Private technology companies are worth roughly $5 trillion in aggregate, and Vaibhav Agrawal — the early-stage investor who goes by Dr. V — thinks venture capital is still undervalued against that number.
The standard reading of a private market that size is that technology has been bid up too far. Agrawal's starting point, learned from one of Lightspeed's founders, is that the same reading has been offered for decades and has been wrong each time.
"He said the market always seems to be overvaluing technology. And ten years later, you realize it wasn't."
He spent nearly a decade at Lightspeed Venture Partners before spinning out to start his own early-stage fund, ODDBIRD VC; he trained in medicine in India and worked in emergency rooms before he ever saw a cap table, and he says Lightspeed's two managing partners wrote the first check into his new fund.
The full interview is covered here so you can skip it. 34 minutes of audio, 21 minutes of reading.
Here are the 14 arguments that matter.
👤 Guest: Vaibhav Agrawal, who goes by Dr. V, Founder of ODDBIRD VC and previously an investor at Lightspeed Venture Partners for close to a decade
🎙️ Host: David Weisburd, who hosts How I Invest and co-hosts The 10X Capital Podcast
📰 Published: 14 September 2026 on How I Invest with David Weisburd
🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 34 min | ✅ Time saved: 13 min
Key Takeaways
Venture capital is undervalued relative to the size of the private market it is chasing, and AI has so far only disrupted one job
Most of Anthropic's and OpenAI's revenue comes from coding, which he treats as the first of several industries, not the whole opportunity
Which work AI takes first is predictable: low cost of being wrong, and a right answer the model can check itself against
Coding sits at one end of that test and medicine at the other
Sourcing deals is now a media business, and exits are now an internal function with its own team
Lightspeed built a capital markets team that makes investors defend a selling plan in a Monday meeting
Co-investment is the thing limited partners ask for most and use least — he puts the real participation rate at about 2%
The blocker is governance: the best rounds need a decision in under five days and most investment committees meet twice a month
Family offices investing directly are the most poorly selected buyers in the market, in his view, because the good deals never reach them
Venture firms are turning into diversified asset managers because recurring fee income, not performance, is what makes an asset manager valuable
The failure mode is a fund size that grows faster than the firm's actual advantage
Markups have become the industry's scoreboard, and that pushes every junior investor toward the same crowded deals
He calls it a systemic momentum trade that runs from founders through investors to the people who allocate to them
Pattern matching produces returns; pattern breaking produces legendary ones
His best Lightspeed investment was a business-to-business marketplace nobody had a category name for in 2018
Consumer and fintech are the two sectors he thinks the Valley has oversold
Fifteen of his fund's seventeen investments are repeat founders, many of them recent arrivals in the US
1. Venture Is Undervalued
Agrawal opened on valuation, and specifically on why he does not trust the reflex that says technology is expensive. He credited the frame to Chris Schaepe, one of Lightspeed's founders.
"He said the market always seems to be overvaluing technology. And ten years later, you realize it wasn't."
His argument is a ratio, not a level: the aggregate value of private technology companies against the pool of venture capital chasing it. He put the private market at about $5 trillion in aggregate today, against a period around the financial crisis when the largest private company was Facebook and the biggest Silicon Valley funds raised a fraction of what they raise now
He said the scale of capital has reset rather than inflated. There was no SoftBank Vision Fund and there were no mega hedge funds in private markets then; Lightspeed, Andreessen Horowitz, Sequoia and increasingly Benchmark now run in excess of $50 billion to $70 billion
"The markets have completely reset in terms of the capital scale and expected value"
The part of the argument he leans on hardest is how little of AI's opportunity has been realized. Most of the revenue at Anthropic and OpenAI comes from coding, and medicine, engineering and physical automation have barely been touched
"When I look at the degrees of possibility, it just feels very, very large"
2. Which Labor AI Takes First
Weisburd put it to him that the big unknown is whether AI will actually disrupt labor. Agrawal rejected the framing.
"I don't think it's an unknown, right?"
He said coding is already the proof. Tools like Cursor and Claude are eating into work that ranges from research-level code to basic systems implementation, which is engineering labor by any definition
The useful question, in his account, is sequencing rather than possibility. "The question is what kinds of labor get disrupted first?"
