Crypto VCs Are Crowding Into Late-Stage Deals, and Varun Datta Says That Is Herd Behavior

Crypto venture capital has spent the last several quarters pulling back from founding-stage rounds and piling into proven companies, and Varun Datta, founder of Truth Ventures, argues in the latest Crypto Long & Short column that the pattern is consensus disguised as discipline. The contention rests on a single statistic: 57% of last quarter's crypto venture capital flowed to already-proven companies. For Datta, that concentration is the tell, because a market where every allocator reaches for the same handful of names is not exercising judgment; it is following the herd. The piece, dated September 2, 2026, is written as a practitioner's challenge to the prevailing mood inside crypto funds, where late-stage checks have come to feel like the responsible choice during a multi-year risk-off stretch.

Why is the late-stage pile-up happening now?

Crypto venture entered 2026 in a defensive posture that took shape well before this quarter. Token prices spent much of the prior two years range-bound, several high-profile venture portfolios were marked down after the 2022 shocks, and limited partners grew louder about DPI, distributions to paid-in capital, rather than paper IRRs. Allocators responded by narrowing their check writing to companies with revenue, with users, or with named institutional customers. The effect has been visible across the data the industry itself publishes: median pre-seed check sizes shrank, seed rounds increasingly included SAFEs with MFN clauses, and Series A activity thinned out except at the top of the market. In that environment, writing a follow-on into a company that already cleared its last round felt safer than underwriting a brand-new protocol.

The pullback also tracks the broader cost of capital. Public-market investors in 2023 and 2024 demanded higher hurdle rates from growth-stage software companies in particular, and crypto funds benchmarked to those comps raised their own internal rates. A founding-stage protocol, by construction, sits years away from any comparable multiple, so it is the first thing cut from a model's output when discount rates rise. Datta's argument is that this mechanical response is being mistaken for strategic discipline. A consensus trade can look like prudence in a spreadsheet, but it is still a consensus trade, with all the competition for entry and all the compressed upside that implies.

Finally, the concentration reflects the gravitational pull of a small number of brands. Late-stage rounds in the most-watched infrastructure and stablecoin names have anchored headlines, and the funds that were not invited into those rounds have often defaulted to their nearest comps rather than to genuinely new bets. Datta's read is that this is a marketing artifact as much as a financial one.

What does Datta actually mean by consensus versus discipline?

Discipline, in venture, is usually defined as price sensitivity, willingness to walk away from a round, and a clear thesis about why a specific company at a specific valuation will compound. Consensus is defined as everyone else reaching the same conclusion for the same reasons at the same time. The two look identical in a deal log. The difference shows up in returns. If a category is genuinely well-understood and the leading companies are correctly priced, late-stage capital earns a venture return. If the category is mispriced because every allocator anchors on the same comparables, late-stage capital earns a beta return at best.

Datta leans on the 57% figure to make this concrete. When nearly three out of every five dollars go to companies that already raised at the valuations the market is now anchoring on, the average entry price is set by auctions, not by underwriters. That tends to mean the marginal dollar earns less than the first dollar, because the founders in those rounds have leverage and the funds do not. By contrast, founding-stage rounds, where there is no comp and no auction, are where an individual investor's edge can actually move an outcome. The claim is not that late-stage is bad per se; it is that late-stage pursued in a crowd is bad, and that the current data shows it is being pursued in a crowd.

This framing matters because it reframes the 57% number from a comfort signal into a warning. The same statistic that makes a quarterly LP letter sound prudent is, in Datta's reading, evidence that the average allocator is doing the same thing at the same time. For a fund whose pitch is independent judgment, that is a problem.

What is the background on Truth Ventures and the Crypto Long & Short column?

Truth Ventures is an early-stage crypto venture firm founded by Varun Datta, focused on infrastructure and application-layer investments at the seed and pre-seed stages. The firm's posture is explicitly pro-founding-stage, which shapes how Datta reads the data; his critique of the late-stage pile-up is also, in part, a defense of the part of the market he operates in. Crypto Long & Short is the regular CoinDesk newsletter and column that frames crypto markets through a practitioner lens, and it routinely invites outside contributors to argue a specific point of view rather than report consensus. The column is read by allocators, founders, and analysts who want a structured take rather than a market summary.

The September 2, 2026 installment sits inside that tradition. It is an op-ed, not a data report, but the data point it builds around, the 57% concentration figure, is the kind of statistic that gets circulated widely when published in the column. For readers tracking the venture side of crypto, the column tends to be a marker of what serious allocators are arguing about among themselves, which is why the framing of consensus versus discipline is likely to be quoted in subsequent investor memos even if Datta's specific prescriptions are debated.

