OpenAI to develop AI chips in-house

OpenAI is preparing to make a significant leap into the AI chip market by planning their move into the chip industry since last year. They are exploring various possibilities, such as acquiring an AI chip manufacturer or buckling down to craft their own chips.

• CEO's Ambition: Sam Altman, OpenAI's CEO, has a clear objective - secure more AI chips.

• Existing Dependencies: At present, OpenAI relies heavily on GPU-based hardware for developing advanced models like ChatGPT, GPT-4, and DALL-E 3. These GPUs are adept at handling multiple computations simultaneously.

• Supply Chain Challenges: However, the AI boom has left the GPU supply chain gasping. Major players like Microsoft are grappling with hardware shortages, and Nvidia's premium AI chips are sold out until 2024.

• Cost Implications: Assuming ChatGPT queries amount to only a tenth of Google Search scale, an initial investment of a staggering $48.1B worth of GPUs would be required, with an additional $16B per year just for operational costs.

• Following the Trend: OpenAI isn't pioneering the trend of AI companies developing their own chips. Google, Amazon, and reportedly Microsoft, have already taken the plunge.

• Financial Strength: Backed by $11B in venture capital and close to $1B in annual revenue, OpenAI is well-positioned to invest heavily in research and development.

• Potential Risks: However, the hardware business is not always smooth sailing. AI chipmaker Graphcore's valuation fell by $1B last year, Habana Labs had to let go of 10% of its workforce, and even Meta faced difficulties with their custom AI chip efforts.

With the potential risks and rewards, it remains to be seen whether OpenAI's investors, including Microsoft, are prepared for this venture.