I guess we are in an era where the zeitgeist is not to teach the man how to fish, but the AI model, ensuring that the AI lab - Anthropic in this case - would then nosh on a lifetime's worth of customized silicon chips.

We reported back in May that Anthropic's passing remarks on "logic chips" in relation to its evolving partnership with Samsung had unleashed a flood of speculation regarding a possible ASIC-related tie-up between the two.

Well, just around two months later, Anthropic all but confirmed to Business Insider earlier this week that it indeed plans to build such a custom chip by acknowledging its ongoing efforts to assemble an in-house silicon design team , going on to add that it has and will continue to take a "multi-chip approach."

As Anthropic continues to beef up its silicon design team, however, an interesting pay disparity has emerged, with the AI lab willing to pay between $500,000 and $850,000 per year to research engineers who are teaching its AI models how to design silicon chips, but just between $320,000 and $485,000 to the silicon engineers who are actually designing its first ASIC.

What's more, this material pay disparity has emerged despite the convergent skillsets required for these two roles, including expertise on full ASIC/FPGA flow, RTL to tape-out, UVM/formal, physical design, PPA, DFT, and EDA tools.

Of course, Anthropic is not the first AI lab to try to leverage AI models for chip design processes. After all, Moonshot's viral Kimi K3 model was recently able to design a viable silicon chip completely autonomously within 48 hours, producing a chip design with a 4.0 mm² area and a Nangate 45nm library, all by leveraging open-source EDA tools and building a GPU compiler from scratch. The virtual chip even managed to achieve over 8,700 tokens/second decoding throughput in simulation.

Looks like even silicon engineers are not immune from the AI's roving wrath of obsolescence.

Follow Wccftech on Google to get more of our news coverage in your feeds.