By Zaheer Kachwala and Max A. Cherney
(Reuters) -Broadcom CEO on Thursday assuaged investor worries about AI chip demand with a strong second-quarter forecast and hinted about new potential customers that could boost revenue in a highly competitive market.
Its forecast eased concerns about the thirst for AI chips, a day after Marvell Technology’s spooked the market with a tepid forecast. Broadcom shares, which closed down 6%, surged 14% in extended trading after the results.
The chipmaker expects revenue of around $14.90 billion, compared with estimates of $14.76 billion, according to data compiled by LSEG.
Broadcom is seeing red-hot demand for its custom artificial intelligence chips from cloud computing companies looking for an alternative to the costly processors designed by Nvidia as they expand their AI infrastructure.
Broadcom CEO Hock Tan said the company expects revenue of $4.4 billion in the second quarter for its AI semiconductors as hyperscale customers invest in custom AI chips for data centers.
Analysts expect Broadcom to benefit further from large tech companies moving away from off-the-shelf chips to in-house processors as computing needs for AI tasks get more complex and personalized.
Broadcom now has four more hyperscale customers who are “deeply engaged” with it to create their own custom chips separate from the current three that use its processors, CEO Tan said in a post-earnings call.
These four are not included in its estimate of revenue opportunity of $60 billion to $90 billion in 2027, he said.
Reuters reported last month OpenAI is working with Broadcom to finalize its first custom chip design and reduce its reliance on Nvidia.
“The company is enabling hyperscalers and others who want to build custom AI accelerators to control their designs and costs,” said Anshel Sag, an analyst at Moor Insights & Strategy.
Broadcom is one of several companies that is evaluating Intel’s most advanced manufacturing process, known as 18A. It has run test wafers through the Intel factories that evaluate portions of chips, not complete designs.
“Broadcom is much better positioned compared to its peers as its exposure in AI market is relatively more diversified with multiple AI ASIC customers,” Summit Insights analyst Kinngai Chan said.
Application-Specific Integrated Circuit, or ASICs, are chips designed for specific tasks. Designing and manufacturing an ASIC helps companies save the amount of energy chip use and boost performance.
Broadcom reported first-quarter revenue of $14.92 billion, beating estimates of $14.61 billion. AI revenue surged more than 77% to $4.1 billion on strong adoption of its custom-made accelerators.
Revenue in its infrastructure software segment rose more than 47% to $6.70 billion, while analysts expected $6.49 billion.
(Reporting by Zaheer Kachwala in Bengaluru and Max A. Cherney in San Francisco; Editing by Arun Koyyur)
