Broadcom (AVGO) CEO says raises FY26 AI revenue guidance to USD 58bln from USD 56bln, which is a 186% increase Y/Y

• Production shipments of next-generation TPU version 8i for Google began during the quarter.
• New Google TPU is the same or better than NVIDIA Vera Rubin.
• Now shipping TPU version 8i ahead of MediaTek’s version V8t.
• Expects to accelerate shipments of Ironwood to Anthropic in Q4.
• Expects to ramp up high-volume shipments of TPU version 8i to Google in Q4.
• Expects production shipments of Meta’s custom MTIA accelerator at scale in Q4.

A raise to AI revenue guidance from the dominant custom-silicon vendor follows a familiar pattern in this cycle: the read-through is judged less by the increment itself than by what it implies about hyperscaler capex intentions, since guidance of this kind is effectively a statement about customers' build-out plans rather than the company's own demand generation. The comparison drawn against the leading merchant GPU roadmap is notable, as it speaks directly to the share question that has defined the AI accelerator trade, namely whether custom ASIC programs erode or merely supplement merchant silicon volumes. In past episodes, named-customer ramp commentary of this sort has tended to lift the custom-silicon peer set while pressuring the merchant GPU complex at the margin, with the dispersion between the two the sharper signal than the index reaction. The sequencing detail matters here: multiple hyperscaler programs ramping in the same quarter concentrates execution risk and raises the sensitivity of the next few quarters' AI revenue prints to any slippage in production timelines. Follow-ons worth noting are how the merchant GPU incumbent frames competitive positioning on its next interaction, and whether the guidance raise is corroborated by commentary from the named customers themselves. As earnings guidance, the print carries the usual caveat that AI-line definitions and backlog conversion have been points of scrutiny for this issuer previously.