Etched AI is gaining attention in the tech world with its bold plan to develop an AI chip called the Sohu chip. This chip is unique because it is designed to run only one type of model: transformers, which are widely used in generative AI. Founders Gavin Uberti and Chris Zhu, along with their team, believe that this specialized design will significantly enhance AI chip performance. Try this Ginny AI Personal Tutor For Student 2024
What is Etched AI?
Etched AI was founded in 2022 by two Harvard dropouts, Gavin Uberti and Chris Zhu, who wanted to transform AI hardware. They teamed up with Robert Wachen and Mark Ross, the former CTO of Cypress Semiconductor, to bring their vision to life.
Their goal was to meet a specific need in the AI industry by creating a chip that excels at running transformer models. Unlike many general-purpose AI chips, the Sohu chip is a specialized application-specific integrated circuit (ASIC) designed only for transformers.
Sohu is designed specifically for transformers, a key model in generative AI, unlike general-purpose AI chips.
Transformers have become crucial in the AI world since Google researchers introduced them in 2017. They power many generative AI models, like OpenAI’s video generator Sora, text generators such as Anthropic’s Claude and Google’s Gemini, and art generators like Stable Diffusion.
Understanding the importance of transformers, Etched AI focused all their efforts on optimizing their chip for this model.
What sets Sohu chip?
Forget bulky GPUs Etched AI’s new Sohu chip, built with cutting-edge 4nm technology, promises a greener and faster way to run complex AI tasks, especially those involving text, images, and videos. Unlike traditional GPUs, Sohu is specifically designed for these jobs, making it much faster and more energy-efficient. In fact, reports suggest a single Sohu server can outperform a whopping 160 Nvidia H100 GPUs – that’s a massive leap in performance with a smaller environmental footprint. This innovative chip could revolutionize how AI is powered, paving the way for a new era of efficient and eco-friendly artificial intelligence. Read this also The Latest NVIDIA Blackwell AI Superchip and Architecture
Unlike most graphics cards (GPUs) that try to do everything, Sohu is laser-focused on a specific type of AI task: handling complex jobs that involve text, images, and videos. This focus is what makes Sohu so special. By ditching unnecessary features and streamlining its design, Sohu runs much faster and uses less energy than traditional GPUs. It’s kind of like having a super-powered tool specifically designed for these tasks, instead of a bulky toolbox with everything under the sun. This targeted approach also helps tackle the growing concern of AI’s environmental impact, as Sohu’s efficiency reduces the overall power consumption needed for complex AI tasks. With Sohu, Etched AI is offering a greener and faster alternative for companies looking to harness the power of AI without breaking the bank or the environment.
Competitive Market
Etched AI is entering a highly competitive market dominated by companies like Nvidia, which controls much of the AI chip market. Major cloud providers like Meta and Microsoft rely heavily on Nvidia’s GPUs. However, this dependence has led generative AI vendors and venture capitalists to look for alternative solutions.
Etched AI stands out by focusing solely on transformer models. Unlike other companies like Meta and Amazon, which have developed their own inferencing chips, Etched AI’s unique selling point is its specialization in transformers. This specific focus has attracted significant investment, recently raising $120 million in a Series A funding round, bringing their total funding to $125.36 million.
High-profile investors include:
- Peter Thiel
- GitHub CEO Thomas Dohmke
- Quora co-founder Charlie Cheever
Despite challenges such as potential new competitors and changes in AI model preferences, Etched AI remains optimistic. They are preparing to launch the Sohu Developer Cloud, an online platform for customers to preview Sohu’s capabilities, which is expected to boost interest and sales.
The success of Etched AI depends on its ability to scale production and meet the demand for efficient, specialized AI chips. Co-founder Gavin Uberti believes that the future of AI technology depends on faster and more efficient infrastructure. His bold claims about Sohu’s capabilities and the company’s aggressive market strategy highlight Etched AI’s commitment to shaping the future of AI hardware.