- The analyst said that Nvidia is now turning its focus to inference architecture as the next catalyst for the company.
- Nvidia stated that its architecture can deliver up to 35 times higher throughput per watt when Groq 3 LPX racks are paired with its next-generation Vera Rubin GPUs.
- The company also expects Groq to help increase the token output per second by 15 times.
Nvidia Corp.’s (NVDA) $20 billion acquisition of chip startup Groq will be the AI bellwether’s next revenue catalyst, according to IO Fund’s Beth Kindig.
In a note on Friday, Kindig highlighted that the Groq acquisition is aimed at driving up token usage, which would boost the company’s revenue and profits.
“The 256-chip LPX rack introduces Groq’s unique SRAM‑based architecture that allows Nvidia to offload decode‑phase workloads and massively increase token throughput,” Kindig said.
Nvidia shares were down more than 1% in Friday’s opening trade. Retail sentiment on Stocktwits around the company trended in the ‘bullish’ territory at the time of writing.
Parallels To Mellanox Acquisition
Kindig added that Nvidia’s Groq deal has some parallels to the company’s acquisition of Mellanox.
Nvidia acquired networking products maker Mellanox Technologies in 2020 for $6.9 billion, which helped the company clear the networking bottleneck with its GPUs. Kindig states that this allowed the company to turn its accelerators into clusters by removing the limiter at the time, which was scale-out networking.
The analyst added that the Groq acquisition is also on similar lines, allowing Nvidia to remove the current limiter, which is inference throughput per watt.
“Nvidia is preparing to position its GPUs to be among the best inference options available, utilizing Groq’s unique SRAM-based architecture to significantly turbocharge token throughput and accelerate inference performance,” she added.
Kindig said that Nvidia is now turning its focus to inference architecture as the next catalyst for the company.
Nvidia Touts 35 Times Higher Throughput Per Watt
Nvidia stated that its architecture can deliver up to 35 times higher throughput per watt when Groq 3 LPX racks are paired with its next-generation Vera Rubin GPUs.
The company also expects Groq to help increase the token output per second by 15 times. “If these claims hold true, then cheaper inference will unlock more usage, and more usage should lead to higher revenue and higher profits as the AI monetization wave plays out,” Kindig added.
She said that Nvidia is positioning Groq 3 LPX as a “token accelerator” that works in tandem with the Vera Rubin GPUs, with the company’s eyes set on a multi-agent future.
“As the AI industry now faces power and infrastructure constraints rather than compute, the key differentiator in the upcoming AI inference monetization wave will be how to extract the highest number of tokens per megawatt to maximize revenue,” she said.
NVDA stock is down 6% year-to-date, but up 49% over the past 12 months. The SPDR S&P 500 ETF Trust (SPY) is up 16% over the past 12 months, while the Invesco QQQ Trust (QQQ) is up 23%.
For updates and corrections, email newsroom[at]stocktwits[dot]com.