D
+7
Cerebras: Demand and supply look ample—data-center capacity the binding constraint?
Earnings aren't the issue. The bottleneck is data centers.
+7Cerebras
D
Cerebras (Trans): Q4 guide raised across the board; OpenAI deliveries remain behind schedule
Q4 revenue guide raised to $310–330 mn (up nearly 40% QoQ), with core GPM jumping to 48%.
The 750MW OpenAI contract is delayed due to data center land and power constraints, and RPO is net of delay credits; catch-up is now expected in H2 2027.Meanwhile, Microsoft (MSFT) became the third hyperscaler customer, and the 2027 revenue target still calls for a 3x increase.D![ESE EiiW18823 cerebras 5 The World's Largest Chip 日 H 舒 点 WSE-3 器 Cerebras Wafer]()
![CIv GIG A]()
![WSE-3 Die Core 10.7k Cores 84 Die 900k Cores Source： Cerebras： ICerebras ONGPORT]()
![Processing element Waveletin Input queues 8colors， task triggers Router Compute]()
+27
Cerebras: Wafer-scale 'monster' chip—can it challenge NVDA?
Most chip designers, including NVIDIA, lay out circuits on a large wafer, then dice it into smaller dies. Those dies are interconnected to build AI accelerators, improving yield by avoiding scrapping an entire wafer due to defects.
One company refuses to follow that path. It treats the whole wafer as a single chip—a 21.5 cm x 21.5 cm square roughly matching the largest rectangle you can cut from a 12-inch wafer—packing 4 trillion transistors and 900k cores; the company is Cerebras. The bottleneck for GPU inference is the 'memory wall'...