Broadcom stock rose on Thursday after its earnings report. Investors look more convinced about the growth of its custom artificial-intelligence chips than the AI processors of its rival Nvidia.
Broadcom’s first-quarter AI revenue of $8.4 billion more than doubled from the previous year. However, more importantly CEO Hock Tan said Wednesday that it has “line of sight” to sales from that business significantly exceeding $100 billion in 2027.
Broadcom shares closed up 4.8% at $332.77 on Thursday.
Expectations were high coming into the print. Broadcom trades at 27.2 times earnings expected over the next 12-months. That’s more expensive than chip giant Nvidia, which is trading at 21.6 times forward earnings, and AMD, which is trading at 27 times forward earnings.
Nvidia stock fell after its own earnings report last week despite beating analysts’ expectations and was down 0.9% early Thursday.
“Following the disappointing market reception of Nvidia’s stronger than projected results and outlook, the Street was poised for Broadcom’s shares to react with similar lethargy,” wrote Benchmark Research analyst Cody Acree in a research note. “However, Broadcom stock rose…with investors most notably impressed by the company’s much higher 2Q outlook and its forecast for its AI revenue to grow to significantly more than $100 billion next year.”
Acree has a Buy rating and $485 target price on Broadcom stock.
Broadcom reported fiscal first-quarter adjusted earnings of $2.05 a share, which beat analysts’ estimates of $2.03 a share, according to FactSet. Revenue for the quarter of $19.31 billion also came in higher than expectations of $19.26 billion and jumped 29% from the prior year.
Revenue for the semiconductor solutions segment was $12.52 billion, compared with Wall Street estimates of $12.4 billion. Infrastructure software revenue for the quarter of $6.8 billion just missed estimates of $6.99 billion.
Broadcom said that it expects second-quarter revenue to be about $22 billion, which was above the consensus call among analysts for $20.5 billion.
“Broadcom seems to have strong visibility on spend until 2028. This is longer than what others have talked about most notably Nvidia,” said Ryuta Makino, research analyst at Gabelli Funds.
Broadcom shares have jumped 77% over the past 12 months as investors bet the chip company will continue to benefit from AI demand.
However, the stock had fallen 8.3% this year so far through to Wednesday’s close, partly dogged by fears over its software segment. Some of those software fears appeared to ease on Thursday after CEO Tan said the company’s infrastructure software is not disrupted by AI.
Investors are becoming more selective about the AI trade. Tech stocks have had a massive run leading up to 2026 as Wall Street bets on who will benefit from the future of AI. But now, the geopolitical environment is uncertain, and questions remain about whether or not companies will start to pull back their AI spending.
The growth of Google’s Tensor Processing Units (TPUs)—which Broadcom helped design—has brought the biggest competitive challenge yet to Nvidia in the AI chips sector. Under the right circumstances they can be more cost efficient than Nvidia’s graphics-processing units.
Broadcom said there was strong demand this year for the seventh-generation Ironwood TPU and it expects that to increase in 2027 with future TPU generations. It also said customers were increasingly using its custom AI chips—which it calls XPUs—for training AI models, an area where Nvidia’s GPUs have dominated.
“The momentum of Google and its success and scale around TPU is showing up in [Broadcom’s] numbers while it also continues to deliver across its broad network and software portfolio,” said Daniel Newman, CEO of technology research firm Futurum.
In addition to Google, Broadcom counts Facebook-parent Meta Platforms and AI start-ups OpenAI and Anthropic among its significant AI-chip clients.
Nvidia may now look to strike back with hardware announcements at its own GTC conference this month. Nvidia is planning to release a new AI processor specialized for inference—the process of generating answers or results from AI models—at GTC, The Wall Street Journal reported recently, citing people familiar with the plans.
Comments