Key takeaways
- Nvidia is set to kick off its annual GPU Technology Conference (GTC) with workshops on Sunday and a keynote address from founder and CEO Jensen Huang on Monday.
- Industry leaders including OpenAI, Microsoft, Amazon and Google will participate in the GTC, which could highlight Nvidia’s artificial intelligence (AI) initiatives and partnerships.
- Nvidia is expected to provide details on the B100 Blackwell which could be its most capable AI GPU yet, such as price, performance and demand.
- The chipmaker could also provide more information on its sovereign AI initiatives, as data sharing has been thrust into the spotlight as the US government takes interest in Chinese companies ByteDance and TikTok.
NVIDIA (NVDA) is preparing to launch its annual edition GPU technology conference (GTC) with workshops on Sunday and a keynote address from founder and CEO Jensen Huang on Monday.
Analysts expect the chipmaker to share updates on its new products, partnerships with industry leaders and sovereign deal. artificial intelligence (AI) initiatives during the conference.
Update on the next AI chip, B100 Blackwell
The chipmaker is expected to provide more details on the B100 chip, which could be its most capable AI GPU yet.
The B100 chip is an upgrade of the H200 chip and is expected to be available sometime in 2024, based on Nvidia’s development roadmap released in October 2023.
“NVIDIA’s current flagship H200 chip in the high-performance computing market adopts the GPU Hopper architecture combined with HBM3e memory chips, considered the industry’s highest performance chip for AI computing,” according to TrendForce.
Investors could get new details about the chip, including price, performance and demand. Nvidia could also provide more color on the chip hardware, such as the liquid cooling component and DRAM memory.
“We expect NVDA’s new Blackwell architecture to deliver significant performance gains over Hopper,” Wedbush analysts said, noting that “NVDA has touted the H100 as delivering up to 9x faster training and 30x faster inference performance than A100”, the previous version.
Wedbush analysts said they are “interested in any additional information NVDA chooses to provide regarding the initial demand the company is seeing for the upcoming B100 (management has already characterized demand as initially expected to significantly exceed supply capabilities ).”
They added that the company would monitor “any incremental color regarding NVDA’s future GPU and AI products beyond the 2025 roadmap presented by NVDA to date.”
AI leaders gather
Other AI leaders are also expected to join the conference, which could provide investors with more insight into the AI space in general and the role Nvidia’s technology plays.
Nvidia partners including OpenAI, Microsoft (MSFT), Amazon (AMZN), Google’s parent alphabet (GOOGLE), Meta (META), Micronic Technology (UM), Oracle(ORCL), Super Micro Computer (SMCI), Dell (DELL) and Intel (INTC), among others, will participate in events and workshops during the GTC.
While Nvidia has established itself as one of the early leaders in the AI race, companies within its AI ecosystem have gained amid growing demand for AI technology.
Bank of America analysts said they expected to assess “the impact of growing competition from custom chips” affecting companies like Broadcom (AVGO) and Marvell (MRVL).
The conference could also shed light on competition in the merchant silicon market from Advanced Micro Devices (AMD) and Google. Bank of America noted that silicon chipmakers claimed their products were “cheaper (especially for AI inference) and/or less expensive (30-50% less than comparable NVDA hardware ). »
Nvidia’s Sovereign AI Efforts
During the conference, Nvidia may share details about its sovereign AI initiatives, as the scope of AI technology is rapidly expanding.
Bank of America analysts said they “expect NVDA to highlight TAM expansion opportunities beyond the four largest U.S. hyperscalers,” Microsoft, Google, Amazon and Meta.
They added that there could be a multi-billion dollar opportunity for Nvidia with “tier 2/3 webscalers (CoreWeave, Lambda Labs)” and “sovereign nations.”
Sovereign AI “refers to a country’s capabilities to produce artificial intelligence using its own infrastructure, data, workforce and enterprise networks,” Nvidia said.
Nvidia has spoken about its sovereign AI initiatives. The stance could help differentiate Nvidia, particularly amid concerns about potential national security risks from sharing data, with the U.S. government. consider forcing Chinese ByteDance will sell TikTok under penalty of banning the application in the United States
Ahead of GTC, analysts at Bank of America raised their price target for Nvidia from $925 to $1,100, while analysts at Wedbush increased their target from $850 to $1,000. Both companies gave Nvidia a “buy” rating.
Nvidia shares were up 1% Friday at $888.26 as of 3 p.m. ET. Despite a recent sell-off, the stock has gained more than 84% year to date.
