On the evening of January 6, local time, Jensen Huang, the CEO of NVIDIA, delivered an impressive 90-minute keynote address at a major conference, unveiling multiple innovative products spanning various technological sectors including graphics cards, world modeling, autonomous driving, and roboticsThe anticipation around this event had reached a fever pitch within the market, made evident by NVIDIA's soaring stock prices which had recently hit an all-time high, valuing the company at approximately $3.66 trillion before Huang took the stage.

However, following his presentation, NVIDIA's stock market performance did not quite reflect the brilliance of Huang's leather jacketOn January 7, shares dropped by 6.22%, settling at $140.14. This steep decline wiped out the previous day's gains of 3.4% and marked the largest single-day drop in four monthsAnalysts pointed to a number of factors contributing to this downturn, particularly noting that Huang provided scant details about the company’s most lucrative business: chips used for training AI modelsHis only indication was that the Blackwell AI processor had entered full production, leaving investors craving for more in-depth updates.

Moreover, Huang’s omission of any information about the developmental progress of the next-generation GPU platform, Rubin, seemingly crucial to the future of NVIDIA, contributed significantly to the market's cooling expectations.

At the outset of his address, Huang introduced the latest addition to the RTX graphics card series, the RTX 5090. Being primarily aimed at gamers, this represents a historical revenue source for NVIDIA, which brought in a staggering 47% of its total income during the 2021 fiscal year, just before the dawn of large-scale AI modelsThe RTX 5090 was noted to boast an astounding 92 billion transistors, delivering a computational performance of 4000 TOPS—almost double that of its predecessor, the RTX 4090.

The announced retail price for this gaming powerhouse reached a lofty $1,999, and it was set to hit the market as early as January of this year

Advertisements

Furthermore, pricing for laptops equipped with the RTX 5090 also found its way into the limelight, with the devices priced at $2,899 anticipating market release by March.

However, the launch of gaming graphics cards was merely an appetizer compared to the main course—the revenue contributions from NVIDIA's data center division have surged to 90% following the AI boomThus, investors remained attuned to the advancements NVIDIA was making with its AI chips.

Next, Huang hopped onto the AI bandwagon, introducing NVIDIA's version of AI agents, branded “Blueprint for AI Agents.” This platform operates on NVIDIA's Metropolis system, allowing developers to harness visual perception capabilities to generate and analyze video content at speeds thirty times quicker than traditional real-time methods.

Huang posited a future where every company's IT department would essentially become the human resources division for AI agents, a prospect which he claimed could present opportunities exceeding a trillion dollarsNotably, in recent months, there have been whispers about NVIDIA entering the PC market with AI PC processorsIn response, Huang confirmed that NVIDIA is indeed working on integrating AI technologies into personal computing, starting with devices operating on Windows 11 and utilizing the Windows Subsystem for Linux.

As generative AI applications begin to proliferate, physical AI emerged as a focal aspect of Huang’s conferenceHe launched “NVIDIA Cosmos,” a foundational model designed specifically for robotics and autonomous driving technologiesThis tool provides a means of generating highly realistic, physically plausible video for training purposes without the costly need for real-world data collection.

Much like large language models have driven innovations in platforms like ChatGPT, world models stand to serve as a fundamental engine for smart driving and roboticsCosmos would leverage AI alongside generative technologies to supply "synthetic data" for training and testing robots and autonomous vehicles.

Huang heralded that Cosmos could potentially have a transformative impact on the robotics and industrial AI sectors akin to what Llama3 (the large language model launched by Meta) has done for enterprise AI

Advertisements

However, he was quick to address that Cosmos requires further data and optimization before reaching a maturity level comparable to ChatGPT.

Turning his attention to practical applications of Cosmos, Huang emphasized the vital need for understanding the physical world, especially within the realms of autonomous driving and AI robotics, where data must be transformed into a machine-readable language, ultimately yielding visual output.

With predictions from NVIDIA projecting that the humanoid robotics market could balloon to $38 billion over the next two decades, the potential for this technology is significant.

In the closing moments of the keynote, during the much-anticipated “One More Thing” segment, Huang unveiled an intriguing product—Project Digits, a portable AI supercomputer priced at $3,000. Slightly larger than Apple's Mac Mini, this device is touted to deliver data center-equivalent compute powerIt supports models with up to 200 billion parameters, effectively allowing personal computers to function like miniature data centers.

“This is an AI supercomputer running an entire stack of NVIDIA's AI technologies,” Huang stated, asserting that this device could harness substantial AI processing capabilities for educational and research purposes.

Despite Huang’s ambitious showcase of flagship graphics cards, super PCs, world models, and advancements in both autonomous driving and robotics, NVIDIA's stock did not mirror this excitementThe downturn in stock performance could partly be attributed to the overall drop in the US stock market on January 7, where major indices, including the NASDAQ and S&P 500, saw significant declines.

Nevertheless, the prevailing sentiment among investors appears to be that NVIDIA's struggles stem more from internal factors than external pressuresSpecifically, stakeholders were looking for details regarding the advancement of the Blackwell chip and insights on the next-generation GPU platform, Rubin

Advertisements

With substantial numbers of NVIDIA call options trading hands, much attention is trained on how NVIDIA responds in the months to come.

While Huang set ambitious market targets, such as the mass adoption of humanoid robots, automation factories, and a significant increase in autonomous vehicles, these market potentials could take time to materialize, leaving their tangible impacts challenged by NVIDIA's sky-high valuation.

As industry trends shift, many speculate that the slowing of the so-called “Scaling Law,” vital to data and computational growth, could serve as a precursor to a reevaluation of NVIDIA's worthOpenAI co-founder Ilya Sutskever has previously noted a diminishing ROI in large-scale language model investments, pointing to increasingly complex challenges faced by AI labs.

Even within China's burgeoning “AI Six Dragons,” rumors have emerged recently about cutting back on large-scale pre-training investmentsLee Kai-Fu, founder of Sinovation Ventures, directly expressed in a recent interview a departure from the pursuit of Artificial General Intelligence (AGI).

Ultimately, NVIDIA's extraordinary market value over the past couple of years has been predicated on sky-high expectations for growth within the AI sectorConsequently, the company has accelerated its GPU release schedule from every two years to annually to keep pace with this elevated growth trajectory.

In spite of the stock plunge, numerous analysts maintain an optimistic outlook for NVIDIA’s futureRosenblatt Securities analyst Hans Mosesmann upheld a buy rating with a target price of $220, signaling an upside potential of over 47% based on NVIDIA’s peak prices.

Meanwhile, Wedbush analyst Daniel Ives proposed that Wall Street may have underestimated NVIDIA's growth prospects by at least 30%, asserting that Huang’s presentation substantiated NVIDIA's intention to amplify its “substantial technological lead.” Ives projected the burgeoning opportunities within the robotics and autonomous driving markets could yield an additional trillion-dollar market for NVIDIA in the coming years, suggesting that NVIDIA’s market cap could eventually exceed $4 trillion and could see itself approaching $5 trillion within the following 12 to 18 months.

Advertisements

Advertisements