**WARNING** The Cerebras IPO [CBRS]

Key Concepts

• Wafer Scale Engine (WSE): Cerebras’s proprietary chip architecture that keeps the entire wafer intact rather than dicing it into individual chips, aiming to reduce latency and eliminate the need for complex cabling.
• SRAM (Static Random-Access Memory): High-speed memory used by Cerebras on-chip to improve processing efficiency.
• CFIUS (Committee on Foreign Investment in the United States): The regulatory body that delayed the IPO due to concerns over UAE-based investors and potential Chinese portfolio exposure.
• Contra Revenue: An accounting method where the value of warrants (given to partners like OpenAI) reduces reported GAAP revenue.
• Related Party Transactions: A significant portion of revenue derived from entities that are also major investors in the company (e.g., G42, UAE entities).
• Penny Warrants: Options allowing holders to purchase shares at a nominal price, creating significant potential dilution and financial gain for early partners.

---

1. Company Overview and IPO Details

• Ticker: CBRS (NASDAQ).
• Market Context: Cerebras is entering a highly competitive market dominated by Nvidia (GPUs), Marvell (ASICs), and AMD (Processors).
• Valuation: Expected market cap of approximately $48 billion.
• IPO Status: 20x oversubscribed as of May 10th, following a year-long delay due to CFIUS reviews regarding foreign ownership and Chinese portfolio divestment.

2. Technical Differentiation: The "Mousetrap"

• Traditional Approach (Nvidia/Others): Manufacturers create a large wafer, slice it into individual chips, and discard defective ones. The remaining functional chips are then connected via high-speed communication technologies like Infiniband or fiber optics.
• Cerebras Approach: The WSE keeps the entire wafer as one unit.
  • Redundancy: Instead of discarding defective sections, Cerebras uses software to route around them, claiming the defects are statistically insignificant.
  • Performance: By keeping memory and processing on a single piece of silicon, they claim to eliminate the latency and costs associated with external cabling.
  • Scale: The WSE is reported to be 29x the size of an Nvidia B200 and 57x the size of an H100.

3. Financial Analysis and Red Flags

• Profitability Concerns: While the company reports positive EPS, this is largely attributed to a $390 million one-time gain from the remeasurement of warrant liabilities. Without this, the company shows a growing loss from operations.
• Revenue Concentration: 86% of 2025 income originated from related parties (62% from Muhammad bin Zayed University of AI and 24% from G42).
• Balance Sheet: The company maintains a strong cash position (over $1.1 billion) and low debt, providing sufficient capital for R&D and operations.
• Future Volatility: The transition of warrants into "contra revenue" is expected to negatively impact future GAAP revenue reports.

4. The OpenAI Partnership and "Pump" Concerns

• The Deal: OpenAI provided a $1 billion working capital loan to Cerebras in exchange for 33.4 million "penny warrants."
• Circular Economics: OpenAI has committed to $20 billion in compute contracts with Cerebras.
• Conflict of Interest: Greg Brockman (President of OpenAI) holds a significant personal position in Cerebras. The speaker argues this structure creates a "pump and dump" scenario where OpenAI and its executives stand to gain billions in equity value through the IPO, while the revenue is heavily dependent on the very partners who stand to profit from the stock's appreciation.

5. Synthesis and Conclusion

Cerebras presents a compelling technological innovation with its Wafer Scale Engine, potentially offering a faster alternative to traditional GPU clusters. However, the company’s path to public markets is clouded by significant red flags:

1. Artificial Profitability: The current positive EPS is a result of accounting adjustments rather than operational success.
2. Related Party Dependency: The reliance on investors (UAE/G42) and partners (OpenAI) for the vast majority of revenue raises questions about the commercial viability of the product in the broader, non-affiliated market.
3. Governance/Incentives: The concentration of voting power (99.2% in Class B shares) and the circular nature of the OpenAI contracts suggest that the IPO may be designed to benefit early insiders rather than reflect organic market demand.

Takeaway: While the stock may experience significant momentum-driven price action at the open, investors should be wary of the underlying financial structure and the potential for volatility once the "one-time" accounting gains fade and the true operational costs become apparent in future earnings reports.