Cerebras Systems
Cerebras Systems, founded in 2016 by ex-SeaMicro executives including CEO Andrew Feldman and CTO Gary Lauterbach, is a Sunnyvale, California-based AI compute company that designs wafer-scale processors—single chips spanning an entire silicon wafer. Its WSE-3 (Wafer-Scale Engine 3), announced on March 13, 2024, integrates 4 trillion transistors and 900,000 AI-optimized cores with 44 GB of on-chip SRAM, delivering 21 PB/s of memory bandwidth. The accompanying CS-3 system executes workloads on one chip, eliminating the networking bottlenecks of GPU clusters. In July 2023, the company and G42 revealed the Condor Galaxy supercomputer network: the first node, Condor Galaxy 1 (CG-1), combines 64 CS-2 systems to reach 4 exaFLOPS of AI compute, with a planned build-out across multiple sites. The company’s earlier Andromeda supercomputer, composed of 16 CS-2 systems, demonstrated near-linear scaling on large language models. Cerebras matters now because its single-chip architecture and memory-bandwidth advantage offer a concrete alternative to Nvidia’s multi-GPU infrastructure, directly addressing the industry’s critical scaling and power-efficiency bottlenecks.
Signal Radar
Five-axis snapshot of this entity's footprint
Mentions × Lab Attention
Weekly mentions (solid) and average article relevance (dotted)
Timeline
15- Product LaunchJul 19, 2026
Cerebras CS-3 launched achieving 500+ token/s for Llama 2 70B
View source- speed:
- 500+ token/s
- PartnershipJul 9, 2026
Cerebras and Flex announce 7x production expansion of CS-3 at Milpitas facility
View source- production increase:
- 7x
- location:
- Milpitas, California
- Product LaunchJun 4, 2026
Disclosed three mechanical innovations for wafer-scale chip cooling: vertical power delivery, flexible interposers, direct-impingement cooling
View source - Product LaunchMay 27, 2026
Announced CS4 wafer-scale chip staying on 5nm due to SRAM scaling flattening.
View source - Product LaunchMay 23, 2026
Cerebras published benchmark results for Kimi K2.6 on CS-3, claiming 981 tokens/sec and 6.7× speedup over GPU cloud.
View source - Research MilestoneMay 18, 2026
Claims 10x training speed over Nvidia H100 for GPT-3-scale models using WSE-3
View source- speed claim:
- 10x
- model scale:
- 175B parameters
- transistors:
- 4 trillion
- Regulatory ActionMay 7, 2026
SemiAnalysis noted that Cerebras understates on-chip SRAM by 8x on its website.
View source - IpoApr 21, 2026
Confidentially filed paperwork with the SEC for an initial public offering.
View source- filing type:
- confidential
- regulator:
- SEC
- IpoApr 21, 2026
Filed for IPO on April 21, 2026, betting wafer-scale chips can disrupt Nvidia's GPU cluster model.
View source - Product LaunchSep 30, 2025
Cerebras reported $187M revenue for first 9 months of 2025, up from $78M year prior
View source- revenue:
- $187M
- prior revenue:
- $78M
- IpoJan 1, 2025
Cerebras shares open at $385, 108% above IPO price, market cap ~$68B
View source- opening price:
- $385
- market cap:
- $68B
- IpoJan 1, 2025
Cerebras IPO priced at $185 per share, raising $5.5 billion
View source- amount:
- $5.5B
- price per share:
- $185
- PartnershipFeb 21, 2023
Strategic partnership with G42 yields breakthrough AI training results 100x faster than GPU clusters
View source- performance gain:
- 100x
Relationships
26Competes With
Developed
Uses
Regulated
Frequently appears with
10Entities that show up in the same articles — shared coverage, not a stated relationship.
Recent Articles
4AMD-Cerebras Disaggregated Inference: 5× T/s/W, Prompt vs. Decode Split
+AMD and Cerebras launched a disaggregated inference platform splitting prompt processing on Helios from decode on WSE, claiming up to 5× T/s/W.
100 relevanceGPT-5.6 Sol on Cerebras Hits 750 Token/s
+GPT-5.6 Sol on Cerebras claimed at 750 token/s, but no official data or model release exists. Unverified claim needs vendor confirmation.
97 relevanceCerebras, Flex Expand CS-3 Production 7x at Milpitas Facility
+Cerebras and Flex expand CS-3 production 7x at Milpitas facility. The partnership keeps wafer-scale AI manufacturing in the U.S. as Nvidia faces delay
85 relevanceNvidia's Next-Gen AI Rack Delayed to 2028, SemiAnalysis Says
+Nvidia's next-gen AI rack delayed to 2028 on manufacturing snags per SemiAnalysis. Delay benefits AMD and custom silicon rivals.
95 relevance
Predictions
No predictions linked to this entity.
AI Discoveries
2- observationactive6d ago
Lifecycle: Cerebras Systems
Cerebras Systems is in 'active' phase (1 mentions/3d, 2/14d, 21 total)
90% confidence - hypothesisactiveJul 19, 2026
H: Within 90 days, Alibaba Cloud will announce a managed inference service for GPT-5.6 Sol on Cerebras
Within 90 days, Alibaba Cloud will announce a managed inference service for GPT-5.6 Sol on Cerebras hardware, directly competing with Nvidia's GPU cloud offerings and leveraging the SAIL stack.
65% confidence
Sentiment History
| Week | Avg Sentiment | Mentions |
|---|---|---|
| 2026-W24 | 0.50 | 1 |
| 2026-W28 | 0.60 | 2 |
| 2026-W29 | 0.30 | 1 |
| 2026-W30 | 0.70 | 1 |