On Thursday, US semiconductor stocks came under notable selling pressure following Broadcom’s latest earnings report and forward outlook. The primary reason behind the weaker market sentiment stemmed from Broadcom’s recent results and guidance. Although the company reported second-quarter revenue of US$22.2 billion, up 48% year-on-year, including US$10.8 billion in AI-related revenue, and forecast third-quarter revenue of US$29.4 billion, investors chose to take profits because actual revenue came in slightly below market expectations and CEO Hock Tan did not raise the company’s full-year sales target. As a result, Broadcom shares fell more than 15% in pre-market trading.

At the same time, investment bank Macquarie downgraded Broadcom from “Outperform” to “Neutral” and lowered its 12-month price target. Macquarie believes Broadcom’s leadership position in the AI ASIC market is facing increasing challenges. Google, which has historically relied heavily on Broadcom’s chips, is accelerating its in-house AI chip development strategy while simultaneously partnering with MediaTek to diversify its supply chain. Macquarie suggested that Broadcom could become attractive again around the US$400 level.

Macquarie forecasts that Broadcom’s share of revenue generated from Google’s TPU program will decline from 95% in 2026 to 80% in 2027, before falling further to 65% in 2028.

Despite the downgrade, Macquarie has not turned completely bearish on Broadcom. The firm also raised its earnings forecasts for fiscal 2026 and fiscal 2027 by 12% and 14%, respectively, reflecting the view that the current boom in AI infrastructure investment will continue to support industry growth. However, as major technology companies increasingly strengthen their in-house chip capabilities, the competitive structure of the AI semiconductor supply chain could undergo significant changes in the years ahead.

Market Analysis:

This downgrade not only signals growing caution regarding Broadcom’s long-term growth outlook, but also reflects a broader shift in investor focus toward a key question facing the AI industry:

As demand for AI computing power continues to surge, will hyperscale cloud providers remain dependent on external chip suppliers, or will they increasingly transition toward internally developed solutions?

The answer to that question is likely to become one of the most important variables shaping the future competitive landscape of the AI semiconductor industry over the next several years.