Opportuna Newsletter #4

🚀 Successful Launch

Our launch generated positive media interest, across a number of titles bolstering our confidence in the opportunities ahead. We have enjoyed engaging in interviews, with more to come. Ryan Hibbison from Venture Capital Journal featured us here , as did Secondaries Investor.

If you are a founder, executive, BA, or GP seeking liquidity solutions, please reach out for a confidential chat. For those exploring private market exposure with faster liquidity and downside protection, book a call with us.

🚨 Highlights of the Month

• A good long article on Plaid’s strategy. The company bets on credit-risk analytics, fraud prevention and pay-by-bank to revitalize growth. It is, reportedly, growing 20% in 2024 to $360m in Rev at 80% Gross Margin, for a loss of $50m or less. It has $140m in the bank and plans an IPO in 2026. Caplight estimates at $3.8bn imply 10x Sales, which is at the top-end of Fintech public peers, growing above 15% with 80% Gross Margins.

• Nasdaq’s Chief Economist sees the IPO extending to 2025. Falling interest rates positively affect valuations, and margins, since 40% of small caps’ debt is floating – compared to just 7% for large caps.

• Generative AI’s Act o1 from Sequoia’s Sonya Huang and Pat Grady’s is worth reading to get a glimpse of how one of the most influential VC firm sees the ecosystem evolving. The authors argue OpenAI’s o1 model exhibits true general reasoning capabilities through inference-time compute. As a result, it moves us beyond simply generating text or code based on patterns in data to developing AI systems capable of reasoning and problem-solving at inference time. Agentic reasoning will allow software companies to go after the services market, measured in trillions of dollars.

• Pitchbook released its Global M&A Report Q3 24. They see a modest M&A recovery compared to the tech bubble and GFC, which experienced peak-to-trough declines of 60-70% during the 2001-2002 and 2007-2008 downturns, vs 34.7% in 2022-2023. The current recovery started in Q4 2023 and has continued to gain momentum.

  • Private Equity Activity began catching up in Q2 2024, with buyout volumes increasing by 24% in value and 10.4% in count year-over-year, after it initially lagged behind corporate M&A due to high borrowing costs. PE buyers are now actively participating in deals, demonstrated by their prominent presence in top deals across sectors like IT (60%), B2C (60% in Q3 compared to 50% in Q2 and 0% in Q4 2023), and North American M&A (60% in Q3 compared to 50% in Q2 and 0% in Q4 2023).

  • Lower borrowing costs have been a crucial driver of the M&A rebound, even if modest. All-in borrowing rates on US leveraged loans for B-rated issuers in the BSL market decreased from a peak of 11% in May 2023 to 9.7% in September 2024. A similar trend played out in the Eurozone, with borrowing costs declining from 9.1% in June 2023 to 8.4% a year later.

  • Outlook. With another 100-125bps of rate cuts over the next 12 months, reduced borrowing costs should support M&A volumes for at least another year. Were private valuations to catch up with public ones, it could significantly boost M&A dealmaking.

  • Finally, the IT sector is leading the M&A recovery, with a 60.4% year-over-year increase in deal value in Q3 2024. PE buyers play a significant role in this sector.

📈 Chart of the Month: One-Year Returns by Market Cap Buckets in US Tech Stocks

This month, we looked into how US tech stocks performed depending on their market capitalization from small to large (1 to 5). Median return (bar) in the smallest (1) and mid (3) buckets have matched those of the top bucket (5). This is encouraging for exits. We dive deeper into this in the next section.

Note: Central bars represent the median return in each market cap buckets; the red dots are the simple average; the blue boxes mark the 25th to 75th percentiles and the ranges show the 10th to 90th percentiles.

🧭LT: Investing in GenAI Infrastructure vs Application

The disappointing returns on enterprise investments in generative AI (GenAI) applications have raised questions about the durability of the GenAI spending surge. Investors and decision-makers should distinguish between infrastructure investments—such as GPUs, AI servers, and large language models (LLMs)—and application-level investments, including GenAI-powered apps and agents. While infrastructure spending is expected to remain robust, application-layer investments will take time to yield returns.

GenAI adoption unfolds in three phases:

1. Initial Excitement - marked by increased spending on infrastructure and early application development;

2. Consolidation - as investment concentrates among a few dominant players;

3. Broad Adoption - where applications drive substantial revenue growth.

The market currently lies between phases one and two.

Supporting LLMs requires extensive capital. For instance, OpenAI’s annual costs reportedly reach $8.5 billion, with $3 billion allocated to model training and $4 billion to Microsoft server rentals. Their recent $10.6 billion funding round—$6.6 billion in equity and $4 billion in debt—illustrates the high cash burn rate, adding pressure on other LLM developers to keep pace. Anthropic, which raised $4.5 billion over the past 13 months, is already planning its next round, highlighting the intensive training demands and the critical role of funding access.

The backing of tech giants—hyperscalers like Amazon, Microsoft, Meta, Google, and Nvidia—offers significant advantages beyond computing power.

First, it provides a strategic exit path. Hyperscalers face competitive pressure to maintain control over LLMs as a key part of the tech stack and are likely to keep investing heavily until viable alternatives emerge. Additionally, companies like Microsoft, Amazon, and Nvidia gain economically from partnerships with LLM developers, with expenses like OpenAI’s $4 billion annual server rental directly benefiting Microsoft. At Anthropic, 25–50% of revenue is driven by Amazon reselling its Claude models to cloud clients.

Moreover, hyperscalers and Nvidia generate substantial free cash flow (FCF) to support LLM initiatives. According to Koyfin, these firms are projected to produce $330 billion in FCF over the next year, making annual funding of $10 billion for LLM development a manageable expenditure. The comments around increased capital spend from last week’s hyperscalers reports further make the point.

In summary, GenAI infrastructure spending is likely to remain strong among well-capitalized players, although consolidation around a few dominant entities is anticipated.

📌 Conclusion

In this edition we have tried to highlight key shifts in tech investing: a broader rally boosting exit prospects, steady gains in M&A supported by lower borrowing costs, and sustained investment in GenAI infrastructure despite slower returns from applications.

As these trends unfold, they signal both immediate opportunities in exits and M&A, and longer-term potential in AI-driven infrastructure. We will continue tracking these developments in upcoming issues to keep you informed on the evolving investment landscape.

In the meantime, we would love to hear your feedback and ideas.