• Nvidia unveils next-generation Vera Rubin GPUs and "personal AI supercomputers" at record-breaking GTC 2025.
  • The company faces growing competition from AI chip startups and major tech firms developing in-house alternatives.
  • Geopolitical concerns prompt pledge of massive U.S. manufacturing investment despite potential margin impact.

Nvidia dominated San Jose this week with its annual GPU Technology Conference (GTC), drawing a record-breaking 25,000 attendees who packed workshops and panels to capacity.

While CEO Jensen Huang unveiled powerful new products in a show of strength, the event highlighted the growing challenges facing the AI chip giant despite its current market dominance.

Huang delivered an exhaustive sales pitch aimed at reassuring investors concerned about Nvidia's recent stock slide.

"The more you buy, the more you save," he told attendees during Tuesday's keynote. "It's even better than that. Now, the more you buy, the more you make."

The company's flagship announcement was its next-generation Vera Rubin GPUs, which Huang claimed will perform inference at roughly double the rate of Nvidia's current Blackwell chips.

This emphasis on inference capabilities directly addresses concerns raised by Chinese AI lab DeepSeek's highly efficient "reasoning" model R1, which prompted investor fears that Nvidia's powerful chips might become unnecessary for training competitive AI models.

However, Huang spent noticeably less time addressing the growing threat from AI chip competitors like Cerebras, Groq, and major cloud providers developing their custom silicon.

More concerning for Nvidia, key customers including OpenAI and Meta are actively pursuing in-house hardware development to reduce their dependence on Nvidia's products.

The conference also revealed Nvidia's response to potential tariff disruptions, with Huang pledging hundreds of billions in U.S. manufacturing investments.

While this diversification could shield the company from supply chain risks, it represents a significant cost that could impact Nvidia's profit margins.

Looking beyond its core chip business, Nvidia announced a new quantum computing center in Boston (NVAQC) and unveiled its "personal AI supercomputers" – the DGX Spark and DGX Station.

Though priced at thousands of dollars, Huang positioned these devices as the future of personal computing in the AI era.

Despite the confident presentation, investors appeared unconvinced, with Nvidia's share price dipping approximately 4% following Huang's keynote.

As one analyst noted, the market might have been expecting a surprise announcement or accelerated product timeline that never materialized.

As Nvidia navigates these challenges, the company's ability to maintain its leadership position will depend not just on technical innovation but also on managing geopolitical risks, competitor advances, and changing customer dynamics in the rapidly evolving AI landscape.

Edited by Annette George