like, single digits to 15 percent in buyout and probably similar with, like, with real assets."
In a period of disruption — he pointed at 2022 through 2024 and 2025, when a run of consecutive rate increases hit private markets — "Those discount ranges at least double."
"But the higher quality managers and the higher quality assets, even in those market cycles, can actually defend better pricing." The pattern is the same one he keeps returning to: more capital flowing to the perceived winners, whether that is a fund manager or an operating company.
His ranking of what the wealth channel will buy: "I think private equity, primary fundraising, will have a problem with the wealth channel. I think secondaries and private credit make a lot of sense." Secondaries is a very easy story to understand. Private credit is income, floating rate, no interest rate sensitivity, which people like.
Private equity is the hard sell, and he made the pitch out loud to show why: "All right, I'm going to lock my money up for eight to 10 years. Maybe I'll beat the S&P by two to three percent, maybe if they're really good a little bit more. I don't know. That doesn't sound that attractive to me, especially when advisors and clients are looking in the rearview mirror."
The rearview mirror is the problem, because the index has been strong. "But the reality is we've gotten 14% for the last decade." He was explicit that he is describing how people decide rather than what they should expect.
Strip out the borrowing and the comparison gets worse: "So your actual unlevered returns may be more like 8 to 10 or 12% but whatever."
The lock-up depends entirely on the wrapper. A 1940 Act closed-end fund with structured liquidity — an interval fund or a tender offer fund — lets an investor take partial liquidity during the redemption windows. A traditional drawdown fund for qualified purchasers does not. "Shocker, they're never actually 10 years, especially if you're in venture, it's more like, 15 to 20 years."
11. VanEck's Rifle Shot
Batnick thanked him for using the data plainly, said the backdrop he had described sounds relatively negative, and asked what he is actually doing with it.
The method: "So our approach with building out this kind of later stage private growth strategy at VanEck is focus on the major themes, look at the companies that have demonstrated significant traction." He calls it a rifle shot approach.
The themes are the same ones VanEck's listed thematic funds are built on — reindustrialization, electrification, de-dollarization — applied to private companies.
The traction test is specific: "So the technology works, clear product market fit, a lot of customers, hundreds of millions to billions in revenue." On top of that, operating metrics that can actually be examined, seasoned operators and good governance. He is not buying two people with a whiteboard, which he said is a perfectly fine place to put money at a different risk and reward.
The risk he says he owns is the Airbnb risk in reverse: "Like, did we miss the run up by waiting for this underlying asset to be too de-risked? And then the answer is yes, then we don't do it." The exception is where a discounted secondary and a friendly capital structure let him sit slightly below the last private round.
"As I say, the market's concentrating in fewer names." So the portfolio is built as diversification across the major themes, inside the set of perceived winners.
12. Frontier vs Open Models
Batnick noted that two of the companies most affecting how capital is allocated in the United States — OpenAI and Anthropic — are private, and asked how VanEck sees that.
Munafo would not discuss specific holdings but said his team follows the space closely, and that the two leading frontier models are currently in a private wrapper. He added that Elon Musk's Grok should not be left out of the picture.
The demand case for the expensive models is a particular kind of customer: "Our view is you're going to continue to have a use case, especially for enterprise level super users" and for continuously running agent systems, as long as the frontier models keep their performance advantage over the cheaper open ones. He flagged that condition himself: if the gap closes, the question changes.
"But we also think you're going to increasingly see orchestration across the open models that are more efficient and frankly more cost-effective."
Asked how much of a threat that is to the two big companies, he split it by revenue rather than by usage: more of the revenue goes to the large frontier models because the heavy users pay for them, while the open models get a smaller share of revenue and more ubiquitous use. "We don't think it's like one model takes all. That's not our view."
Carlson stopped him to ask what open source even is: "Can I ask you a basic question? I know nothing about this world."
Munafo's explanation was about visibility: "So the whole idea is if you go with like a frontier model, a closed model, you don't really have a view as to how all of the data is being processed, how it's being interpreted." Every model is trained; the question is whether you can see how it turns a prompt into an answer, which is a question about the model's weights.
"Those on the open side want to have more kind of visibility and transparency into how the sausage is getting made, right?" The second motive is data sovereignty — he pointed at Palantir's chief executive making the argument that closed models take your data and turn you into a competitor.
The practical split he drew: curing cancer, heavy capital-markets work or agents running non-stop means paying up for the best model even though it is closed. Building your own agent workforce for specific tasks means using something like Perplexity's orchestration to match you with the right open model, more cheaply and with visibility into how it runs.
