Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
The 2023 Condor Galaxy 1 (CG-1) project showed how Cerebras aimed to compete in AI infrastructure: not just by selling processors, but by building and operating large-scale systems with G42. ServeTheHome’s Patrick Kennedy called Cerebras a “post-legacy silicon AI winner,” but that was his analysis—not a formal ranking or an independently measured verdict. Later company announcements point to further commercial ambitions, not proof that Cerebras is universally better than GPU-based alternatives.
What was Cerebras Condor Galaxy 1?
Condor Galaxy 1 was the first project in the Condor Galaxy network that Cerebras introduced with Abu Dhabi-based G42 in 2023. In his July 20, 2023 article, ServeTheHome’s Patrick Kennedy described CG-1 as a $100 million-plus AI supercomputer project, with its initial Phase 1 deployment in Santa Clara.
Kennedy reported that Phase 1 comprised 32 Cerebras CS-2 systems and more than 550 AMD EPYC 7003 “Milan” CPUs. These are the configuration figures reported in that 2023 article; they should not be read as a verified inventory of every later Condor Galaxy deployment.
Cerebras’ company history describes CG-1 as delivering 4 exaFLOPs of FP16 performance and containing 54 million cores. Those are company-reported specifications, not independent benchmark results. Cerebras’ company history places CG-1 within the network launched with G42 in 2023.
#1 Best Overall
- A USB accessory that brings machine learning inferencing to existing systems. Works with Raspberry Pi and other Linux systems
- Performs high-speed ML inferencing: the on-board edge TPU Coprocessor is capable of performing 4 trillion operations (tera-operations) per second (tops), using 0.5 watts for each tops (2 tops per watt). For example, it can execute state-of-the-art mobile vision models such as mobilenet V2 AT 400 FPS, in a power efficient manner
- Works with Debian Linux: connects to any debian-based Linux system with an included USB 3.0 Type-C cable
- Supports tensorflow Lite: no need to build models from the ground up. Tensorflow Lite models can be compiled to run on the edge TPE
- Supports automl vision edge: easily build and deploy fast, high-accuracy custom image classification models to your device with automl vision edge
Why did the project lead to the “AI winner” claim?
Kennedy’s thesis was about Cerebras’ business model as much as its hardware. Rather than only selling large processors, the company was building AI clusters and seeking to sell available cloud capacity. In principle, operating infrastructure could add recurring service revenue and distinguish Cerebras from companies whose main offering is hardware.
The article also outlined a hypothetical path to $1 billion in AI revenue. That figure was a projection based on assumptions about cluster buildout—not revenue Cerebras had reported earning. Kennedy’s “winner” language was therefore a forward-looking interpretation of Cerebras’ strategy, not an established market status or measured comparison with Nvidia.
Rank #2
- High-Performance AI Processing: The MX3 is designed to handle the most demanding AI computer vision workloads, delivering exceptional performance and efficiency.
- Flexible Integration: The MX3 can be easily integrated into your existing systems via its M.2 M-key form factor and support for Linux operating systems.
- Energy Efficient: The MX3 is designed to provide high performance while minimizing power consumption.
- Comprehensive Software Development Kit (SDK): The MX3 is supported by a comprehensive SDK that simplifies development and deployment.
- Hardware compatability: The MX3 is compatible with the PCI-SIG M.2 M-key 2280 Specification. It can be used with the Raspberry Pi 5 with a M-key 2280 HAT.
What changed after the 2023 project?
Later announcements provide evidence of continued commercial activity, but they need to be distinguished from completed deployments and independently verified performance:
- OpenAI agreement: In January 2026, Cerebras announced a multi-year agreement with OpenAI covering 750 megawatts of wafer-scale systems. The company said deployment would roll out in stages beginning in 2026; the announcement is not evidence that the full capacity had already been installed. Cerebras’ announcement.
- AMD inference partnership: On July 23, 2026, AMD and Cerebras announced a disaggregated inference design. Their stated plan assigns AMD Helios the high-throughput prompt and context processing role, while Cerebras wafer-scale technology handles low-latency decode and token generation. The companies said the joint solution was expected first through Cerebras Cloud in the second half of 2026. This is a company-stated timetable. AMD’s announcement.
- Reported financial results: Cerebras reported $193.4 million in GAAP revenue and a $14.0 million GAAP net loss for the quarter ended March 31, 2026. These are company-reported results for that quarter, not confirmation of the 2023 article’s $1 billion projection. Cerebras’ Q1 2026 results.
In its June 23, 2026 results release, CEO Andrew Feldman said, “AI has moved from being a novelty to being useful and productive.” That is the executive’s view, not an independent assessment of Cerebras’ competitive position.
Is Cerebras better than Nvidia for AI?
There is no defensible universal answer from the available figures. Cerebras’ wafer-scale systems and GPU-based infrastructure are different system approaches; a fair comparison needs to match the workload and configuration. A system’s performance for model training, prompt processing (prefill), or token generation (decode) may differ, and a result in one phase does not establish superiority in another.
For a useful comparison, examine the specific model and context size, latency, throughput, scale, software support, deployment option, and cost. Check whether performance figures come from an independent, comparable benchmark or from a vendor announcement. The AMD-Cerebras partnership itself frames inference as a set of trade-offs involving latency, throughput, token capacity, cost, and scale; AMD CEO Lisa Su said, “AI inference is becoming one of the largest infrastructure opportunities in AI, and its growing diversity requires a more flexible approach.” This is a vendor executive statement, not independent proof that the announced design outperforms alternatives.
Rank #4
What the $100M figure does—and does not—tell you
The $100 million-plus figure belongs to Kennedy’s account of the 2023 Cerebras-G42 project. It conveys the scale of the planned infrastructure effort, but it is not a price per CS-2, a benchmark result, or evidence that every planned cluster phase was completed. The article discussed expansion in the United States and internationally; the available later announcements do not establish that all those plans were built on the schedule projected in 2023.
CG-1 is best understood as an early example of Cerebras’ broader strategy: pair wafer-scale systems with cloud capacity and large infrastructure partnerships. The project made that strategy concrete, while later announcements suggest that the company continued pursuing it. Neither the 2023 forecast nor vendor-reported specifications establish a universal performance win over GPU infrastructure.
Quick Recap
Best Value
- ✅Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor. 2.5W typical power consumption
- ✅Scalable, enabling simultaneous processing of multi-streams & multi-models
- ✅Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
- ✅Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
- ✅Supports Linux and Windows. Supports the temperature range of -40°C to 85°C
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
