Broadcom Q4 2025 Earnings: 8 Things To Know From AI Strength, VMware Growth To Anthropic Deal Reveal

A focus on the accelerators, networking, and other components needed to build infrastructures, plus the signing of huge deals like the new $11 billion sale to Anthropic on top of a massive and growing backlog spike optimism on the future of Broadcom, President and CEO Hock Tan told analysts after his company’s fiscal year 2025 financial report.

Broadcom, the Palo Alto, Calif.-based designer and producer of chips and infrastructure software, reported fiscal year 2025 total revenue growth of 24 percent over that of fiscal 2024, led by a huge growth in its AI semiconductor and its VMware businesses.

Broadcom President and CEO Hock Tan, in both his prepared remarks for the company’s fiscal year 2025 financial report and his responses to financial analysts’ questions on Thursday, painted a picture of a company for whom AI shows no limits to growth.

Tan said Broadcom’s fiscal fourth quarter AI semiconductor business grew 74 percent year-over-year, with custom AI semiconductor sales doubling over last year.

At the same time, strong growth in VMware Cloud Foundation revenue pushed infrastructure software revenue growth of 26 percent, he said.

“In summary, 2025 was another strong year for Broadcom, and we see the spending momentum by our customers for AI continuing to accelerate in 2026,” he said.

Tan also revealed that the mystery customer that Broadcom introduced in September as signing a $10 billion custom chip deal was Anthropic, and said that Anthropic in the fourth fiscal quarter signed another deal worth $11 billion.

There’s a lot to unpack from Broadcom’s fiscal 2025 report. To learn more, read CRN’s look at eight of the key results for the year as presented by Tan.

‘Another Strong Year’ For AI, VMware

Broadcom had a blockbuster fiscal fourth quarter 2025 and full year 2025, Tan said.

Total fiscal 2025 revenue grew 24 percent year-over-year to a record $64 billion, driven by AI semiconductors and VMware. AI revenue grew 65 percent year-over-year to $20 billion, which drove Broadcom’s semiconductor revenue to a record $37 billion for the year. On the infrastructure software side, strong adoption of VMware Cloud Foundation or VCF pushed revenue growth of 26 percent year-over-year to $27 billion, he said.

“In summary, 2025 was another strong year for Broadcom, and we see the spending momentum by our customers for AI continuing to accelerate in 2026,” he said.

AI Semiconductor Business Driving Overall Semiconductor Revenue

Broadcom’s total fiscal fourth quarter reached a record $18 billion, up 28 percent year-over-year on better-than-expected growth in AI semiconductors and infrastructure software, Tan said. Semiconductor revenue reached $11.1 billion, up 35 percent year-over-year driven by a 74 percent growth in AI semiconductor revenue to $6.5 billion, Tan said.

“This represents a growth trajectory exceeding 10 times over the 11 quarters we have reported this line of business,” he said. “Our custom accelerator business more than doubled year-over-year as we see our customers increase adoption of XPUs, as we call those custom accelerators, in training their LLMs and monetizing their platforms through inferencing APIs and applications. These XPUs, I may add, [have] not only been used to train and inference internal workloads by our customers. The same experience in some situations has been extended externally to other LLM peers, best exemplified at Google, where the TPUs used in creating Gemini are also being used for AI cloud computing by Apple, Cohere, and SSI (Safe Superintelligence) as a sample.”

Big Customers For AI

The scale Broadcom is seeing is significant, Tan said, as exemplified by a massive order from Anthropic.

“As you are aware, last quarter, Q3 ‘25, we received a $10 billion order to sell the latest TPU Ironwood racks to Anthropic,” he said. “And this was our fourth custom that we mentioned [previously]. And in this quarter, Q4, we received an additional $11 billion order from this same customer for delivery in late 2026.”

In the fourth quarter, Broadcom acquired a fifth XPU customer through a $1 billion order placed for delivery in late 2026, Tan said. When asked by financial analysts if that customer was OpenAI, which in October signed a multi-year partnership for accelerator and networking systems for next-generation AI clusters, Tan declined to answer.

AI Networking Booming

AI networking demand in the quarter was strong as Broadcom saw customers build out their data center infrastructure ahead of deploying AI accelerators, Tan said.

“Our current order backlog for AI switches exceeds $10 billion as our latest 102-terabit-per-second Tomahawk 6 switch, the first and only one of its capability out there, continues to book at record rates. And this is just a subset of what we have. We have also secured record orders on DSPs, optical components like lasers, and PCI Express switches all to be deployed in AI data centers.”

Massive Revenue Growth Already Booked

Broadcom’s AI-related orders on hand for its XPU and its networking components are currently in excess of $73 billion today, which is almost half of Broadcom’s consolidated backlog of $162 billion, Tan said.

“We expect this $73 billion in the AI backlog to be delivered over the next 18 months,” he said. “And in Q1 fiscal 26, we expect our AI revenue to double year-over-year to $8.2 billion.”

Actually, Broadcom expects the backlog to only grow, Tan said.

“We fully expect more bookings to come in over that period of time, and so don’t take that $73 [billion] as the revenue that we ship over the next 18 months,” he said. “I’m just saying we have that now, and in that bookings have been accelerating.”

Non-AI Semiconductor Business Flat

Broadcom’s fiscal fourth quarter 2025 non-AI semiconductor revenue reached $4.6 billion, which was up only 2 percent year-over-year, Tan said. Broadband semiconductor revenue showed what Tan called “solid recovery,” while wireless was flat and all other end semiconductor revenue fell as enterprise spending continue to show limited signs of recovery, he said.

“Accordingly, in Q1 we forecast non-AI semiconductor revenue to be approximately $4.1 billion, flat from a year ago [and] down sequentially due to wireless seasonality,” he said.

Infrastructure Software Revenue Up On VMware Sales

Broadcom reported fourth fiscal quarter 2025 infrastructure software revenue of $6.9 billion, up 19 percent year-over-year and above the company’s outlook of $6.7 billion, Tan said. The total contract value of bookings in the fourth quarter exceeded $10.4 billion versus $8.2 billion. Broadcom ended the year with $73 billion of infrastructure software backlog, up from last year’s $49 billion. Infrastructure software revenue is slated to grow by a low double-digit percentage, he said.

“Here’s what we see in 2026,” he said. “Directionally, we expect AI revenue to continue to accelerate and drive more software growth and non-AI semiconductor revenue to be stable. Infrastructure software revenue will continue to be driven by VMware growth at low double digits. And for Q1 ‘26, we expect consolidated revenue of approximately $19.1 billion, up 28 percent year-over-year.”

Customers Will Not Develop Custom Chips Like Broadcom’s XPU

When asked by an analyst about industry chatter over the possibility that large customers are looking at custom tooling their own AI accelerators, Tan said he does not believe that will happen to any extent.

“Don’t follow what you hear out there as gospel,” he said. “[Developing AI accelerators] a trajectory. It’s a multi-year journey. And not too many players doing LLM want to do their own custom AI accelerator for very good reason.”

It is possible to use general-purpose GPUs or custom software and kernels to accelerate AI, Tan said.

"Now, will that mean that over time, they all want to go do it themselves,” he said. “Not necessarily. [The] technology in silicon keeps updating, keeps evolving. And if you are an LLM player, where do you put your resources in order to compete in this space, especially when you have to compete at the end of the day against merchant GPUs, who are not slowing down in their rate of evolution. So I see that this concept of customer tooling is an overblown hypothesis, which, frankly, I don’t think will happen.”