He gave two tests that decide the order. The first is a low cost of being wrong, where there is room to course-correct and the user does not have to extend much trust. The second is a deterministic right answer, so the feedback loop closes and the model improves continuously
Coding scores on both: code either performs as expected or it does not
Medicine sits at the opposite end. The cost of being wrong is very high because biology is multi-system, and it is hard to attribute a response — or an unintended one — to the algorithm alone
3. Venture as a Media Business
Asked why the old venture playbook no longer works, Agrawal described two changes at opposite ends of the fund's life: how deals arrive, and how they leave.
At the front end, sourcing has become a distribution problem. "Ten years ago, venture was not media business. It was a reputation business where investors would invest in over decades and they were known for investing in certain companies and therefore became aspirational founders to raise capital from"
He pointed to TBPN being acquired and to Harry Stebbings having deployed over a billion dollars, by his account, as evidence the shift is settled. "It is now widely established that distribution is important for a venture firm to be able to see the best opportunities"
At the back end, exits used to require no effort. The conventional assumption was ten investments, one or two that hit, eight that go sideways and need no management, and the winners going public on their own schedule
That has been replaced by continuous position management. An investor now has to keep asking whether to sell a company or buy more of it, which requires a standing set of relationships with buyers and sellers that did not have to exist before
Two trends are pulling against each other. Companies are staying private much longer, while the average time an investor stays at a big firm is falling
4. Why Investors Keep Moving
Weisburd asked whether investors are being poached. Agrawal said yes, and described the staffing problem it creates.
He splits investors into three groups. The ones who do not perform get transitioned out over time; the high performers are at risk of leaving to start their own firm or being hired away
"And then what you're left with is the stable middle, right? And that is the dilemma of a large firm"
The structural change underneath it is that pieces of an investor's job have been taken over by dedicated teams. Marketing is the clearest case — large funds now have marketing departments investors collaborate with, where once an investor's public profile was their own blog
He named Bill Gurley's blogging and Fred Wilson's writing as the earlier model
Exits have been institutionalized the same way. Lightspeed built a capital markets team to impose that discipline at scale, and an investor now comes into a Monday meeting to talk through a plan for exiting a position
That conversation now happens with a different set of people and on a much more regular cadence
5. The Capital Markets Team
Asked what such a team actually does, Agrawal gave two jobs.
The first is portfolio segmentation: taking the list of companies the fund owns and sorting it into buy, hold and sell. At a firm with thirty, forty or fifty investors and thousands of companies, he said simply reaching that level of clarity takes a great deal of work
The second is building the relationships that let an exit happen at all. Large funds now use continuation funds, general-partner secondaries, limited-partner secondaries and encouragement of mergers and acquisitions
Firms like Lightspeed are also doing buyouts, which makes them both buyer and seller in the same market. Keeping up with how other firms' strategies are changing is therefore continuous work rather than a deal-by-deal exercise
His summary of the problem is a question rather than a plan. "Saying that you want to manage liquidity and you want to provide liquidity to LPs sounds extremely obvious and sounds like everybody should be for that. But the question then becomes, who is buying these assets?"
6. Who Buys Private Stakes Now
Agrawal's answer to his own question is that the buyer base has widened well beyond the corporate acquirers and crossover funds of a decade ago.
Family offices are the first group. He said the family office world is far more switched on to alternative assets than it was, and that these buyers do not hesitate to make direct investments
His own fund backed a company rolling up small wealth management firms, part of whose offer is putting alternative products in front of a younger generation inheriting wealth
Sovereign wealth funds are the second, and the larger one. He sized the sovereign wealth market at about $15 trillion of capital, of which roughly a third is going into private markets including venture and buyouts, and about two-thirds of that third into direct investments
"By and large, sovereign wealth funds are extremely under allocated to venture"
Put another way, for every dollar they put into a private equity fund, several dollars go into buying stakes in companies directly
He said the Nordics have historically never invested in alternatives at all, which he offered as a measure of how much room is left
GIC and Temasek were the early adopters of the model. They would write large checks into funds such as Sequoia and then ask for direct investing and introductions into the follow-on rounds
7. Co-Invest Rarely Happens
Weisburd asked whether the constraint is infrastructure — whether these investors can process deals fast enough. Agrawal said that is exactly the constraint, and that it is a governance problem rather than a willingness problem.
