Larry Aschebrook, Founder & MP @GSquared: How We Lost Money on Uber and Made Millions on Lyft
Summary
- Larry Aschebrook built G Squared around a five-to-seven-year liquidity strategy: convert late-stage private-company access into cash DPI. The portfolio “lands” with small checks, expands through repeated transactions, and concentrates 80%—hopefully 90%—of risk in about 10 companies. He calls TVPI and MOIC “fake numbers”; DPI “is the only thing you can use to buy food.”
- Spotify validated the model and changed G Squared’s scale permanently. After six straight days of being denied a meeting in Stockholm, Aschebrook saw 25% Swedish penetration and learned the record labels were shareholders. Spotify then offered $150 million of stock, forcing him to assemble the money in 60 days and borrow the final $9 million from an early investor. G Squared ultimately put 40% of its $380 million third fund plus roughly $700 million of co-investment into Spotify, producing about $1 billion for LPs.
- The realized record came from selling into demand, not perfectly forecasting eventual winners. G Squared owned 16% of Coursera, returned $800 million to LPs after selling around $36, and watched the stock later trade near $8; Lyft returned roughly 3x while Uber lost about 20 cents on the dollar, or approximately $50 million. “We made our multiple and went home and distributed the cash.”
- The 2020 vintage broke when G Squared mistook a booming market and prior liquidity for proof that its judgment could not miss. It deployed about $900 million from early COVID through 2021, treated 12x LTM ARR-to-enterprise-value pricing for SaaS as conservative against public multiples near 25x, and later saw the comparable multiple fall toward 4x. Toast at $76 became Aschebrook’s “canary in the coal mine”: “We’ve overpaid for all of it.”
- The rescue required admitting the error, raising another $300 million and buying protection while markets burned. G Squared sold inflated positions, lowered cost bases through secondaries, and negotiated structured equity paying a 25% IRR or 2.5x, whichever was greater; about 70% of the vintage became primary exposure, with structure on 40% of that. “As the house is on fire, we’re running in the front door with cash.”
- His worst mistakes separate bad underwriting from bad behavior after underwriting. Theranos cost him personally a few million dollars to escape a binding agreement to buy roughly $50 million of stock; 23andMe could have produced about a 2x return when he began selling, but he chased a larger multiple and later said the position lost about $70 million. At Getir, the damaging decision was investing another $100 million rather than accepting the first loss: “You were dead man walking without knowing it.”
- Co-investment now amplifies only positions the fund itself has already designated as core. Earlier vintages used co-investment at up to four times fund capital, and the 2020 vehicle even accommodated LP-requested thematic one-offs; Aschebrook said those experiences taught him that investors may blame the manager when a single-shot position fails. The current model restricts co-investment to conviction names such as Anthropic, Fanatics, Wiz, Databricks, Turo and Monzo.
- On AI, Aschebrook would pay for leaders that have reached “escape velocity” rather than hunt for another foundation-model entrant. He sees little room beyond OpenAI and Anthropic, would buy Anthropic around the stated $61 billion valuation “all day long and twice on Sunday,” and agreed that it is an extraordinary business after Harry described a $350 billion-to-$1.5 trillion OpenAI scenario over five years. The hedge is picks-and-shovels exposure such as Lambda and Scale AI—and avoiding legacy companies that cannot rebuild AI into their DNA.
Deep dive
1. A $50,000 household stake became a private-market thesis
Aschebrook began as an academic fundraiser, watching donors create family wealth through private companies and deciding, “They’re not that much different than me.” He returned to business school late, made the emerging smartphone economy his thesis and started asking classmates whether he could buy their Twitter, Uber or Spotify shares.
The capital was genuinely scarce: he cashed out retirement savings after a divorce, paid the tax, and started again with the $50,000 his new wife brought into the marriage. “Half of nothing is nothing,” he said of risking what they had on Twitter and Alibaba shares, including stock sourced from Jack Ma’s family office.
His advantage was partly “ignorance is bliss.” Not knowing private shares were considered difficult to transfer, he created a binding one-page purchase form at Arizona State; 15 years later, a broker sent him back the literal form he had created in 2010. Its binding nature later cost him dearly when his instinct changed after signing.
Raising the first fund took three years, from 2010 to 2013, and produced about $34 million or $35 million through multiple closes. He invested as money arrived—a practice he still favors for emerging managers: close available capital, start deploying it and build a differentiated model rather than waiting for a perfect final close.
