Nvidia Accelerates AI Chip Production with New Packaging Tech to Meet Soaring Demand

Nvidia Corp NVDA plans to use Fan-Out Panel Level Packaging (FOPLP) technology for its GB200 AI server chips earlier than scheduled to address the production constraints of Chip on Wafer on Substrate (CoWoS) packaging at Taiwan Semiconductor Manufacturing Co TSM.

Industry insiders see FOPLP as a viable alternative to CoWoS due to TSMC’s limited CoWoS capacity and the rising demand for AI chips driven by generative AI applications.

A senior executive from China Wafer Level CSP stated that FOPLP could reduce costs and enhance capacity despite having weaker technical specifications than CoWoS due to a more extensive process size, the Digi Times reports.

Also Read: AMD Grabs 33% of Server CPU Market and Prepares for Major Processor Launch: Report

Packaging companies are preparing to offer FOPLP services, and Yole Group predicts its market share will grow from 2% in 2022 to 8% by 2028, driven by higher yields and cost efficiency.

Fan-Out Wafer Level Packaging (FOWLP) remains the mainstream fan-out packaging type.

Meanwhile, analysts hailed Nvidia’s earning beat last week as the semiconductor industry’s highlight. They noted Nvidia as the premier accelerated compute & Gen AI supplier backed by Blackwell chip production plans and U.S. Big Tech customers.

Nvidia stock gained over 184% in the last 12 months. Investors can gain exposure to the stock via VanEck Semiconductor ETF SMH

and Return Stacked Bonds & Managed Futures ETF Grizzle Growth ETF DARP.

Price Actions: NVDA shares traded higher by 0.89% at $1,149.10 premarket at the last check on Wednesday.

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Photo via Shutterstock