A Concurring Opinion: Who Actually Discovers the Winners?

A16z's investor relations team published an essay a few weeks ago arguing that institutional LPs are underexposed to venture — particularly to the "best," i.e. brand-name, managers — and that the remedy is to concentrate more capital with them. Euclid VC responded with a detailed rebuttal, showing that much of a16z's headline evidence measures individual investors' cumulative profit dollars rather than fund-level net returns, and that the case for scale is considerably weaker than the essay's charts suggest. Odin's Dan G. made a related point the same week, describing the "power law keeps getting stronger" framing as doing more rhetorical work than the underlying data supports.

We agree with both responses, and this piece is not a rebuttal of either. It is an attempt to isolate one particular thread in the broader debate that we think deserves its own treatment, separate from the fund-return-methodology questions Euclid and Odin have already covered well: not whether mega-funds have superior access to winning companies — that seems likely true — but whether they discover those companies first. That is a narrower and, we think, more interesting question, and one the debate so far has mostly assumed rather than tested.

We should acknowledge the obvious conflict of interest. We are emerging managers. We have a vested interest in persuading allocators that the venture universe contains worthwhile managers beyond the usual suspects. Neither Euclid nor Odin nor GrayHair expects a large institutional allocator to wake up tomorrow, pull capital from a16z, and wire it to an emerging manager because of something we have written. We think the question is worth examining on its merits regardless.

Discovery is not access

The argument for concentration rests partly on a straightforward observation. Venture is a power-law business. A very small number of companies generate a very large proportion of the returns. And if the best venture firms have accumulated relationships, reputation, information and follow-on rights from previous winners, success can compound. That is difficult to dispute.

But there is a subtle leap between two propositions:

• —The best firms have disproportionate access to successful companies.
• —The best firms disproportionately discover successful companies before they are obviously successful.

Those are not the same thing. We think venture has at least three distinct skills:

• —Discovery: finding an exceptional company before the market has established that it is exceptional.
• —Selection: deciding which companies are likely to become exceptional.
• —Concentration: having sufficient conviction and capital to put meaningful money behind that judgment.

Large platforms can have enormous advantages in the second and third categories without possessing anything approaching a monopoly on the first. A mega-fund may be able to buy the winner after the market has identified it. That is an advantage. It is not evidence that the mega-fund was the one that knew it was a winner first.

So we looked at the companies

This is difficult to test perfectly, and we want to be upfront about the biggest limitation before presenting any numbers: our sample is outcome-based. We started with companies that became large and worked backwards through their financing histories. That tells us where discovery occurred among known winners. It tells us nothing about whether emerging managers are better at discovering companies that go on to succeed, because we have not looked at the much larger universe of early-stage bets — by mega-funds or anyone else — that didn't work out. A firm's win rate on the deals it chose to make is a different, and to us unanswerable, question from the one we're examining here. We return to this at the end, but we'd rather flag it now than let it look like an afterthought.

With that limitation stated plainly: venture financing histories are also messy on their own terms. "Seed" means different things to different companies; investor syndicates evolve; private-company valuations go stale; some companies disclose very little about their early financing. We have tried to be deliberately conservative throughout.

We looked at three sectors with enough large outcomes and sufficiently visible financing histories to make the exercise useful: legal technology, cybersecurity, and B2B insurtech. Legal tech is our vertical; cybersecurity and B2B insurtech were chosen for their structural similarities to legal tech, discussed below.

For purposes of the exercise, we classify an investor as a generalist mega-platform only where it resembles the large, multi-stage generalist platforms at the center of the access argument — Sequoia, Andreessen Horowitz, GV, and Lightspeed (included for comparable breadth and scale). We do not count every large venture firm as a mega-platform. Accel, Index, Bessemer, Norwest, Benchmark, Greylock and similar firms are substantial investors, but lumping every successful institutional VC into one category would make the exercise meaningless. Where one of these firms appears alongside a category-one platform in a company's earliest round, we note it, but it doesn't affect the classification.

Most importantly, we look at who was present at the earliest institutional financing, rather than whether a large platform eventually invested. We are focusing on discovery, not on whether someone eventually managed to buy a ticket.

