Samsung to Manufacture Nvidia’s Groq 3 LPU; Shipments Expected in Q3

Samsung to manufacture Nvidia’s new inference chip Groq 3 LPU; shipments likely in Q3

Nvidia has announced that Samsung Electronics will manufacture its upcoming Groq 3 LPU inference chip, with shipments scheduled to begin in the third quarter of 2026. The partnership marks a significant step in Nvidia’s strategy to expand its AI inference hardware portfolio under the Vera Rubin platform, following its $20 billion licensing deal with Groq.

Key Details

• Manufacturing Partner: Samsung Electronics
• Process Technology: Advanced 4nm node
• Architecture: SRAM-based design, optimized for inference workloads
• Role: Functions as a decode-phase co-processor within Nvidia’s Vera Rubin AI platform
• Deployment: Integrated into 256-chip inference racks for ultra-low latency performance
• Shipments: Expected to start in Q3 2026

Strategic Context

The Groq 3 LPU builds on Nvidia’s December 2025 agreement with Groq, which included licensing and talent acquisition valued at $20 billion. By combining Groq’s inference-focused architecture with Samsung’s manufacturing expertise, Nvidia aims to address the growing demand for efficient large-scale inference solutions. The company projects $1 trillion in platform orders through 2027, doubling its earlier forecast.

Market Impact

The announcement had an immediate effect on investor sentiment, with Samsung shares rising 5.3%. The deal also strengthens Samsung’s role in the AI hardware ecosystem, complementing its leadership in memory technologies such as HBM4E. For Nvidia, the Groq 3 LPU positions the company ahead of competitors in inference hardware, adding to its dominance in AI training with GPUs.

Risks and Considerations

• Production Scale: Samsung must ensure smooth ramp-up on its 4nm process to meet shipment timelines.
• Market Adoption: Success depends on enterprise uptake of Nvidia’s Vera Rubin racks.
• Competitive Landscape: Rivals such as AMD and Qualcomm are also investing heavily in inference-focused chip development.

Conclusion

The Groq 3 LPU represents Nvidia’s boldest move yet to dominate AI inference, leveraging Samsung’s advanced manufacturing capabilities. With shipments slated for Q3 2026, the chip could redefine inference efficiency at scale and further cement Nvidia’s leadership in the trillion-dollar AI hardware market.