NVIDIA has swept San Jose this year, with 25,000 participants flocking to the San Jose Convention Center and surrounding downtown buildings. Many workshops, speeches and panels were so crowded that people had to lean against walls or sit on the floor—and suffered from anger from organizers yelling orders to get them in line properly.
Nvidia is currently at the top of the AI world, Record financehigh profit margins, no serious competitors yet. But the next few months also present unprecedented risks for companies when facing U.S. tariffs. DeepSeekand shift priority from advanced AI customers.
exist GTC 2025NVIDIA CEO Jensen Huang tried to present confidence, unveiled Strong new chips,,,,, Personal “supercomputer”, certainly, Really cute robot. This is an exhaustive sale designed to investors from Nvidia nose.
“The more you buy, the more you save,” Huang said in his keynote on Tuesday. “It’s even better than that. Now, the more you buy, the more you make.”
Prosperity of reasoning
On top of that, NVIDIA is trying to ensure that the demand for chips to be watched by attendees and the rest of the world doesn’t slow down anytime soon.
During the keynote speech, Huang Claiming that almost “the whole world is wrong” exist Traditional AI zooming is reduced from fashion. DeepSeek, China’s artificial intelligence laboratory, earlier this year Released an efficient “inference” model Called R1raising concerns among investors that Nvidia’s monster chips may no longer require training for competitive AI.
But Huang has repeatedly insisted that a desire for a strong reasoning model will actually drive more demand for the company’s chips, rather than less. That’s why in GTC, Huang Xuanyao NVIDIA’s next line, Vera Rubin GPU, claims they will reason with twice the speed of NVIDIA’s current best Blackwell chips (i.e., running an AI model).
Threats to Nvidia businesses take less time, such as Cerebras, Groq and other upstarts such as low-cost inference hardware and cloud providers. Almost every high standard person is developing a custom chip for reasoning, even without training. AWS has Graviton and Pebleentia (It is reportedly actively discounting), Google has TPU, Microsoft has Cobalt 100.
Along the same lines, tech giants currently relying heavily on NVIDIA chips including OpenAI and Meta are looking to pass internal hardware effort. If they and other competitors mentioned above succeed, it almost certainly undermines Nvidia’s shackles in the AI chip market.
Maybe it’s why NVIDIA shares fall After Huang’s keynote speech, about 4%. Investors may have been hopeful for the “last thing” or perhaps an accelerated release window. In the end, neither of them.
Tariff tensions
Nvidia is also trying to alleviate GTC’s 2025 tariff concerns.
The United States has not imposed any tariffs on Taiwan (Nvidia has obtained most of its chips) and Huang claim In the short term, tariffs will not cause “significant harm”. However, he no longer promises to protect NVIDIA from long-term economic impacts, but whatever form they end up taking.
NVIDIA apparently received the “America First” message from the Trump administration, with Huang Committed on GTC To spend hundreds of billions of dollars on U.S. manufacturing, this will help the company diversify its supply chain, which is also a huge cost for NVIDIA, whose huge valuation depends on healthy profit margins.
New business
GTC’s NVIDIA, in addition to seed cultivation and growing its business, has attracted attention to Quantum’s new investment, which the industry has historically overlooked. On the first quantum day of GTC, Huang apologizes For the CEO of a large quantum company, he proposed in January 2025 that the technology will not be very useful in the next 15 to 30 years.
Tuesday, Envidia Announce It will open a new center in Boston, NVAQC, to advance quantum computing in partnership with “leading” hardware and software markers. Of course, the center will be equipped with NVIDIA chips, which the company says will enable researchers to simulate the models needed for quantum systems and quantum error correction.
In a more direct future, Nvidia sees what it is Called “personal AI supercomputer” As a potential new revenue manufacturer.
In GTC, company emission DGX Spark (formerly known as Project Number) and DGX stations, both of which allow users to prototype, fine-tune and run AI models in various size ranges at the edge. Neither of these two are very cheap—they retail for thousands of dollars—but Huang boldly claims that they represent the future of personal PCs.
“This is computers in the AI era,” Huang said in his keynote speech. “This is what a computer should be, and this is what the computer will run in the future.”
We will see soon if the customer agrees.
