NVIDIA Corporation (NVDA.US) partners with Groq: The era of AI inference is approaching, the high-profit myth may face a test.
In the context of GPUs almost monopolizing AI computing power, why does NVIDIA still want to "introduce other people's chips"?
Global AI chip leader NVIDIA Corporation (NVDA.US) recently purchased a license from the artificial intelligence chip startup Groq, sparking market discussions. Although the specific terms of the transaction have not been disclosed, a key question has arisen: in the context of GPUs almost monopolizing AI computing power, why does NVIDIA Corporation still need to "introduce others' chips"?
The market generally compares this move to Meta Platforms' acquisition of Instagram in 2012. This is a typical defensive transaction, as Meta's founder, Mark Zuckerberg, was concerned about emerging social platforms threatening Facebook's core position. It was later proven that this strategic move not only mitigated competitive risks but also brought a younger user base to Meta. Now, NVIDIA Corporation's investment in Groq is also seen as a preemptive defense strategy.
However, this choice also implies a shift in NVIDIA Corporation's assessment of future competition. Analysts believe that this move, while maintaining the company's long-term growth potential, also indicates that NVIDIA Corporation is preparing for potential declines in profitability.
Over the past two years, NVIDIA Corporation's GPUs have been the "main engine" of the AI revolution. Since the release of ChatGPT in November 2022, the company's quarterly revenue has increased nearly tenfold. With a gross margin of 73% in the latest quarter, it further highlights its nearly unassailable market position in the AI training field.
The advantage of GPUs lies in their "versatility," as they can be used for both large-scale model training and inference tasks (such as chatbots, image generation, etc.). However, as AI applications are implemented, inference is becoming the fastest-growing part of the computing demand, and competition is accelerating here.
As early as 2015, Alphabet Inc. Class C (GOOG.US, GOOGL.US) launched its proprietary TPU, while Microsoft Corporation (MSFT.US) and Amazon.com, Inc. (AMZN.US) have also entered the custom AI chip market. Groq is one of these representatives, with its founder previously being a member of Alphabet Inc.'s TPU team and will now work for NVIDIA Corporation after the deal.
Currently, most competitors have given up challenging NVIDIA Corporation on the "training" front, but in the inference field, specialized chips are showing advantages. Bank of America Corp analyst Vivek Arya points out that this transaction indicates that NVIDIA Corporation has realized that even though GPUs dominate the training market, the rapid rise of inference may require more specialized chips.
Groq's flagship product, the Language Processing Unit (LPU), is designed for efficient inference. Previously, Groq's prospects were dim due to many potential clients already having their own custom inference chips. The authorization cooperation with NVIDIA Corporation has brought new opportunities for Groq.
From an industry integration perspective, the combination of the two sides is not a simple competitive relationship. In the future, the market may see NVIDIA Corporation servers that integrate both GPUs and LPUs. Groq will access NVIDIA Corporation's AI software ecosystem accumulated over the past two decades, and its own inference software capabilities may also benefit NVIDIA Corporation, allowing the two types of chips to complement each other, similar to the relationship between Facebook and Instagram in the past.
But this cooperation is also seen as a turning point signal in the AI investment wave. If inference becomes the main battlefield of AI computing power, and core hardware shifts more towards specialized chips like LPUs and TPUs, NVIDIA corporations must be "present."
In the short term, investor sentiment remains optimistic. NVIDIA Corporation's stock price rose by 1.02% on Friday, indicating that the market is not yet concerned about the transaction. However, in the medium to long term, if the proportion of Groq LPUs in NVIDIA Corporation's product portfolio increases, its high gross margin will inevitably face downward pressure.
Related Articles
.png)
HK Bull/Bear Outstanding Qty Ratio(59:41) | December 27th

Overnight US stocks | The three major indexes fell, with the S&P 500 index hitting a new high intraday. Spot silver prices surged more than 10%.

US Stock Market Move | Tesla, Inc. (TSLA.US) fell more than 2%, with bulls expecting fourth-quarter electric vehicle deliveries to be below 440,000 units.
HK Bull/Bear Outstanding Qty Ratio(59:41) | December 27th
.png)
Overnight US stocks | The three major indexes fell, with the S&P 500 index hitting a new high intraday. Spot silver prices surged more than 10%.

US Stock Market Move | Tesla, Inc. (TSLA.US) fell more than 2%, with bulls expecting fourth-quarter electric vehicle deliveries to be below 440,000 units.

RECOMMEND

Not Just “Power Shortages,” Delays Will Become The Key Theme For U.S. Data Centers In 2026
26/12/2025

Hang Seng Index Rises 33% This Year, Best Five‑Year Performance; Multiple Institutions Forecast Breakthrough Above 30,000 Next Year
26/12/2025

Gold Rally Has Further To Run, JPMorgan Bullish: Prices Could Reach USD 5,055 By Year‑End 2026
26/12/2025