2. Scarcity, discounts and concentration drove the early portfolio
The structural thesis was that fewer institutions helping companies go public, combined with larger private funding rounds, would extend company lifetimes. Average inception-to-IPO time moved from roughly three years before 2010 to seven or eight by 2018; G Squared’s portfolio companies now average about 15 years old, creating persistent liquidity needs among employees and early shareholders.
Alibaba’s 2014 IPO was the first material confirmation. With almost no internal staff, Aschebrook had outsourced early investment-committee research to analysts in India, combining public information and personal contacts into “mosaic theory.” The outcome convinced him there was a business rather than merely a lucky personal trade.
The first vehicle held only about seven companies, with most capital in Alibaba, Spotify, Palantir and Twitter. There were failures too, including cleantech investments that established venture firms put into the syndicate while contributing little themselves and directing substantial LP co-investment toward the deals. That imbalance became a lasting alarm bell.
What marketing later named “land and expand” began from necessity: make a small purchase, obtain better data, then concentrate into the few companies proving strongest. Early secondary purchases could come at roughly 35 cents on the dollar versus primary buyers; today, 10 companies are intended to represent 80%—hopefully 90%—of portfolio risk.
3. Micro-secondaries are an information system, not an index
Harry challenged whether a late-stage secondary investor is detached from founders and operating data. Aschebrook argued the opposite: companies with large liquidity needs value a trusted buyer operating under the required regulatory structure, so G Squared receives primary-level information and founder access rather than merely arbitraging anonymous blocks.
Transactions below $2 million—even inside a $2 billion fund—create frequent touchpoints and interim data. The first small purchase acts as a “Trojan horse”: conviction begins with mosaic research, but access lets the team decide whether to triple down, as it did in Wiz before the outcome Aschebrook said they expect.
The operational burden is the moat. Some $75 million positions required about 50 transactions; others reached $200 million in four. G Squared consolidates cap tables, runs employee or shareholder tenders and buys awkward departing-holder stakes—the unglamorous repetition that another manager cannot reproduce merely by announcing a secondary strategy.
4. Spotify turned a cold trip to Stockholm into a billion-dollar outcome
Spencer Mlot, then a young Berkeley graduate, heard Aschebrook reminiscing about downloading Metallica through Napster and replied, “You’ve heard of Spotify, right?” They found an initial $4 million block from a celebrity going through a divorce, then flew to Stockholm without an appointment and spent a week pursuing company approval. Spotify declined them for six straight days before a young lawyer agreed to meet.
Once Spotify met them, two facts transformed the underwriting: roughly 25% of Sweden already used the service, and “the record labels were also investors.” With Apple Music gaining share and artists including Adele and Taylor Swift pulling music, the labels’ decision to exercise options for more stock became Aschebrook’s signal to buy from frightened sellers.
Spotify offered approximately $150 million of stock and gave them around 60 days to close. By day 59 they had assembled $141 million; lacking the final $9 million personally, Aschebrook borrowed it from the early investor to whom they had sent the money. His wife’s response was practical: if he loved the business that much, they had to determine how to own the exposure.
G Squared kept buying at about a 50% discount to the current financing for the next two years, even posting “we’ll buy your shares” signs in Spotify’s break room. Roughly 40% of the $380 million third fund and another $700 million of co-investment went into Spotify; the firm became a top-10 global shareholder and generated about $1 billion for LPs.
5. Exit timing made Lyft a 3x and Uber a loss
G Squared sometimes backs two companies attacking the same market. Lyft was the value trade against Uber: the firm sold most shares privately before listing, often to brand-name investors following one another into the position, and realized about 3x rather than waiting to see which ride-hailing narrative ultimately dominated.
Uber produced the opposite result. Its IPO arrived during a difficult period, and because G Squared rarely holds companies after listing, the converted public price left the fund down roughly 20 cents on the dollar—about $50 million—even though Aschebrook believes today’s gross-profit valuation framework fits the business better.
Coursera showed the benefit of concentration plus timely distribution. G Squared owned about 16% at listing, made roughly 3x and returned $800 million to LPs by selling around $36; Aschebrook contrasted that realized outcome with the roughly $8 price he cited during the conversation.
