![**WARNING** The Cerebras IPO [CRBS] | SCAM?!?!](/_next/image?url=https%3A%2F%2Fimg.youtube.com%2Fvi%2FKK_ccR1T8U4%2Fhqdefault.jpg&w=1080&q=75)
**WARNING** The Cerebras IPO [CRBS] | SCAM?!?!
Audio Summary
AI Summary
Cerebrus is set to go public on the NASDAQ under the ticker symbol CBRS on May 14th, with anticipation for a significant IPO, indicated by being 20 times oversubscribed as of May 10th. The IPO experienced a year-long delay due to scrutiny of its United Arab Emirates (UAE) investors and potential divestment of its Chinese holdings.
Cerebrus positions itself as a provider of the fastest artificial intelligence data centers, with its core innovation being the "wafer scale engine." This is presented as a novel approach to chip manufacturing, aiming to compete with established players like Nvidia (GPUs), AMD and Intel (CPUs), and Marvell (ASICs). Unlike traditional chip manufacturing, where large wafers are cut into smaller, individual chips, with defects leading to lower yields for higher-quality chips (e.g., 4090 vs. 4080), Cerebrus proposes leaving the wafer intact. This "wafer scale engine" integrates all necessary components, including memory, directly onto the entire wafer. The purported benefit is eliminating the need for extensive cabling, which connects individual chips in traditional systems and introduces latency. Cerebrus claims its software can manage defects by rerouting or ignoring faulty sections, rendering them insignificant.
This wafer-scale approach results in significantly larger chips. For instance, the Cerebrus wafer scale engine is stated to be 29 times the size of Nvidia's B200 Blackwell chip and 57 times the size of the H100. The Cerebrus chip also boasts 44 GB of SRAM directly on-chip, utilizing a 5-nanometer production process, while the Nvidia B200 has 126 MB of on-chip SRAM and additional off-chip memory that requires cabling. Cerebrus asserts that its architecture is capable of outperforming Nvidia's chips.
Financially, Cerebrus shows growth in hardware and cloud sales, with total growth averaging 75% and costs rising by 62%, suggesting pricing power. However, a significant portion of its net income appears to be driven by "other income," particularly a $390 million gain from the remeasurement of warrant and forward contract liabilities. This gain is described as a one-time event that artificially inflates its earnings per share (EPS), masking growing operating losses due to high research and development expenses. The company's balance sheet is strong, with minimal debt and over $1.1 billion in cash and investments, with the IPO expected to raise further capital. Despite negative cash flow, the substantial cash reserves and anticipated IPO proceeds alleviate immediate liquidity concerns.
A key concern for investors is the sustainability of Cerebrus's profitability. The positive EPS is attributed to the one-time remeasurement gain. Future earnings are expected to become negative once this gain disappears and due to the impact of warrants issued to OpenAI. OpenAI received a $1 billion working capital loan from Cerebrus at 6% interest and was granted 33.4 million penny warrants for Cerebrus's Class N shares, convertible to common stock. These warrants, estimated to be worth around $5.8 billion if Cerebrus stock reaches $175 per share, will be recognized as a contra-revenue item upon vesting, further reducing reported GAAP revenue.
The company's voting structure is concentrated, with Class B shares holding 99.2% of the voting power, primarily held by insiders. Post-IPO, insiders are expected to control 50.9% of the voting power.
Revenue concentration is another red flag. In 2025, 86% of Cerebrus's income came from related parties: Customer A (62%) and Customer B (24%). These customers are identified as UAE entities, specifically the Mohammed bin Zayed University of Artificial Intelligence and G42. This means a substantial portion of their business comes from entities that are also major investors and stand to benefit from the IPO's success. The $20 billion compute contract with OpenAI, which began in April 2026, also represents a significant revenue stream. Notably, Greg Brockman, President of OpenAI and a significant Cerebrus shareholder, is involved in establishing these revenue streams. This interconnectedness between investors, major customers, and the company's leadership raises concerns about a potential "pump and dump" scenario, where the stock is artificially inflated to benefit insiders and related parties before a subsequent decline.
While the product itself is innovative, its commercial viability beyond these related parties remains unproven. The initial public earnings report in August may provide more clarity with potential revenue from entities like Amazon, with whom Cerebrus has a letter of intent. However, the current landscape presents numerous red flags, though the potential for short-term trading gains due to momentum is acknowledged.