The best companies raise on their own timetable and give almost no notice. He asked how much warning any shareholder of Anthropic gets before the next round, and answered it himself: very little, and even insiders hear things as the market hears them
A co-investment opportunity can require a decision in under five days on a meaningful amount of capital, which he said most institutional limited partners are simply not built for. Their underwriting process is slower and their investment committees typically meet once or twice a month
He named committee cadence as a marker of the strongest allocators. Meeting once or twice a month rather than quarterly or twice a year is, in his words, very unusual
His verdict on the whole practice is blunt. "Co invest is one of those things that 99% of LPs say they want to do and probably 10% do on a regular basis. I think 10% is an overstatement. I think it'll be like 2%"
He said his own young fund is actively willing to show its best positions to its investors as those companies come up for follow-on rounds, so the gap is not on the general partner's side
8. Family Offices Picked Off
The conversation turned to family offices making direct investments, and Agrawal was the most skeptical he was anywhere in the interview.
His argument is about deal selection, not fees. Outside the specific niches where a family already has an operating edge — his example is a family with a widget business looking at a widget deal — he said family offices are the most poorly selected buyers in the market, because the deals that reach them are the ones nobody else wanted
He said paying no fees is not a saving if the investment goes to zero, which is the point of a line he uses about fee structures: the only thing worse than paying a manager is paying nothing and losing the principal
He was careful to say families differ. A first-generation family where the entrepreneur who made the money is still allocating is in wealth-maximization mode; a third- or fourth-generation family is usually in preservation mode and taking risk at the margin
He applies the same standard to himself, and it is unusually strict. "My personal point of view is that it's really hard to deliver returns consistently"
"So much so that I have made zero investments outside Lightspeed or my new fund in the years that I've been in venture over the last decade. I have purchased zero public stocks"
He said people ask why he does not also trade, the way some doctors keep practicing, and his answer is that he does not know how to be good at both at once
9. VCs Become Asset Managers
Weisburd raised Lightspeed moving into buyouts and General Catalyst, which he said he thought was approaching $50 billion in assets, and asked where multi-stage firms are heading. Agrawal's answer was a direct analogy to private equity two decades ago.
The template is KKR, Blackstone and BlackRock. Each broke out on one part of the business, and other parts took root later and became larger than the original
"Blackstone was not a real estate private equity investor in the early days. Now it's their biggest business"
He expects the same in venture, with firms that became famous for one thing adding adjacent asset management businesses — more products, more regions, more scale, and in some cases public-market strategies alongside the private ones
He drew a line between different kinds of buyout. What a firm like Lightspeed aspires to do is, in his account, very different from what Summit or TA does, in the risks taken and the premiums paid
"And so I do think there is room for innovation. There is room for multiple strategies"
The reason this is happening, he said, is how asset managers are valued. Citing his mother, a certified public accountant, and her advice to follow the money, he said the value of an asset management business is tied disproportionately to recurring income and is much less correlated to performance, which was always hard for an investor to model
The paradox he raised is that a firm's right to win genuinely does improve with each vintage — relationships, brand and institutional know-how all compound — and that this is precisely what creates the trap
"The problem is that oftentimes their fund size outpaces their right to win because a lot of things do compound"
10. Strategy Drift
Weisburd pushed back, suggesting a growing franchise has real advantages that justify some natural growth rate. Agrawal agreed on the principle and then named the ways it goes wrong.
He accepts the compounding argument. "Venture is a compounding business. Any success that you have in the first fund, any learning, all compounds in the second fund"
The complication is that the people running funds are human. They have the same ambitions and insecurities as anyone, and do not want to be left behind, which pulls them away from the strategy they originally stated
Drift takes two shapes. One is fund size — going from raising $100 million to raising $300 million. The other is adding new areas of investment, such as a software fund that starts doing software and crypto, which he called very different animals
He agreed with the host's framing on the payoff for not drifting. Consistency of strategy means getting better at the craft, and better odds of success
11. The Markup Scoreboard
Asked which biases smart venture investors repeat, Agrawal gave two: where deals come from, and how investors are scored before any exit happens.
