Anthropic offers high salaries for self-developed AI chips, and the salary for model training positions is higher than that of actual core-making engineers

IT House News on August 7, according to Wccftech reported today, Anthropic is currently expanding its internal chip design team and plans to develop self-developed ASIC chips. At the same time, there are obvious salary differences between different chip-related positions in the company: the annual salary of a research engineer responsible for training AI models for chip design can reach up to 850,000 US dollars (IT House Note: the current exchange rate is approximately 5.749 million yuan), while the salary range of the engineer position actually responsible for designing the first ASIC chip of Anthropic is approximately 320,000 to 485,000 US dollars (the current exchange rate is approximately 2.164 million to 328,000 yuan). million yuan). Wccftech calls it a “pay inversion.” The recruitment information released by Anthropic shows that the research engineer position responsible for the direction of "Chip Design RL (Chip Design Reinforcement Learning)" is mainly responsible for building a reinforcement learning environment to allow AI models such as Claude to learn the chip design process, including RTL code generation, verification, and physical design optimization. The salary range for this position is US$500,000 to US$850,000 (current exchange rate is approximately RMB 3.382 million to RMB 5.749 million). In comparison, the salary range for Silicon Engineer positions, which are responsible for actual chip design work, is approximately US$320,000 to US$485,000 (current exchange rate is approximately 2.164 million yuan to 3.28 million yuan). Wccftech pointed out that there is a high degree of overlap in the skills required for the two types of positions, including the complete ASIC/FPGA design process, RTL to tape-out, UVM and formal verification, physical design, PPA (performance, power consumption, area) optimization, DFT (design for testability) and EDA tool usage experience, etc., and both require actual chip development experience. Currently, Anthropic has not announced its first self-developed ASI