How does this fit the bigger trend in crypto capital allocation?

The wider trend is one of bifurcation. Crypto venture is splitting into a small group of mega-rounds at the top, where strategic investors and crossover funds compete for ownership, and a much larger group of small, founder-led rounds at the bottom, where the check sizes barely cover a runway. The middle, the Series A and Series B band where companies used to graduate from seed traction into institutional capital, has hollowed out. The 57% figure is consistent with that hollowing: when capital concentrates at the top, the proportion going to proven companies rises mechanically, even before any fund makes a deliberate choice.

A second trend is the rise of strategic capital. Public-company treasuries, exchanges, and infrastructure providers have become more active lead investors in late-stage rounds, both for product alignment and for optionality on token launches. Their participation has crowded out some financial investors at the entry point and compressed the return profile for everyone else. Datta's critique fits here too, because strategic-led rounds often price on relationships rather than on the kind of return math a venture LP would run.

A third trend, more specific to this cycle, is the way stablecoins and payments rails have absorbed disproportionate attention. Several of the most-funded late-stage rounds of the last several quarters sit in stablecoin issuance, payment orchestration, and onchain FX, areas where the underlying economics resemble fintech more than crypto. That has drawn capital away from the application and protocol layers where founding-stage bets are more common, and it has reinforced the late-stage skew Datta is calling out. None of these trends is new on its own, but together they explain why a single quarter can deliver a 57% figure without any individual allocator feeling like they are doing anything aggressive.

What does Datta say the founding-stage gap actually looks like?

The founding-stage gap, in Datta's framing, is the widening distance between the volume of pre-seed and seed-quality deals being formed in crypto and the volume of capital willing to lead them. He argues that this is where asymmetric returns sit, because at this stage the market has not yet priced in the company's progress, and the investors who underwrite the round set the comp rather than anchor to one. Historically, venture math holds that a small number of outlier wins fund an entire fund, and that those outliers are far more likely to be found in the first professional round than in any subsequent one. When capital retreats from that layer, the expected value of the entire asset class falls, even if the late-stage entries feel safer in the moment.

Datta also points out that founding-stage discipline is harder to fake, which is precisely why so few funds attempt it. It requires saying no to categories that are working in the public narrative, writing smaller checks into companies that do not yet have traction metrics, and accepting illiquidity for a longer window. In a quarter where every LP meeting is about DPI, those are uncomfortable choices. His argument is that the discomfort is the edge, and that funds confusing discomfort for risk are the ones most likely to underperform the cycle they are trying to play it safe through.

What are the three things to look for next?

Datta closes the column by laying out three markers that he says will reveal whether the late-stage retreat is genuinely durable or whether it has already begun to reverse. The first is the share of capital going to founding-stage rounds in the next two quarters; if that share recovers from its current low, the consensus trade is unwinding. The second is the number of new lead investors emerging at the seed layer, since the existence of willing leads is the bottleneck on the whole founding-stage pipeline. The third is the pricing behavior of those seed leads, specifically whether they are willing to anchor rounds at valuations that do not import late-stage comps.

For traders and investors who watch crypto venture as a leading indicator of where the next cycle of tokens, protocols, and listed companies will come from, those three markers are the ones to track. A reversal in the founding-stage share would suggest that capital is preparing to underwrite the next wave of projects rather than just defend the last one, and that would matter for token issuance windows, airdrop calendars, and the broader risk appetite of the ecosystem heading into 2027. Until then, the 57% figure is the line Datta wants allocators to remember whenever a late-stage check starts to feel prudent.

What should readers watch next?

The next concrete data points will come from the Q3 2026 venture aggregations that trackers like Galaxy Digital, Messari, The Block Research, and Electric Capital publish in the weeks ahead. Each of those reports will refresh the 57% number, will break it down by stage, and will show whether stablecoin and payments deals continue to dominate the top of the leaderboard. Any move in the founding-stage share, in either direction, will set the tone for the next round of LP letters and is likely to be referenced in the next Crypto Long & Short installment.

The other thing to watch is the behavior of the funds that publicly commit to founding-stage discipline. If their deal counts and check counts hold up while the broader market retreats further, that is evidence Datta's framing is correct and that the edge is real. If those funds soften and quietly write follow-ons instead, his critique applies to them too, and the 57% concentration is the kind of number that gets cited again the next time the column wants to make the consensus-versus-discipline point.