Investment Review №345. Treasuries vs Stocks

On May 13, Cerebras Systems (CBRS) debuted on Nasdaq at $185/share and briefly approached ~$390 intraday—an almost 110% first-day surge and the largest U.S. tech IPO since Snowflake in 2020. The company raised $5.55bn at an implied valuation of roughly $56bn, with demand reportedly oversubscribed by ~20x. The scale of the move is particularly notable given the current backdrop of geopolitical uncertainty, tariff volatility and partial rollbacks, alongside concerns over inflation and its implications for GDP growth. However, to properly assess how successful the IPO truly was—and for whom—it is important to examine the company’s most recent private funding rounds.

In September 2025, Cerebras Systems closed its Series G round, raising $1.1bn at an $8.1bn valuation. Just five months later, in February 2026, the company completed a $1bn Series H round at a $23bn valuation—a ~2.8x step-up in valuation over a very short period. By the IPO, valuation had reached ~$56bn, implying another 143% increase in just three months. The Series H round was led by Tiger Global Management, with participation from Fidelity Investments, Coatue Management, Altimeter Capital, Benchmark, 1789 Capital, and Advanced Micro Devices (AMD). The breadth of the cap table, pace of funding rounds, and caliber of investors suggest that most major institutional AI allocators had already secured their exposure. As a result, IPO oversubscription appears to have been driven largely by retail demand, reflecting the broader desire among mass-market investors to gain exposure to AI-linked deals. Among institutional participants, AMD stands out in particular, as the investment is highly atypical for the company. Advanced Micro Devices Ventures has completed roughly 87 investments historically, primarily focused on companies either consuming AMD chips or operating within the company’s broader ecosystem. Cerebras, however, is a direct competitor, making this effectively AMD’s first strategic investment into a rival, a move that likely strengthened confidence among other investors even post-IPO.

One of the key risks highlighted in the S-1 filing was revenue concentration: 87% of Cerebras revenue in 1H24 came from UAE’s G42. This concentration issue was the main reason the company’s initial IPO attempt in October 2025 was withdrawn following a national security review by the Committee on Foreign Investment in the United States (CFIUS). Ahead of the current listing, Cerebras significantly diversified its revenue base via a reported $10bn contract with OpenAI. Still, having two customers account for 99% of total revenue would normally represent a major institutional red flag. In the current semiconductor environment, however, the market appears willing to overlook that concentration risk due to extraordinary AI infrastructure demand. Industry CapEx growth continues accelerating—for example, Anthropic reportedly expanded annualized revenue from ~$9bn to ~$30bn within a single quarter. The implication is straightforward: compute demand is arriving faster than supply can be built. Against that backdrop, Cerebras is already generating ~$238m in net income with ~47% margins, materially reinforcing the company’s fundamental positioning. 

Another major risk is the expiration of insider lock-ups around November 2026. Retail investors were effectively paying ~$350/share in early May trading for stock that Tiger Global acquired in February at an implied ~$76/share equivalent in Series H. The issue is not necessarily the quality of the asset itself, but the magnitude of valuation expansion in an extremely compressed timeframe, creating strong incentives for institutional holders to monetize gains. Notably, our investors had the opportunity to accumulate shares in the ~$220–250 range during pre-IPO trading, materially improving average entry levels and providing a more attractive risk/reward profile post-listing. Against the backdrop of extreme demand for AI-related IPOs, pre-IPO access is increasingly becoming a key competitive advantage for our retail investor base. To maximize what is likely a short-duration but highly attractive opportunity set, investors should closely monitor all of our AI-related transactions and allocations.