In the past two months, it surged 70%, with a market value exceeding one trillion. AI chips are booming, but is Broadcom too expensive?

Wallstreetcn
2025.06.09 02:17
portai
I'm PortAI, I can summarize articles.

Broadcom's expected price-to-earnings ratio reaches 33.6 times, a premium of 23% over the Philadelphia Semiconductor Index, close to the highest premium in a decade, and a 13% premium over NVIDIA. In addition, Broadcom's AI revenue is highly dependent on Alphabet, with TPU orders accounting for more than 80% of its AI revenue, and it also faces challenges related to the cost and performance of custom chips

As the "new darling" of the AI boom, can Broadcom meet the high expectations of the market?

In the past two months, chip manufacturer Broadcom's stock price has soared by 70%, with a market capitalization exceeding $1 trillion, making it the seventh largest publicly traded company in the United States, reflecting investors' high hopes for its explosive growth in the AI chip sector.

According to the company's latest financial report, Broadcom expects its AI processing and networking chip revenue to reach $5.1 billion in the third quarter, a year-on-year increase of 60%, accounting for about one-third of total sales. However, this guidance is still not as impressive as some optimistic investors had anticipated. After the earnings report was released, Broadcom's stock price has fallen for two consecutive days.

According to Wind Information, Bank of America analyst Vivek Arya and his team pointed out in their latest report that behind the enthusiasm for Broadcom lies high development costs, performance bottlenecks, and valuation risks.

High Customer Concentration and High Valuation

As a pioneer in the custom AI chip field, Broadcom began developing custom AI chips for Google several years ago, and this once "side business" has now become a major growth engine amid the AI boom.

Bank of America stated that custom chip projects inherently carry a high customer concentration risk, and Broadcom's AI revenue may heavily rely on Google.

The report noted that orders for Google's TPU may account for over 80% of Broadcom's ASIC revenue, which means that if Google adjusts its strategy or turns to other suppliers, Broadcom will face significant volatility.

According to media reports, although Broadcom expects to diversify its AI customer base, this transition may not take effect until the second half of fiscal year 2026, and there are almost no external indicators to track the company's collaboration progress with potential clients, especially those deeply integrated into the NVIDIA ecosystem.

Analysts suggest that in the long run, Broadcom's biggest challenge may be that its market is more limited than NVIDIA's, as custom AI chips may still remain the exclusive domain of financially strong tech companies.

Broadcom's valuation issues are also quite prominent.

Relevant data shows that Broadcom's expected price-to-earnings ratio is 33.6 times, a premium of 23% over the Philadelphia Semiconductor Index, close to the highest premium in the past decade, and a 13% premium over NVIDIA.

Facing Performance, Cost, and Market Challenges

Moreover, the custom AI chip market is not without challenges.

A report from TD Cowen analysts pointed out that the performance of Google's latest custom AI chips is only about half that of NVIDIA's corresponding products, barely reaching the threshold of financial viability. Morgan Stanley analysts also stated that performance issues with custom chips are "more common than people think." Secondly, even if the production cost of custom chips is far lower than NVIDIA's high-end chips, which often exceed $30,000, the supporting systems (such as fiber optic network technology) are often more expensive than NVIDIA's copper wire systems, offsetting some of the cost advantages. Not to mention the additional burden of developing compatible software—NVIDIA's customers can almost directly "move in," while Broadcom's customers need to be self-reliant.

In addition, Broadcom is not only facing increasingly fierce competition from NVIDIA, but also from chip manufacturers like Marvell and MediaTek, who are gradually eating into its market share.

AI Infrastructure Demand Will Continue to Support Long-Term Growth

Despite the challenges, Bank of America remains optimistic about Broadcom's growth prospects overall.

The report points out that although competition from NVIDIA may intensify, the overall rise of the AI wave should be able to support Broadcom's demand, and Broadcom's network product portfolio is well-positioned in the expanding demand for AI clusters.

The report predicts that Broadcom's earnings per share for the fiscal year 2027 will be approximately $10 billion based on $45 billion in AI sales, and if the company reaches its target midpoint ($75 billion), this could bring a growth potential of $13-14 per share.

Bank of America raised Broadcom's target price from $240 to $300, reaffirming its "Buy" rating and maintaining its position as a "top stock in the computing sector and one of the top five industry picks."

Morgan Stanley also expects that by 2027, the custom chip market size will grow from $12 billion last year to approximately $30 billion. Although this growth trend is not as strong as the broader AI chip market where NVIDIA operates (which is expected to grow at an annual rate of 23% over the next five years, approaching $500 billion by 2030), it still provides Broadcom with considerable growth space