Legal technology

We examined twelve significant legal-tech outcomes (latest disclosed valuations or transaction values in parentheses): Harvey ($15.5bn), Legora ($5.55bn), Clio ($5bn), Relativity ($4bn — rumoured IPO value), Filevine ($3.65bn), Ironclad ($3.2bn), Everlaw ($2bn), EvenUp ($2bn), Enter ($1.2bn), Norm ($1.2bn), Eve ($1bn) and Lawhive ($685m) — roughly $45bn in total.

Only three of the twelve — Enter, Eve, and Lawhive — had a category-one generalist mega-platform in the earliest institutional financing group. That's 25% of the companies in the sample, representing about $2.9bn of the $45bn, or roughly 6.4%.

Even that gives large platforms the benefit of the doubt. Harvey, now valued at $15.5bn, is the clearest example of the distinction: its earliest institutional capital came from the OpenAI Startup Fund and angels, with Sequoia arriving later. Today virtually every major investor wants exposure to Harvey — access and discovery were different events. Legora's early financing included Benchmark, Hummingbird, SV Angel, and Y Combinator, with Accel, Bessemer, General Catalyst, ICONIQ and Redpoint arriving in the much larger later rounds. Clio's early institutional backers were not the mega-platforms at issue here. Ironclad's $3.2bn valuation came after Accel and Sequoia arrived well past its earliest stage. Everlaw's Series D syndicate — TPG, H.I.G., CapitalG, Menlo, a16z, and others — reflects access to a successful company, not exclusive early discovery of one. EvenUp crossed $2bn long after early backing from investors including SignalFire.

The three clean category-one examples: Sequoia was an early investor in Enter, which reached $1.2bn. Lightspeed backed Eve at seed, ahead of its $1bn+ valuation. GV led Lawhive's seed round, ahead of a valuation approaching $700m. Those are genuine examples of large platforms discovering winners early — and a minority of the sample.

We exclude Norm from the category-one count. Coatue invested around the company's founding and later led rounds, but Coatue is a different type of investment platform from the generalist venture firms whose access thesis we're testing here. Norm's later financing became a broad syndicate including Blackstone, Bain Capital Ventures, Vanguard, New York Life and others, with Khosla leading the 2026 Series C. We would rather exclude a borderline case than quietly move the goalposts to flatter the statistic.

The point is not that large platforms are bad investors — several have demonstrated an extraordinary ability to recognize, finance, and concentrate in exceptional companies. The point is that their eventual ownership of successful companies doesn't establish that they uniquely discovered them.

Cybersecurity

A larger, more mature sample: Palo Alto Networks, CrowdStrike, Zscaler, Cloudflare, Okta, SentinelOne, Wiz, Snyk, Tanium, and 1Password.

Here the category-one discovery count is two of ten: Palo Alto Networks and Wiz. Palo Alto Networks is telling — Sequoia (our category-one criterion) led its earliest institutional round, with Greylock alongside it. Wiz is similar: Sequoia invested in 2020, well ahead of Google's eventual $32bn acquisition, with Cyberstarts — with its specialist cyber expertise — right there alongside it.

The rest tell a different story. CrowdStrike's early institutional backing included Accel. Zscaler's included Norwest. Okta's seed included Floodgate. Cloudflare's early financing included Venrock and Pelion. Snyk's early investors included Boldstart and Heavybit. SentinelOne's early backers included Accel, DCVC and UpWest rather than a category-one platform. 1Password was initially built without venture capital. Tanium similarly offers no clean example of mega-platform discovery at inception.

Using the headline Google transaction value for Wiz and Palo Alto Networks' current market capitalization, the two category-one discoveries represent a large share of the sector's combined value — well over a third by our estimate. (It is worth noting that Palo Alto Networks' valuation has fluctuated quite a bit lately - we are not investing in this sector, so cannot claim any inside knowledge, and thus went with one of the numbers that has been publicly reported). That is a large economic share from just two companies, but it is still two of ten. A platform can discover one enormous winner and generate extraordinary absolute dollars from it without that establishing it discovers most winners.

B2B insurtech

Our sample: NEXT Insurance, Coalition, Tractable, Shift Technology, Akur8, CoverWallet, and Embroker.

