Nvidia faces a major turning point

Nvidia - the company that benefits the most from the AI wave is continuing to stand on the threshold of another important event called the global AI virtual conference (GTC).

Even though the semiconductor giant's latest AI chips have only just begun shipping, the tech world is already turning its attention to the next generation.

As the most valuable semiconductor company on the planet, the pressure on Nvidia to continue to innovate and grow is enormous.

Pressure to step up

GTC, once a modest event, has become a mecca for the AI world. From around 9,000 attendees before the pandemic, this year's event is expected to draw more than 25,000 experts and developers to Nvidia's headquarters.

Everyone gathered to listen to the "trump cards" that this semiconductor giant will unleash in the coming time. In 2024, the Blackwell AI chip was in the spotlight.

Concerns about Blackwell design flaws, leading to delayed deliveries, affecting investor confidence. Photo: NVIDIA.

According to the chip giant, processors based on Blackwell architecture, such as GB200, bring huge performance upgrades for businesses applying AI.

Specifically, the GB200's computing power is 20 petaflops, five times larger than the H100's 4 petaflops. Nvidia said this additional processing power will allow AI companies to train larger and more complex models.

However, it was only recently that the Blackwell began to be shipped in large numbers following delays associated with mass producing the complex design.

Rumors swirled in the industry about design flaws in the platform, with the initial delivery window of Q3 2024 now pushed to 2025. Half of NVIDIA’s investors questioned Blackwell’s position in a post-earnings call.

To "compensate," analysts predict that Nvidia CEO Jensen Huang will introduce an upgraded version of this chip line, called Blackwell Ultra, in an upcoming keynote.

New "trump card"

However, WSJ says that particular attention will be on Rubin, the next generation of AI chips that Nvidia has previously revealed. Deutsche Bank analyst Ross Seymore expects Rubin to deliver “very impressive performance improvements” over Blackwell.

Meanwhile, Citigroup expert Atif Malik noted that Blackwell was 30 times faster than the previous generation in AI inference.

“We do not rule out the possibility that Rubin will make similar improvements,” Malik wrote in a note to clients.

While Rubin-based products aren't expected to start shipping until 2026, analysts paint a promising picture for Nvidia's future.

The GPU architecture codenamed Rubin will be a notable "unknown" at GTC 2025. Image: NVIDIA.

They forecast the company's data center revenue to reach about $237 billion in the fiscal year ending January 2027. For comparison, that's more than double Nvidia's current size.

The report even estimates that this figure will exceed $300 billion two years later, representing an impressive average annual growth rate of 30% over four years.

However, these optimistic figures are entirely dependent on whether the world's tech giants continue to invest massively in AI infrastructure. These expenditures can only be justified if there is truly large demand from consumers and businesses for generative AI services.

Nvidia also faces a number of challenges. The risk of a global recession due to factors such as tariffs, geopolitical instability and inflation could affect demand for AI services.

Nvidia also faces many challenges when it comes to over-inflated AI company valuations. Photo: Paddle.

Competition from in-house chips designed by Nvidia’s biggest customers like Amazon and Google is also a concern, as are claims of performance breakthroughs from Chinese AI startup DeepSeek.

The R1's incredible performance could reduce the need for the expensive AI chip clusters that Nvidia is offering.

All of these factors have created an interesting backdrop for GTC 2025. The tech world still has high hopes for Nvidia and its latest chips.

However, at least the market has not "priced" such resounding successes with such certainty as before. Can Nvidia meet these expectations and continue to assert its leading position in the fierce AI race? The answer will probably be revealed at the GTC event.