Interview, Fireside Chat
Larry Aschebrook, Founder & MP @GSquared: How We Lost Money on Uber and Made Millions on Lyft
- The firm intends to pivot strategy and raise an additional $300 million immediately to address past errors involving overpayment and herd mentality, with plans to close the raise as soon as possible while avoiding the identity of a standard early-stage seed manager or crossover fund.
- Raising funds in the $1 billion to $2 billion range is deemed feasible based on current capital availability, with the expectation that building a portfolio and improving Net Asset Value will generate inertia for future fundraising.
- The firm plans to tighten deployment control internally to ensure clear attribution of decisions and avoid the capital dispersion associated with the "Silicon Valley mentality."
- A specific outcome is predicted for the 2022 vintage, where a "land and expand" strategy using small initial checks aims to generate 3x returns for founders within 18 months.
- For LPs backing back-to-back funds, the firm expects a 4x cash-on-cash return over 10 years, leveraging the optionality of their capital.
- The 2020 vintage is expected to maintain 70% primary exposure, with 40% of that amount structured with IRR hurdles and multiples embedded to protect against valuation bubbles.
- Historical analysis indicates that 2020 and 2021 featured irrational capital flows and public market multiples "off the charts," creating a difficult environment for achieving returns; the speaker notes the historic valuation floor cannot be lower than 10x, whereas the 2025 multiple is four.
- The speaker anticipates that the direct secondary market will offer significant value discounts due to a massive need for liquidity, with LPs seeking trusted partners to engage directly with founders.
- The 2021 environment is identified as the point where the ability to secure a 2.5x net return within five years ended if entry multiples were too high.
- The firm warns against chasing Multiple of Invested Capital (MoIC) over Distributed to Paid-In Capital (DPI), citing a 23andMe investment that resulted in a $70 million loss to LPs due to catastrophic losses from process failures.
- Investments in 23andMe and Theranos are attributed to bad processes, leading to a strict policy of only signing binding transactions when certain to avoid personal liability, while the 23andMe case specifically highlighted the risk of lost capital and investor attrition.
- The "Getir" investment is identified as a monumental miss where the second capital tranche cost the partnership approximately $500 million in lost capital and investor relations, despite the company having over 10,000 employees and reaching a $10 billion valuation.
- OpenAI is predicted to become a $1.5 trillion company within five years, representing a high-confidence 5x return, with the speaker stating they would invest the entire fund in it.
- OpenAI and Anthropic are identified as the sole winners in the LLM foundation model space, with no rivals expected to challenge them, prompting a strategy to invest in "picks and shovels" like Scale AI and Lambda.
- The LLM space is predicted to be challenging for new entrants due to high capital and time requirements, leaving limited room for additional foundation model companies.
- "Zombie" companies lacking AI integration are predicted to fail, while "vampire" companies are expected to survive the transition, with hundreds of zombies currently in the market.
- Liquidity is predicted to remain difficult, resulting in low DPI numbers, as exiting investments is described as harder than acquiring them.
- While Anthropic possesses some quasi-liquid private positions, moving $1 billion of value in the company is expected to be impossible in a short timeframe.
- Monzo is identified as a value play with potential for a 500% return relative to Revolut, which generated an "awesome return" in the 2018 vintage fund.
- The Whiz investment is characterized as a short position that took three years to build, a pace that would be impossible under a "hockey stick" scenario from day one.
- The firm plans to deploy more capital into winners in the AI space, specifically OpenAI, Anthropic, Databricks, and Wiz, with an appetite to buy all available shares at current valuations.
- OpenAI is predicted to have "escape velocity," with Microsoft executive Kevin Scott confirming no defensibility exists in search because users do not prefer Bing over Google.
- The firm expects to endure for many vintages, attributing its success to a focus on the firm's logo rather than the individual, which has broadened its recognition.
- The current fund life model is predicted to be fundamentally broken, driving the creation of evergreen structures, interval funds, quasi-liquid strategies, and continuation funds to address liquidity issues.
- Endowments are advised to ignore TVPI and MoIC as "fake numbers" and focus exclusively on DPI to ensure they can generate actual cash returns.
- The speaker's personal intensity is compared to Cal Ripken Jr.'s 16-season record of playing every major league game, though this approach is acknowledged to cause some high-performing staff to leave.
- A culture of "productive paranoia" is expected to continue driving daily operational focus and the avoidance of larger losses, similar to the protective actions taken during the Theranos incident.