The catch with open models is the hardware bill: "The other thing about open models is you may also need to manage your own infrastructure, which not everyone can do." Real data sovereignty means owning the machines. "You need to own hardware. You need to own chips, all that stuff. Not everyone's equipped to do that."
13. Diversifying in Private AI
Batnick's last substantive question was about concentration: the leaders change weekly, but a private position is locked up for years, so how do you diversify against that?
Munafo's first answer is patience. Waiting gives a better view of what the outcome actually looks like — and even a company that goes public leaves its holders locked up for a while.
The construction is one name per layer rather than one name overall. Inside AI software, he said, that means a frontier model and an orchestration layer that connects clients to whichever open models suit them. Then separate positions in infrastructure, energy or compute; in fintech and payments; and in defense, including autonomous air, land and sea systems and something in space.
The reason he thinks there is anything left to buy once the famous names list: "Because you got like $6 trillion right now, roughly, of value across these unicorns alone, right?" Some of which, as he had already said, may carry stale valuations.
The specific names he offered as compelling: "So from our perspective, we think companies out there that are quite compelling are companies like Databricks, right, which is the equivalent to, or somewhat the equivalent to like a snowflake." He puts it in the data aggregation layer. He also named healthcare, on both the administrative and technology sides, as a place where use cases are only now appearing.
The constraint behind the whole AI theme is electricity: "But we have a real issue with energy. Like these intelligence units, right, we need watts to create them." Then: "Regardless of what your views are, we need power to fuel these things." And massive infrastructure to fuel them, wherever people are willing to have it built.
"So our view is you've got to be patient, you got to be disciplined" — and, he said, a large number of companies will keep staying private for longer and building value before they reach the public market.
Bonus Insights
Asked what would happen if the locked-up capital finally came free, Munafo said it would go looking for a home and land in the same place everything else has. "So as that capital unlocks, it's going to look for a home." The allocation, though, keeps going to fewer managers and fewer companies — "And I think you're going to have, again, a big reckoning across the broader space as companies have to get repriced."
On who VanEck is building these products for, his answer was three tiers of wrapper rather than three types of investor: a solution for qualified purchasers and institutions, something being designed for the wealth channel, and retail-oriented products still in the thinking stage. The pitch he uses is a translation into listed-market terms: what you used to look for in a small and mid-cap growth strategy in public markets, delivered in a wrapper that suits the end client. He noted that small and mid-cap growth in listed markets does not excite many people now.
Batnick's reaction when Munafo said many of the stale-marked companies have probably already been disrupted was one line: "A thousand percent."
Batnick's read on his own guest, delivered to the audience before the interview ran, was that the down-round figure is what the episode turns on. He called it "the stat of the show."
Munafo's bottom line is that private markets have a reporting problem rather than a pricing problem: roughly 1,000 unicorns can hold valuations set years ago because nothing forces them to set new ones, the capital behind them has not been returned to investors, and the release valve is not a wave of listings but the secondaries market, which now clears roughly a quarter of a trillion dollars a year.
Products, Companies & Tools Mentioned
VanEck (Munafo's firm, where he runs the later-stage private growth strategy and applies the same themes — reindustrialization, electrification, de-dollarization — that its listed thematic funds are built on)
Airbnb (Batnick's case study for the whole conversation: it came public in 2021 at about 70 times sales, unprofitable, and took five or six years of operating to grow into the price)
Pitchbook (Source of both sets of numbers in the episode — the private equity "zombie problem" report Batnick read out, and the unicorn down-round count Munafo turned into a quiz)
OpenAI and Anthropic (The two leading frontier models, both still private; Munafo expects them to keep the heavy enterprise users and most of the revenue as long as their performance lead holds)
Grok (Elon Musk's model, which Munafo said should not be left out of the frontier picture)
Databricks and Snowflake (Databricks is the private company he named as compelling, placed in the data aggregation layer and compared on air to Snowflake)
Perplexity (His example of an orchestration layer that matches a customer with the right open model for a task, more cheaply than a frontier model)
Palantir (Whose chief executive, Munafo said, has made the argument that closed models take a customer's data and turn the customer into a competitor)
SpaceX (Named repeatedly as the exception rather than the rule — the kind of listing that gets written about while most of the unrealized value stays where it is)
Books & Resources Mentioned
The Pitchbook report on the private equity "zombie problem" (Batnick read its figures on air: 13,509 sponsor-backed companies, 33.8% held more than five years, and 2,536 past the usual exit window)
TCAF, The Compound and Friends (The show where Batnick says he first made the point about how 2021 valuations happened)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: