The market opportunity for high-end custom-design semiconductors used to power artificial intelligence projects is surpassing expectations. Analyst Harlan Sur from J.P. Morgan has increased his estimate for application-specific integrated circuits (ASICs), or custom chips, in this sector. His new range for the market size is $13 billion to $15 billion at the bottom end, and $15 billion to $18 billion at the top end.
The aggressive adoption of generative AI is a key driver behind the growing demand for custom chip AI compute accelerators. According to Sur, Broadcom and Marvell Technology will be the biggest beneficiaries of this resurgence in custom chip design.
As of Friday trading, Broadcom stock was down 1.5% at $814.45, while Marvell Technology shares were down 0.1% at $57.51.
Generative AI, which involves using text, images, and videos to create content, has emerged as a popular investment theme this year. OpenAI's release of ChatGPT in late 2020 sparked significant interest in this form of AI.
Sur predicts that the high-end ASIC market will experience an annual growth rate of 20% going forward. He identifies Broadcom as the market leader with a 35% share, followed by Marvell with a 12% share. Since June, Sur believes that Broadcom has secured two or three major new custom chip projects, while Marvell has also captured one or two new deals.
"We believe these customer initiatives are continuing to drive strong orders/design activity for both Broadcom and Marvell," Sur stated.
Sur has assigned an Overweight rating to Marvell stock, along with a $70 price target. He currently does not have a rating on Broadcom as he dropped coverage after it was announced that J.P. Morgan would be acting as a financial advisor to VMware in connection with its sale to Broadcom.
Marvell is expected to report its fiscal second-quarter results next Thursday.