The first bias is network-sourced deal flow. An introduction from a founder you already back is high quality for two reasons — you trust the source, and the source is only incentivized to show you good things
The shadow side is that the founders who mattered most were outsiders. He named Airbnb's and Stripe's founders as people who began entirely outside those networks
Weisburd called it the Peter Thiel formulation, that Silicon Valley is outsiders becoming the ultimate insiders
"It's a bunch of outsiders and they've produced like some of the most iconic companies of the last generation"
He was careful not to treat bias as simply a defect. "Bias is a heuristic that helps cut through a lot of noise and focus on a few things that matter"
The second bias is the markup. Investors' careers move much faster than companies exit, so the interim measure of whether someone is doing well is whether their deals have been marked up by a well-known investor at a healthy price
He said he has no objection to markups themselves, but that a transient mark is not performance, and the trouble starts when the whole system optimizes around it
He walked through what that does to a junior investor. They can chase something already hot inside the firm, or go looking for the contrarian deal a standard deviation away that may never get done — and the incentive points one way
If every ambitious young investor at every large firm makes the same choice, the crowd forms and the returns get competed away
12. Momentum Up the Stack
Weisburd observed that the same dynamic runs all the way up the capital stack, and offered a figure from the show's own research.
The number was the host's, not the guest's. "The average CIO at a pension fund, six point one years. There's a famous study on this"
Agrawal extended it to the people who allocate to funds. Poorly performing allocators struggle or get fired, and the strongest ones get hired away by larger endowments
"And the highest performing LPs will actually get hired by larger endowments because the pay could go up three times"
Moving from a billion-dollar endowment to a $10 billion or $20 billion one can triple the salary, on his account
His conclusion is that the chasing is structural rather than personal. "So there's almost a systemic momentum trade within all of the private markets, not just venture capital"
He drew the practical distinction between stages from that. Early-stage venture requires finding pattern breakers and things that are not yet consensus; later-stage venture is about scale, efficiency of capital, and companies that have already been discovered
He offered his own firm as evidence the two are genuinely different businesses. His Lightspeed managing partners wrote the first check into his fund, which he read as an acknowledgment that this is a different end of the market, and he said his is the only fund he knows of backed by both of them
He also named the reflexive loop between capital and founders. When capital chases whatever is growing fastest or making the most noise, founders get pulled away from their own value drivers and start playing the same game, with each side accelerating the other
13. Chaos and Pattern Breaking
Weisburd noted that Agrawal's wife says he chases chaos, and asked why. The answer ran from a dissection hall in India to the best investment he made at Lightspeed.
They met as students, he in medicine and she in dentistry, and within medicine he ended up in the emergency room, which he called among the most chaotic and action-packed places in the field
He said venture is the equivalent within finance, and that having established himself at a large fund he wanted to do something else again
His working definition is not disorder but rapid change. "Chaos to me means change"
A lot of frenetic change, a lot of noise, and from a distance it is disorienting to work out what is actually happening — which is where he says the intellectual interest sits
The payoff line is the one he built the section around. "Pattern matching produces returns, but pattern breaking produces legendary returns"
His example is a 2018 investment that fit no category at the time. "The best investment I made at Lightspeed was a B2B company that leveraged software, finance services, and discovery to create a supply chain for custom manufacturing"
Everything about it was wrong for the moment: commerce meant consumer, it meant finished branded goods rather than custom manufacturing, and nobody was talking about supply chains. There was no thesis behind the investment
What they had instead was the quality of the founders, a large apparent market, and an agile business model. Eight years on, he says the company is a model for imitators across Europe, Latin America and the US
He used Union Square Ventures and crypto as the outside example. Crypto looked like a strange fringe thing around 2009 and 2010 — his own reaction, while building a clinic business, was to wonder what those people were smoking — and Union Square Ventures did the work, formed a view, published why they were investing, and produced what he called the most legendary investment of the time
His point about the firms that followed in 2019 to 2022 is about entry price: arriving after a trend is established means paying momentum multiples, funding high-burn companies competing with each other, and having no floor if the market does not cooperate
14. Consumer Looks Oversold
Asked where the market has thrown out something good along with the bad, Agrawal named two sectors and then talked about geography and the kind of founder he backs.