The same discipline appeared in smaller stories: Postmates returned 3x in 18 months, while Instacart was sold privately in “the $20s,” with the transcript not specifying the unit, for roughly 3x. Aschebrook jokingly traced the latter decision to repeated bad grocery deliveries—“try to get a ripe avocado from your Uber driver”—but the governing mandate was capital velocity.
6. DPI discipline banked returns but surrendered Palantir’s upside
G Squared’s LPs hired it for a five-to-seven-year liquidity strategy, not an unconstrained 10-year ownership horizon. Aschebrook has now used a fund-extension lever for the first time, and said it “really guts me,” because changing the horizon may improve an individual company’s return while violating the product investors selected.
Palantir is the painful counterexample. G Squared sold it at roughly 3x, at about $9 a share; Aschebrook estimated it later traded around $80-$90 and joked that holding would have put the podcast “in my bubble in outer space.” One early LP retained the distributed shares and funded every later G Squared vintage from an original $50,000 commitment.
Harry’s pushback—worth keeping—was that guardrails may become a negative constraint when outcome sizes expand dramatically. Aschebrook conceded the forfeited upside but defended the mandate: two back-to-back funds returning an aggregate 4x cash-on-cash over 10 years provide LPs exceptional optionality to redeploy elsewhere.
G Squared stopped distributing shares because LPs may blame the manager for losses they incur after choosing to hold. “When your North Star is a DPI figure, there’s no hiding”; to Aschebrook, TVPI and MOIC are “fake numbers,” while DPI “is the only thing you can use to buy food.”
7. The 2021 failure began with believing the prior returns
Entering the boom, the 2018 vintage had already approached 1x DPI through Airbnb, Coursera, privately sold SpaceX, Impossible Foods and other winners. Raising became easy: in 2021 G Squared assembled roughly $1.4 billion quickly and turned down another $700 million. Success encouraged the firm to believe “our own [expletive].”
Toast at $76 was the canary. Looking at the stock from a Montana ski lift, Aschebrook concluded that an excellent company could not justify that public price—and therefore that the roughly $900 million G Squared had deployed from early COVID through 2021 had probably overpaid across the board.
The quantitative error was stark: valuing SaaS at roughly 12x LTM ARR to enterprise value seemed conservative while public comparables traded near 25x, and the team treated 10x as a historical floor. By 2025, he said the multiple was closer to 4x. At late stage, paying $3 billion instead of $2.5 billion can erase a targeted 2.5x net return.
The organizational error was replacing the co-PM model with a traditional distributed venture team, allowing gut feel, syndicate membership and individual attribution to influence deployment. Aschebrook even hired a Silicon Valley coach and later saw that as evidence he had absorbed a culture prioritizing lifestyle, prestigious dinners and logo-dropping over LP outcomes.
8. Another $300 million and structured equity became the rescue
Aschebrook returned to LPs with an unusually direct message: “We [expletive] up. We need to pivot,” and asked for another $300 million to protect the vintage, ultimately taking it from roughly $1.2 billion to $1.5 billion. The additional capital addressed pay-to-play pressure, supported secondaries and financed structured rounds during the collapse.
G Squared negotiated minimum-return paper alongside investors such as Lightspeed, Dragoneer and DST. A company might retain an $8 billion headline valuation, but the new equity required a 25% IRR or 2.5x, whichever was greater; about 70% of the vintage became primary exposure, with structure on roughly 40% of that.
The mechanism could devastate earlier preference holders because the last-money ratchet accumulated value every day. For G Squared, however, it was protection: “As the market’s falling and the house is on fire, we’re running in the front door with cash,” while simultaneously selling weak positions and lowering cost bases through secondaries.
His replay would be simpler: sit on his hands rather than join a $700 million round for a company with only $10 million of ARR, keep deployment authority with Spencer and himself, and build multiple strategy levers before a crisis. When everything had worked, G Squared had mistaken the absence of needed levers for proof they were unnecessary.
9. Theranos converted a bad process into an expensive escape
G Squared signed its binding one-page form to acquire about $50 million of Theranos stock over four months, attracted by a bulk discount to the last round. A management meeting triggered Aschebrook’s “spider sense”; his epidemiologist wife had told him that collecting enough data from blood and saliva samples would not work.
