In a bold forecast that sent shockwaves through the semiconductor industry, chip designer Broadcom projected on Wednesday that its revenue from artificial intelligence components will surpass $100 billion by 2027. The announcement underscores the explosive demand for custom silicon just as the AI arms race among Big Tech intensifies.
Investors reacted enthusiastically, pushing the company’s stock up nearly 5% in after-hours trading. The surge was further fueled by Broadcom’s announcement of a new share repurchase program, authorizing up to $10 billion in buybacks through the end of the year.
Custom Silicon Takes Center Stage
Unlike rival Nvidia, which dominates the market with its off-the-shelf AI accelerators, Broadcom has carved out a lucrative niche by collaborating with tech giants to develop bespoke chips. The company works closely with clients such as Alphabet’s Google on their tensor processing units (TPUs) and with ChatGPT-maker OpenAI on their proprietary in-house processors.
Broadcom’s engineers essentially serve as the architects, translating early-stage designs into the physical blueprints that manufacturing powerhouse TSMC can then produce.
“Our visibility in 2027 has dramatically improved,” CEO Hock Tan revealed during a post-earnings call with analysts. “Today, in fact, we have line of sight to achieve AI revenue from chips in excess of $100 billion in 2027.”
This long-term clarity is unusual for a company that typically only guides one quarter at a time. Analysts view this as a strong indicator of sustained, significant growth in demand. The company’s immediate outlook is equally robust, with second-quarter revenue projected at approximately $22.0 billion—comfortably above the $20.56 billion analysts had anticipated. For the current quarter alone, Broadcom forecasts AI chip revenue of $10.7 billion.
Major Deals and Major Players
The company’s pipeline is filling up with massive contracts. Broadcom is expected to deliver a gigawatt’s worth of TPUs to AI startup Anthropic in 2026, with that demand tripling to 3 gigawatts in 2027. Meanwhile, the company is targeting the shipment of OpenAI’s first dedicated AI chip in 2027, aiming to deliver over a gigawatt’s worth of those processors.
These deal sizes are approaching the scale of those secured by industry titans. For context, Nvidia recently disclosed 5 gigawatts of sales to OpenAI, while AMD has signed agreements worth up to roughly 6 gigawatts with both OpenAI and Meta.
Addressing speculation about Meta’s commitment, Tan firmly dismissed reports that the social media giant was slowing its AI chip efforts. “Meta’s custom accelerator MTIA roadmap is alive and well,” he stated. “We’re shipping now.”
Record Earnings and a Promising Future
The rosy projections come on the back of a strong first-quarter performance. Broadcom reported a 29% revenue increase to $19.31 billion for the period ending February 1, edging past analyst expectations. Adjusted earnings per share hit $2.05, also beating estimates. AI revenue alone more than doubled to $8.4 billion, fueled by demand for both custom accelerators and networking equipment.
While the core semiconductor business is booming, the company’s infrastructure software segment showed signs of slowing, with growth of just 1% to $6.80 billion, slightly missing targets.
Nevertheless, the future looks bright. Earlier this year, Broadcom indicated it expects to sell at least 1 million chips based on its advanced stacked design technology by 2027—a new product line that could unlock billions in additional revenue. As Big Tech firms like Alphabet, Microsoft, Amazon, and Meta collectively plan to spend at least $630 billion on AI infrastructure this year, Broadcom is positioning itself as a formidable alternative in a market long viewed as Nvidia’s to lose.
About the Author
