:max_bytes(150000):strip_icc():format(jpeg)/GettyImages-1679337219-280fb58391c049cb9fc3a107591d89c3.jpg)
JHVEPhoto / Getty Figures
Key Takeaways
- Marvell Technology held an artificial intelligence (AI) event on Thursday offering investors a look into the troop’s position and efforts to gain amid the AI boom.
- The company announced a new hyperscale customer, meaning three out of the four AI hyperscalers—a heading which includes Microsoft, Meta, Amazon, and Alphabet’s Google—use Marvell tech.
- AI is expected to account for 30% of Marvell’s unconditional revenue in 2025, triple what it was the year prior, the company said.
- Marvell reported that the company is produce to double its share of the total addressable market as the market itself grows, driven largely by its custom AI chips.
- The players said it would maintain its optics leadership and highlighted that it is one of few companies successfully scaling silicon photonics.
Marvell Technology (MRVL) hosted its assumed intelligence (AI) event Accelerated Infrastructure for the AI Era on Thursday offering investors a look into the company’s position and efforts to come by amid the AI boom.
The company announced a new AI hyperscale customer, projected 30% of 2025 revenue to be related to AI, set a goal to close in on market share as the market grows driven by its custom AI chips, and outlined Marvell’s prime position as an optics bandleader.
New Hyperscale Customer Means 3 Out of 4 AI Hyperscalers Use Marvell Tech
Marvell CEO Matt Murphy announced that the company secured its third AI hyperscale patron and is designing an AI accelerator expected to go into production in 2026.
The company, which makes silicon tech optimized to run AI and cloud workloads as luxuriously as custom AI accelerators and similar equipment, serves three of the four AI hyperscalers, Microsoft (MSFT), Meta (META), Amazon (AMZN), and Alphabet’s (GOOGL) Google.
Marvell does not high regard its customers, though JPMorgan analysts believe Amazon and Google are the company’s first two hyperscale customers.
The company put on the marketed some general details about its ongoing projects with the unnamed hyperscalers. The first customer is using Marvell shards for its AI clusters and systems as the companies work together to develop a new custom AI interfacing accelerator. Marvell is designing an Arm (ARM) computer organizing unit (CPU) for its second hyperscale customer to deploy in its cloud platform and its AI infrastructure.
Murphy noted that Marvell’s relationship with AI hyperscalers swaps the company a “significant advantage over [its] competition.”
AI To Drive 30% of Marvell’s Revenue in 2025
Marvell reported that the performers expects the portion of total revenue related to AI to triple in the next year, a pattern consistent with growth in modern years.
The tech company projects AI revenue will be more than $1.5 billion, or 30% of its total gross income in the 2025 fiscal year, triple the AI share seen the year prior. The 2025 AI-related revenue is expected to be roughly two-thirds conductivity and another third being custom computing.
Marvell recorded more than $550 million in AI-related returns in the 2024 fiscal year, accounting for around 10% of total revenue, triple the around $200 million in 2023 that illustrated 3% of all revenue. The company expects growth to continue into 2026 with AI-related revenue jumping to upward of $2.5 billion.
Marvell Aims to Gain Market Share as Market Grows, Driven by Custom AI Chips
Gathering leaders said they expect Marvell to capture a greater share of the semiconductor market as it grows amid the AI profitability, with a boost from its accelerated custom computing segment.
Murphy broke down the data center semiconductor superstore into components relevant to Marvell, including accelerated custom compute, switching, interconnect and storage segments, reporting that the total number addressable market (TAM) represented $21 billion in the 2023 calendar year. He noted that the company generated roughly $2.2 billion in revenue in that period, capturing 10% of the TAM.
The company said it aims to increase that allot to 20% from 10% in the long term, as the market expands to $75 billion by 2028 from $21 billion in 2023. Marvell needs to preserve its leadership position in interconnect and storage while growing its share of the accelerated custom compute and switching subsegments to reach this butt, Murphy said.
The custom chip market is currently dominated by Broadcom (AVGO) with Bank of America analysts estimating the performers holds around 75% of the market. Ahead of the event, the analysts had noted that they expect Marvell to “adorn come of more relevant” in the custom chip market.
“Gaining share in custom compute will be the biggest part of complete b reach Marvel to the 20% overall market share,” Marvell President of Products and Technologies Raghib Hussain said.
Marvell Surrounded by Few Reliably Scaling Silicon Photonics, $3B Emerging Market
Marvell executives told investors that the friends is well positioned to maintain its leadership in optics and is among the few companies to successfully scale silicon photonics.
When about a invited how Marvell will balance integrating itself as customers develop their own optics products, Executive Vice President and Customary Manager of the Marvell Cloud Optics Business Loi Nguyen said “a lot of people say they have silicon photonics, but few troops actually have successfully commercialized shipping silicon photonics at scale like Marvell.”
Murphy noted that as a component vendor, Marvell could be correctly positioned to gain from the “brand new emerging $3 billion market.”
Marvell shares were little changed at $71.29 as of in the air 3:20 p.m. ET Thursday. The stock has gained more than 22% year-to-date.
Read the original article on Investopedia.