He tore up the agreement, faced a threat of litigation and personally paid a couple of million dollars to settle, ensuring LPs did not bear the cost. It hurt when he had little personal wealth, but avoided what he and Harry described as a potentially catastrophic fund and brand loss as damaging information emerged.
Aschebrook’s distinction was “bad process and bad outcome”: the escape was fortunate, but signing before completing conviction was indefensible. The incident added formal checklist items and taught him that neither a prestigious herd nor the prospect of instant gains from a discount substitutes for diligence before a binding commitment.
10. 23andMe proved that selling needs its own guardrails
G Squared first invested in 23andMe in 2017, Aschebrook said, believed deeply in its consumer model and built a large position. He later described total exposure as probably $100 million, while earlier describing roughly $50 million in the 2018 vintage including co-investment.
The Branson-associated SPAC traded at $10 and went up. Aschebrook said G Squared could begin selling at “seven,” which he described as about a 2x return, but he held on and said he sold the last share “in the 70s.” He characterized that as chasing the multiple and later confirmed the host’s estimate that the position lost about $70 million.
The resulting guardrail is to “dollar-cost average out just like you dollar-cost average in.” Liquidity should begin privately and continue methodically through listing; waiting for a perfect terminal price turns a viable realized return into an exposed public-market bet that the fund was never designed to hold.
11. Getir showed how fighting harder can compound a loss
The original Gorillas underwriting had a credible process: G Squared had made money across Instacart, Postmates and food delivery, and Gorillas was growing rapidly in Berlin. Its acquisition by Getir gave a small group of Gorillas shareholders favorable preferred equity and some cash. The host framed Getir’s valuation at that point as $10 billion; Larry did not independently state that valuation in his answer.
About $50 million of initial exposure was followed by another roughly $100 million to restructure and pull the equity forward. During the exchange, the host described roughly $200 million at risk including co-investment; Aschebrook’s own figures were the $50 million initial total and the subsequent $100 million check.
Aschebrook’s identified mistake was the second tranche: he should have accepted the first loss rather than assuming more capital and board involvement could will a recovery. Harry argued that the premature global scaling and broken economics looked obvious from outside. Aschebrook’s concession was that hindsight made it clearer, but his poverty-conditioned instinct to keep fighting had become a curse: “You were dead man walking without knowing it. You were already dead. You just kept fighting.”
He estimated the episode cost the partnership roughly half a billion dollars across lost capital and LPs unlikely to return. He remains on the board because Getir still employs more than 10,000 people in Turkey. Harry advocated overcommunicating with LPs during a crisis; Aschebrook did not specifically present that as his own stated response in this exchange.
12. Co-investment works only when it amplifies fund conviction
Early G Squared vehicles sometimes deployed four times as much co-investment as fund capital. Scale was essential: when Spotify offered $150 million, saying no would have sent the transaction elsewhere and closed the relationship. A secondary manager unable to write $100 million when needed cannot remain relevant to companies managing liquidity.
Harry’s “tourist” challenge drew a useful definition: G Squared is a “point-in-time problem solver,” or “the janitor that’s cleaning up the mess.” It buys shares from departing employees, consolidates cap tables and solves structured tender problems, but does not promise indefinite ownership after that specific job is complete.
The 2020 mistake was extending co-investment to LP-requested thematic deals outside core positions. Aschebrook said that even full disclosure of concerns did not prevent investors from blaming the manager when a single-shot position failed. He now runs co-investment only when the fund is investing alongside it with conviction.
The $1.2 billion 2022 vintage limited co-investment to its top 10 positions. Aschebrook said that, on a $1.5 billion fund, co-investment would be about $700 million. The bespoke portfolios are equal-weighted extensions of names such as Anthropic, Fanatics, Wiz, Databricks, Turo and Monzo—not an LP menu of whatever happens to be available.
13. Portfolio construction pairs momentum with grounded scale
G Squared balances roughly 10 conviction positions across SaaS, fintech, consumer internet and mobility. For every high-velocity Anthropic, Aschebrook wants a scaled business such as Fanatics, with billions of revenue and hundreds of millions in EBITDA, because “you can’t just play the momentum” without eventually being caught at the cycle’s end.
The fintech chain ran from SoFi to N26 and Revolut in the 2018 vintage, then Monzo below a $4 billion valuation. Harry framed Monzo as a value play against Revolut; Aschebrook separately described Revolut as a generational business and said Monzo entered in a later vintage. Chime is in the current fund. Fast appreciation can be a problem: dollar-cost averaging cannot build a meaningful position when a company travels “from zero to 100” immediately.
