user nl
Well-known member
The AI chip race isn't just a one-horse sprint led by Nvidia.
As hyperscalers like Google, Meta, and Microsoft race to lower the eye-watering costs of running massive AI models, a second front is opening in the custom silicon wars, with Broadcom as its primary architect.
"Broadcom is projected to retain its leadership as the premier AI Server Compute ASIC design partner with a 60% market share in 2027," according to a recent report from Counterpoint Research.
This dominance is underpinned by a symbiotic relationship with the world's most advanced foundry, Taiwan Semiconductor Manufacturing Company, which remains the "dominant foundry choice ... with close to 99% wafer fabrication share for the top 10 players' AI Server Compute and ASIC shipments."
...........................................................................
Then there is the TSMC factor. As the industry's indispensable builder, TSMC holds a unique monopoly. Whether a company chooses an Nvidia GPU or a Broadcom-designed ASIC, the chips are almost certainly being forged in TSMC's fabs. Furthermore, as individual chips hit their physical size limits, TSMC's advanced packaging — stacking chips to boost power — means it catches more value from every high-end AI chip produced, regardless of whose logo is on the box.
https://finance.yahoo.com/news/broa...ers-in-the-custom-ai-chip-boom-130336239.html
As hyperscalers like Google, Meta, and Microsoft race to lower the eye-watering costs of running massive AI models, a second front is opening in the custom silicon wars, with Broadcom as its primary architect.
"Broadcom is projected to retain its leadership as the premier AI Server Compute ASIC design partner with a 60% market share in 2027," according to a recent report from Counterpoint Research.
This dominance is underpinned by a symbiotic relationship with the world's most advanced foundry, Taiwan Semiconductor Manufacturing Company, which remains the "dominant foundry choice ... with close to 99% wafer fabrication share for the top 10 players' AI Server Compute and ASIC shipments."
...........................................................................
Then there is the TSMC factor. As the industry's indispensable builder, TSMC holds a unique monopoly. Whether a company chooses an Nvidia GPU or a Broadcom-designed ASIC, the chips are almost certainly being forged in TSMC's fabs. Furthermore, as individual chips hit their physical size limits, TSMC's advanced packaging — stacking chips to boost power — means it catches more value from every high-end AI chip produced, regardless of whose logo is on the box.
https://finance.yahoo.com/news/broa...ers-in-the-custom-ai-chip-boom-130336239.html
Also, Broadcom is not the "primary architect" of custom silicon. The cloud companies do their own digital design, and they license selected IP blocks from Broadcom (and Cadence or Synopsis, and other companies), especially for common I/Os like PCIe and Ethernet, and embedded cores. Broadcom does the so-called "back-end" functions (it apparently varies by cloud company), and that is a very important part of successful chip design, but calling them the primary architect of the not-Nvidia market is just marketing. And the article's wording about "turning these internal corporate blueprints into functional hardware" is cringeworthy. In reality, Broadcom is a conglomerate that gets 42% of their revenue from enterprise software.