How would you design a chip for fast LLM inference?
Our Hardware Deep Dive with @alyciazcary follows the choices behind CPUs, GPU parallelism, and TPU systolic arrays, then explains how wafer scale works: weights stay in SRAM beside compute, reducing data movement during token generation.
0:00 What are the fundamentals of a communication system?
0:34 Semiconductor fundamentals
1:49 SRAM vs DRAM
2:48 What do CPUs struggle with?
3:34 What made GPUs uniquely positioned for graphics?
5:16 Training vs inference workloads
6:35 TPU’s systolic arrays
8:35 Why was wafer scale hard to build?
9:47 How did Cerebras solve the yield problem?
11:28 Reducing data movement
$CBRS
• Jan 14, 2026 – OpenAI: $20B+ deal to supply 750 MW of AI power through 2028.
• Mar 13, 2026 – AWS: Partnered to run Cerebras chips on Amazon Bedrock for 5x faster AI.
• Oct 27, 2026 – Oracle: Teamed up with Oracle Cloud to power enterprise AI agents.
• Nov 12, 2026 – Microsoft: Confirmed Microsoft signed on as Cerebras' third major cloud customer.
• Q1 2027 – Microsoft Revenue Starts: Full contract details will be released early in 2027 as Microsoft revenue officially kicks in.
• Summer 2027 – Oracle Cloud Launch: Cerebras-powered AI services go live for Oracle Cloud customers.
• Google Cloud & Meta Next: With AWS, Microsoft, and Oracle locked in, Google and Meta are the next major cloud targets.
• More Enterprise & AI Cloud Deals: The Big Three cloud wins serve as proof to attract specialized AI clouds and major companies in pharma, finance, and defense.
• 3x Revenue in 2027: Management expects total revenue to triple in 2027 (to roughly $2.85B) as these deals roll out.
$CBRS is my favorite semiconductor stock right now.
Based on projected FY2028 revenues:
• $CBRS: 3.48x P/S
• $NVDA: 5.67x P/S
• $AMD: 6.38x P/S
The valuation is absurd.
Cerebras trades at a 39% discount to Nvidia and a 45% discount to AMD.
Despite having far greater revenue growth potential than either.
And the Q3 earnings only strengthened my conviction.
• $26.1B in remaining performance obligations.
• Long-term gross margin target of 60%.
• Partnerships with $MSFT, $AMZN, $ORCL, and $AMD.
Cerebras is becoming a critical part of the AI inference ecosystem.
And the demand for ultra-fast inference is going to explode.
The market is massively underestimating both the size of this opportunity and Cerebras’ ability to capture it.
At a $26B market cap, $CBRS is an absolute steal.
$CBRS Feldman: TSMC “sufficient” capacity; not subject to HBM bottleneck. Yields are “high” and high reliability; there has been a lot of misinformation; don’t trust any information that doesn’t come from Cerebras (pretty clear ding on @SemiAnalysis_)🔥
$CBRS $8B in cash, $24B of revenue backlog with new orders, and a product better than NVDAs ...
Then, $20B Market Cap.
This can double tomorrow and Friday, on good earning beats.
Strong Buy