Consumer is his main answer. "Fast forward today, I would say across the valley, I do think consumers are oversold"
Ten years ago every firm on Sand Hill Road had at least one partner looking for the next Facebook or Snapchat; Lightspeed itself grew up as an enterprise infrastructure firm and then made investments in Snap and Affirm
His reasoning is cultural rather than financial: given how much preferences, ways of working and society itself have shifted, he finds it implausible that no new consumer paradigm emerges
He relayed two points from Niko Bonatsos, who he said spent fifteen years at General Catalyst: that consumer is the larger segment of the market by first principles, and that ChatGPT is itself a consumer product, with several of the most valuable private companies today being consumer companies
"And the other point that he made was that 50% of all of the market cap in an industry is made before it's actually named"
The consequence, on that view, is that raising a fund once a sector has a name means the first half of the returns is already gone — and the next half usually takes another twenty years
He partly disagreed with the rule about naming. He said it holds in spirit but not for every category, and pointed to fintech, where Ribbit was the poster child, as oversold today and a strong opportunity going forward
His general model is a cycle: an opportunity announces itself, gets overbought, then gets oversold, while the underlying opportunity is still there. That last phase is where he thinks money is made
On geography he expects new technology hubs in Germany, Bangalore, Riyadh and elsewhere in the Middle East, driven by these economies' push to diversify trading relationships. He singled out Germany's industrial base and the quality of its engineers
On founders, his fund's book is deliberately tilted. "Of the 17 investments we've made, 15 repeat founders, i.e., they've started something before, but they're not fully fulfilled yet"
Many were born outside the US and are only now setting up there — the "fresh off the boat" thesis in his words, which he clarified is more like fresh off one boat trip
He has been visiting the Valley himself since 2005, when he was a medical student shadowing a cardiovascular surgeon in the East Bay, and sees the same pattern in founders who keep coming back
The claim underneath the thesis is a pricing one: these founders are systematically mispriced because they sit outside an investor's local network. "Nobody's looking for them"
Agrawal's bottom line is that venture's returns now come from deliberately standing outside the network, the trend and the markup that everyone else is optimizing for — because the capital chasing the obvious opportunity has grown faster than the opportunity itself.
Bonus Insights
He argues strategy in venture cannot be written in advance. "I often think strategy in venture is emergent. It's built in the field, right?" — funds that were not raised for AI have been among the best performers, because the market opportunity does not care what the fund's deck said
His test for acting outside the stated mandate is the founder, not the sector. If another Sam Altman were starting a real estate company, he says the right answer is to try to be in business with them even if real estate is nowhere on the deck
He credits the industry's competitiveness, not any single cause, for shorter investor tenure. Asked why people move, his answer was simply the maturity of the industry
Weisburd translated the markup problem into trading language, calling it an overweight to the momentum factor. Agrawal accepted the frame and split venture investors into investors and traders, with momentum investing at its extreme being trading
Products, Companies & Tools Mentioned
Lightspeed Venture Partners (Where he spent close to a decade; his example throughout for how a large firm institutionalizes marketing, exits and now buyouts)
Anthropic and OpenAI (Named as the evidence that AI revenue is still concentrated in coding — and Anthropic as the example of a company whose next round nobody sees coming)
Cursor and Claude (His examples of AI already eating engineering labor, from research-level code down to basic systems implementation)
SoftBank (The Vision Fund as the marker of how far the scale of private capital has reset)
Andreessen Horowitz, Sequoia and Benchmark (Named alongside Lightspeed as firms now running in excess of $50 billion to $70 billion)
General Catalyst (Raised by Weisburd as approaching $50 billion in assets, the prompt for the whole multi-stage asset manager discussion; also Niko Bonatsos's firm for fifteen years)
Blackstone, KKR and BlackRock (The private equity template he expects venture to follow — each broke out on one business and grew a larger one beside it)
Summit Partners and TA Associates (Named to contrast the kind of buyout they do with what a venture firm like Lightspeed would attempt)
GIC and Temasek (Early adopters of writing large fund checks and then asking for direct co-investment alongside them)
Ribbit Capital (The poster child of fintech venture, in a sector he now calls oversold and a strong opportunity going forward)
Union Square Ventures (Did the work on crypto around 2009–2010, published why they were investing, and produced what he called the most legendary investment of the time)
Snap and Affirm (The iconic consumer investments Lightspeed made despite growing up as an enterprise infrastructure firm)
Airbnb and Stripe (His examples of founders who began as total outsiders to the venture network that now sources most deals)
ChatGPT (Bonatsos's argument that the most valuable private companies today include consumer products, and this is one)
Anduril and SpaceX (Defense as the worked example of an industry whose market cap was largely made before anyone named the sector)
TBPN (Its acquisition cited as evidence that venture sourcing has become a media business)
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: