- Marvell-Google AI chip talks target 2 custom ASICs, per Reuters Oct. 8.
- MRVL shares rose 4.6% to $75.92 in after-hours trading.
- Marvell data center revenue up 98% YoY to $1.27B in Q2.
Marvell Technology Inc. (NASDAQ: MRVL) shares surged 4.6% to $75.92 in after-hours trading on Oct. 8, 2024. A Reuters report fueled the rally, detailing Marvell-Google AI chip talks for two custom application-specific integrated circuits (ASICs) Reuters, Oct. 8, 2024 Yahoo Finance.
These ASICs target AI data center workloads. They optimize power efficiency and performance for hyperscale operations.
Marvell-Google AI Chip Talks Target Distinct Architectures
Sources familiar with the matter told Reuters the negotiations center on two chip designs Reuters, Oct. 8, 2024.
One architecture handles AI model training. The second focuses on inference tasks.
Marvell leverages expertise from past custom ASIC projects for hyperscalers Marvell AI solutions.
Google deploys its Tensor Processing Units (TPUs) for AI acceleration Google Cloud TPU documentation.
Marvell already supplies Ethernet switches and processors to major cloud providers. This positions the company to expand partnerships.
Custom AI Chips Challenge Nvidia's Market Dominance
Nvidia commands 80-90% of the AI accelerator market Omdia Q2 2024 semiconductor report.
Custom ASICs deliver 2-3x better power efficiency for tailored workloads versus general-purpose GPUs.
Google trains its Gemini models on TPUs. This approach saves billions in annual energy costs.
Meta develops MTIA chips. Amazon builds Trainium accelerators.
Marvell competes against Broadcom, which won a previous Google custom chip contract Bloomberg, Oct. 12, 2023.
Demand for high-bandwidth interconnects grew 150% in H1 2024 Dell'Oro Group report.
Investor Reaction Highlights AI Revenue Growth Potential
After-hours trading volume exceeded daily averages as investors reacted positively.
Marvell's data center revenue reached $1.27 billion in Q2 FY2025, up 98% year-over-year Marvell Q2 FY2025 earnings release, Aug. 29, 2024.
AI products accounted for 55% of total revenue. Management guided Q3 data center sales to $1.9 billion.
Alphabet reported $12.7 billion in Q2 capex, focused on AI infrastructure Alphabet Q2 2024 earnings release, July 23, 2024.
The PHLX Semiconductor Sector Index (SOXX) climbed 25% year-to-date on AI momentum Yahoo Finance SOXX.
Crypto Markets Link to AI Chip Demand Surge
The Crypto Fear & Greed Index registered 33 (Fear) on Oct. 8, 2024 Alternative.me.
Bitcoin traded at $60,804, up 0.7% with a $1.20 trillion market cap. Ethereum stood at $2,419 ($291 billion cap) per CoinMarketCap data at 18:00 UTC CoinMarketCap Bitcoin CoinMarketCap Ethereum.
Blockchain analytics firms deploy AI chips for rapid machine learning on on-chain data.
DeFi protocols integrate AI-driven risk assessment models to enhance efficiency.
Venture capital flows into AI-blockchain intersections despite cautious sentiment.
Marvell Positions for AI Ecosystem Dominance
Marvell evolved from storage controllers to AI networking specialists.
OCTEON DPU processors enable edge AI inference.
Custom chips pair with 800G Ethernet for data center fabrics.
Development cycles last 18-24 months, including design, TSMC tape-out, and validation.
Google seeks reliable suppliers for sustained high-volume production.
Future Outlook for Marvell-Google AI Chip Talks
A formal deal announcement may follow prototype testing.
Chips will undergo benchmarks in Google labs.
TSMC prepares 3nm and 2nm production ramps.
Marvell's Q3 earnings call in December 2024 could provide updates Marvell investor relations calendar.
Success here propels Marvell toward a $10 billion AI revenue target by 2027.
Frequently Asked Questions
What are the Marvell-Google AI chip talks?
Reuters (Oct. 8, 2024) reports Marvell in early discussions with Google for two custom ASICs targeting AI data center workloads.
How did Marvell-Google AI chip talks impact stock?
MRVL shares rose 4.6% to $75.92 after-hours on Oct. 8 [Yahoo Finance], reflecting optimism for AI revenue growth.
Why pursue custom AI chips in Marvell-Google talks?
ASICs offer better efficiency than GPUs for AI tasks like Gemini training, reducing power and costs in hyperscale data centers.
What is Marvell's role in AI semiconductors?
Marvell designs custom ASICs, processors, and Ethernet for AI infrastructure, supplying hyperscalers like Google.



