GTC 2024: Optimism vs. NVIDIA Challenges

Nvidia's GTC 2024: A Record-Breaking Event

This year, Nvidia significantly impacted San Jose, drawing an unprecedented 25,000 participants to the San Jose Convention Center and the adjacent downtown area. The popularity of numerous workshops, presentations, and panel discussions was such that attendees were often compelled to stand or sit on the floor.

Organizers frequently had to direct the crowds to maintain order amidst the overwhelming attendance. Nvidia presently dominates the artificial intelligence landscape.

Current Market Position and Future Challenges

The company is experiencing record financial performance, boasting substantial profit margins and currently lacking any significant rivals. However, the near future presents considerable challenges for Nvidia.

These challenges include potential U.S. tariffs, the emergence of DeepSeek as a competitor, and evolving demands from key AI clients. These factors introduce unprecedented risk.

GTC 2024: Projecting Confidence

During GTC 2024, Nvidia’s CEO, Jensen Huang, focused on conveying a sense of assurance. He introduced a new generation of high-performance chips and innovative “supercomputers” designed for personal use.

The event also featured the presentation of appealing robotic technologies. This comprehensive presentation served as a sales effort directed towards investors concerned by Nvidia’s recent stock decline.

Incentivizing Investment

Huang emphasized the benefits of increased investment during a keynote address on Tuesday. He stated, “The greater your purchase volume, the more substantial your savings will be.”

He further elaborated, “Moreover, increased purchases translate directly into increased profits.” This message aimed to reassure investors and stimulate further investment in Nvidia.

A Surge in Inference Capabilities

At this year's GTC event, Nvidia primarily aimed to reassure both attendees and the wider public that the demand for its processors is expected to remain strong for the foreseeable future.

During his presentation, Huang asserted that a widespread miscalculation had occurred regarding the anticipated decline in scaling for traditional AI. The emergence of DeepSeek, a Chinese AI laboratory, and its highly effective “reasoning” model, R1, initially sparked concerns among investors.

These concerns centered on the possibility that Nvidia’s high-performance chips might no longer be essential for training competitive AI systems. However, Huang consistently maintained that computationally intensive reasoning models will, in reality, increase the demand for the company’s chips.

To demonstrate this, Nvidia unveiled its upcoming Vera Rubin GPUs at GTC, with claims of approximately double the inference speed – the process of running AI models – compared to its current flagship Blackwell chip.

Competitive Pressures on Nvidia

Huang dedicated less attention to the challenges posed by emerging companies such as Cerebras and Groq, alongside other providers of lower-cost inference hardware. A significant trend is the development of custom chips for inference, and even training, by nearly all major hyperscalers.

For example, Amazon Web Services (AWS) offers Graviton and Inferentia (reportedly offered with substantial discounts), Google utilizes Tensor Processing Units (TPUs), and Microsoft has Cobalt 100.

Furthermore, major technology companies heavily reliant on Nvidia, including OpenAI and Meta, are actively pursuing strategies to lessen their dependence through internal hardware development initiatives.

Success in these endeavors by these rivals would likely diminish Nvidia’s dominant position within the AI chip market. This potential shift in market dynamics may explain the approximately 4% decrease in Nvidia’s stock price following Huang’s keynote address.

Investors may have anticipated a further announcement – a “one last thing” reveal – or a more accelerated product launch timeline, neither of which materialized.

Navigating Tariff Concerns

During GTC 2025, Nvidia addressed anxieties surrounding potential tariffs.

Currently, the United States has not implemented any tariffs on Taiwan, the primary source of Nvidia’s chips. Huang asserted that such levies wouldn't inflict substantial harm in the immediate future.

However, he refrained from guaranteeing Nvidia’s complete protection from the broader, long-term economic consequences, regardless of their eventual manifestation.

Nvidia's Response to "America First"

It is evident that Nvidia has taken note of the previous administration’s “America First” policy.

Huang announced a commitment at GTC to invest hundreds of billions of dollars in U.S.-based manufacturing. This move aims to diversify the company’s supply chains.

Nevertheless, this substantial investment represents a significant expense for Nvidia, given its multitrillion-dollar valuation is reliant on maintaining robust profit margins.

Diversification of supply chains is crucial for mitigating risk.

• Investing in domestic manufacturing offers greater control.
• It reduces dependence on single sources.
• This strategy can buffer against geopolitical instability.

The long-term financial implications of this manufacturing shift remain to be seen.

Nvidia's Expansion Beyond Core Chips

In its pursuit of fostering growth in sectors beyond its primary chip business, Nvidia highlighted new investments in quantum computing at GTC, an area where the company has previously shown limited engagement.

During the inaugural Quantum Day at GTC, Huang extended an apology to the chief executive officers of prominent quantum firms for inadvertently contributing to a stock market dip in January 2025.

This occurred following his remarks suggesting the technology's practical applications were distant, potentially 15 to 30 years away.

Nvidia revealed plans to establish a new facility, the NVAQC, in Boston.

This center will focus on the advancement of quantum computing through collaborations with key players in both hardware and software development.

The NVAQC will be outfitted with Nvidia’s own chips, intended to facilitate the simulation of quantum systems and the development of models for quantum error correction.

Focus on "Personal AI Supercomputers"

Looking to the near term, Nvidia anticipates a new stream of revenue from what it terms “personal AI supercomputers.”

At GTC, the company introduced DGX Spark (formerly known as Project Digits) and DGX Station.

These systems are engineered to enable users to create prototypes, refine, and execute AI models of varying scales at the network edge.

While neither offering is inexpensive – with retail prices in the thousands of dollars – Huang confidently asserted their representation of the future of the personal computer.

“This represents the defining computer for the age of AI,” Huang stated during his keynote address.

“It embodies the future of computing, both in form and function.”

The market’s acceptance of this vision remains to be seen.