Bolt came through Johan Bjurquist, the Spotify executive who had validated G Squared’s model and later said he would become CFO of the Estonian ride-hailing company. The thesis was to enter markets Uber avoided and operate profitably because Bolt lacked abundant capital. Aschebrook said early investments in Spotify, Wiz and Bolt were “probably 10x,” but later estimated Bolt at roughly 6x-7x and said its ultimate outcome was still to be determined.
Wiz was buildable only because the downturn briefly constrained even elite companies. A proposed $3 million opening check became $9 million after the founders made room, then repeated purchases created a roughly $200 million position. The lesson was not to predict the hockey stick, but to use temporary capital scarcity to accumulate into one.
14. In AI, paying for escape velocity beats searching for a third winner
Aschebrook sees little room for another foundation-model company to reach OpenAI or Anthropic scale within three years because both time and capital have become barriers. His answer to “How do you play AI?” was categorical: “Go to the leaders. Go to the winners.”
Harry said he would put an entire fund into OpenAI at the stated $350 billion valuation and could see $1.5 trillion within five years. Aschebrook agreed that it was an amazing business. At Anthropic’s cited $61 billion valuation, he said he would buy available shares “all day long and twice on Sunday.”
Dilution does not disturb that thesis: “I care about the price I pay in dollars and the price I’m going to sell it at in dollars. I am focused on DPI.” G Squared acquired a large position in Anthropic through the FTX bankruptcy process and remained eager for more in Anthropic, OpenAI, Databricks and Wiz.
The surrounding portfolio uses picks and shovels—Lambda and CoreWeave—and includes Scale AI and software intended to help companies scale on the hyperscaler side. Aschebrook also expects AI throughout cybersecurity, SaaS, fintech and consumer businesses. Harry supplied the “vampires and zombies” framing for large legacy private companies: vampires may retrofit AI and survive; zombies are already dead, and Harry said there are more zombies than vampires. Aschebrook agreed that the idea was strong and referred to long-in-the-tooth companies from older vintages.
15. Private markets’ hidden crisis is the difficulty of getting out
LPs underestimate how hard liquidity is, Aschebrook argued—possibly harder than entering the best deals. A handful of companies are quasi-liquid and small blocks may trade readily, but moving $1 billion of Anthropic is a different problem; Harry emphasized that the positions managers most want to sell are precisely those nobody wants to buy.
Harry’s instinct was to retain every winner he could sell tomorrow. Aschebrook offered the uncomfortable inverse: a sustainable finite-life fund must be willing to sell winners, because realizations drive future vehicles and give LPs optionality. Maximizing each company’s MOIC can destroy the liquidity proposition of the portfolio.
Evergreen and interval funds, quasi-liquid private strategies and continuation vehicles are attempts to repair what Aschebrook called a “fundamentally broken” fund-life model. He finds continuation funds more credible now that buyers focus on a manager’s good assets rather than accepting a mixed pool, but still tells endowments to evaluate DPI rather than paper TVPI or MOIC.
16. A durable firm needs the logo, the edge and more voices
Aschebrook opposes individual deal attribution because people magnify wins and deflect losses. “The logo makes the investment” is meant to preserve G Squared through leadership transitions; his aspiration is a firm remembered across generations, invoking Jim Simons’ roughly 40% annualized record as the standard for institutional endurance.
Wealth changed ease, not purpose. Growing up, he imagined $5,000 a month and a farm would mean he was set; now he says, “Money doesn’t make me happy. Money makes my life easier,” while also making life more complex. He endorsed resisting lifestyle inflation and remembering powdered milk, government cheese and the possibility of losing everything again.
The edge is neither solely escaping poverty nor creating a family monument: “The chase of the next win” supplies satisfaction, while every loss remains vivid. Productive paranoia helped him exit Theranos and still tells him to close and invest capital when available because “the wheels are going to fall off”; its danger is fighting merely because fighting became identity.
His model is Cal Ripken Jr.’s 16 straight seasons without missing a Major League Baseball game: a grinder’s longevity and willingness to keep showing up. Aschebrook admitted relentless delivery has cost relationships with people he wishes had stayed; the leadership task is to listen to more voices and soften the grind without surrendering the edge.