Broadcom Earnings

Broadcom’s latest results reminded investors that the Artificial Intelligence investment cycle is spreading across the global technology supply chain.

The company’s AI semiconductor business continues to grow rapidly, supported by demand for custom accelerators, networking chips and data-centre connectivity. But the market reaction also revealed something equally important: expectations for the most visible US AI winners are now exceptionally high.

That combination — robust underlying demand but less forgiving valuations — strengthens the case for looking beyond a narrow group of US technology names and towards Asian companies at the centre of foundry, memory, advanced packaging, equipment and server infrastructure.

On the face of it, Broadcom’s numbers were impressive. The company delivered second-quarter revenue of $22.2 billion, a 48% year-on-year increase, alongside non-GAAP earnings per share of $2.44 and free cash flow of $10.3 billion. Most significantly, AI semiconductor revenue increased by 143% year-on-year to $10.8 billion, largely driven by custom AI accelerators and AI networking. Management also guided to AI semiconductor revenue of around $16 billion in the current quarter, implying growth of more than 200% year-on-year.

Yet the share price fell sharply after the announcement.

That is the key message for investors. Broadcom did not challenge the AI investment case. Instead, it showed that the market is beginning to demand near-perfection from the companies most closely associated with the theme. Strong growth alone is no longer enough when expectations have already been raised so aggressively.

This is consistent with what we have seen elsewhere in the AI infrastructure complex. Dell recently raised its forecast for annual AI server revenue, with AI servers now playing a much larger role in the group’s growth story. Hewlett Packard Enterprise also delivered a strong update, increasing its outlook as demand for AI infrastructure accelerated.

Taken together, Dell, HPE and Broadcom point to the same conclusion: AI demand is being pulled through the entire hardware stack — from servers and storage to custom silicon, switching, networking and connectivity.

The market is therefore not asking whether demand for AI infrastructure exists. It is asking how much of that demand is already reflected in valuations.

The bullish case for Broadcom has been well captured by Bernstein analyst Stacy Rasgon. Following the company’s previous results, he argued that Broadcom had addressed several major investor concerns, raised his price target and highlighted the potential for significant earnings power if the AI ramp continued. That remains an important point.

Broadcom is not simply another semiconductor stock benefiting from the AI halo. It occupies a strategically valuable position in custom silicon, networking and data-centre infrastructure, giving it operating leverage and cash generation that few companies can match.

However, the latest market reaction also illustrates why selectivity now matters. Investors should be cautious about equating a powerful structural growth story with an automatic positive share-price response. As AI winners become more widely owned, the hurdle rate rises.

The implications extend far beyond the US market.

The Asian read-across from Broadcom’s results was particularly notable, with clear dispersion across Taiwan’s AI server supply chain. Several names tied to server manufacturing and hardware infrastructure moved lower, including Wistron, Inventec, Hon Hai and Quanta. TSMC was also weaker.

In other words, this was not a straightforward “AI up” reaction. It was a more discriminating response to valuation, positioning and the sustainability of earnings upgrades.

That nuance is important. Asia remains at the centre of the AI supply chain, but it is not a single trade. Taiwan is critical to advanced foundry, server manufacturing and packaging. Korea is central to memory and high-bandwidth memory. Japan remains highly relevant in semiconductor equipment, testing and precision components.

The opportunity is therefore broad, but uneven. The likely winners are companies with bottleneck exposure, pricing power and the ability to convert volume growth into earnings growth.

This is exactly why regional diversification matters. The AI theme has often been expressed through a small number of US mega-cap technology companies, but the infrastructure required to deliver AI at scale is global. The servers, chips, substrates, memory, networking equipment and production capabilities that underpin the theme are deeply embedded across Asian markets.

For investors, Broadcom’s lesson is not that the AI trade is over. It is that the next phase may become more selective.

The first phase rewarded the most obvious US beneficiaries. The next phase may increasingly reward those parts of the global supply chain where demand remains robust, capacity is limited and earnings revisions still have room to compound.

Broadcom did not destroy the AI trade. It raised the standard of proof.

That makes a diversified approach to AI exposure more important, not less.

This article is for information only and does not constitute investment advice or a personal recommendation. Capital is at risk, and the value of investments can fall as well as rise.