Zero of seven had a category-one generalist mega-platform in the earliest institutional financing. That doesn't mean these were obscure or undercapitalized companies. Coalition had early backing from Vy, Ribbit, Valor, Sam Altman and Deep Nishar before reaching a $5bn valuation. Tractable reached $1bn with early backing from Zetta and Ignition. Shift Technology crossed the billion-dollar threshold after early financing from European specialist investors, with Accel arriving later. NEXT Insurance reached $4bn before being acquired for $2.6bn. Akur8 has raised substantial institutional capital; CoverWallet was acquired by Aon; Embroker raised a $100m Series C.

We do not have access to comparable current valuation data across all seven to construct a sensible aggregate figure, so we will not manufacture one, but the company-level result stands: zero of seven major B2B insurtech outcomes required a category-one mega-platform to discover them.

But isn't this missing the point?

The strongest response is obvious: who cares? A large platform does not need to discover every winner. It needs to identify enough, invest enough capital, retain enough ownership, and generate enough absolute dollars to outperform. We agree — but let's do the math.

A specialist investor puts $1m into a company at a $20m valuation (5%) and ultimately, after multiple rounds and attendant dilution, owns 0.5% of a $20bn company: $100m on $1m, a 100x return. A mega-fund invests $25m in the Series A and $100m later, ending with 20% of the same company: $4bn on $125m, roughly a 32x multiple. The mega-fund has made vastly more money in absolute terms, but the same outcome at the fund returns level can look different, and there is no evidence that the mega-fund's "access" enables it to pick better in the earlier rounds.

There is another problem with "access"

"Access" has become dangerously elastic in venture. A large platform may have extraordinary access to a company once it's obviously successful, but access can be competed away as price is bid up and the information advantage diminishes. At that point the relevant question is no longer who can get into the company, but rather who knew it was exceptional before everyone else did.

The sector matters

Venture is not one market. A frontier-model company, a cybersecurity company, a vertical SaaS company, a legal tech company, and a biotech company have different founder networks, technical communities, buyer ecosystems, and financing patterns — so the relevant information advantage is often sector-specific. In legal technology, a specialist investor may encounter a company through a law firm partner, a peer operating in the space, or one of the hundreds of founders the GPs have met — before that company is interesting to the generalist venture ecosystem. The same principle likely applies in cybersecurity, insurtech, or other specialist markets. We are not arguing specialists are inherently better investors. We are arguing that information does not necessarily aggregate in proportion to assets under management.

Is there a monopoly?

The evidence we have examined does not support the proposition that the largest platforms have a monopoly on discovery. It also does not support the opposite proposition, that emerging managers are inherently better. It suggests something more modest: the concentration of venture outcomes does not imply the concentration of venture discovery.

A handful of firms may capture a disproportionate share of the economic value venture creates. That can happen because they discover winners, because they select winners particularly well, because they can concentrate more capital into them, because they retain ownership, because they get better follow-on access, or because all of those compound together. Those mechanisms should not be collapsed into the single word "access."

One final caveat

We flagged the core limitation up front: our sample is outcome-oriented by construction, so it tells us about discovery among known winners, not about which managers are best at discovering companies that will become winners — that would require examining the full universe of early bets, including the ones that failed, which is a different and much larger exercise we haven't undertaken.

One further caveat worth naming explicitly: being first doesn't necessarily produce the best fund-level return. The specialist who discovers a $20bn company but can't retain meaningful ownership may generate a worse fund outcome than a later investor who concentrates heavily in the same company. Discovery is one source of venture advantage. It is not the whole of venture.

Conclusion: A Concurring Opinion

We find ourselves in agreement with Euclid's and Odin's broader case, and in partial agreement with the concentration thesis itself. The best-known venture firms can possess extraordinary advantages. Large platforms can attract better founders, secure larger allocations, invest more capital, support companies more extensively, and participate in later rounds.

But there is a difference between saying the best-known platforms have disproportionate access to winners and saying they have disproportionate discovery of winners. The evidence we have examined suggests the latter proposition is considerably harder to establish — and we think that distinction should matter to allocators. If discovery is more distributed than outcomes, a venture portfolio built solely around the firms that already captured the largest share of historical outcomes may be concentrating not only capital, but the assumption that historical access necessarily predicts future discovery. We think the argument should be narrower than "the winners are concentrated, therefore the discoverers are concentrated."

David Teten Itamar Novick — if it is not too presumptuous of me - what's your take on this debate?