GTC felt more bullish than ever, but Nvidia’s challenges are piling up

Nvidia took the storm on San Jose this year, with the record breaking 25,000 presence at the San Jose Convention Center and the surrounding suburban buildings. Many workshops, discussions and panels were packed so much that people had to risk against the walls or sit on the floor – and the organizers were angry so that they screamed the command to stand in line properly.

Nvidia is currently sitting at the top of the AI ​​World The recording breaking financialAkash-high profit margin, and still a serious competitor. However, the upcoming months also keeps an unprecedented risk for the organization because it faces our tariffs, DipscAnd transfer priority from top AI customers.

This GTC 2025Nvidia CEO Jensen Huang tried to project confidence, unveiled Strong new chips, Personal “Super Computer,” And, of course, Really beautiful robot. It was a broad sales pitch – aims to investors the reeling from an NVEDIA nosedward stockThe

During a keynote on Tuesday, Huang said at one point, “The more you buy, the more you save”. “It’s better now now, now the more you buy, the more you will make” “

Assumption boom

After all, Nvidia attended this year’s GTC – and tried to assure the rest of the world to observe the rest of the world – the demand for chips will not decrease soon.

During his keynote, Huang Claimed that almost “the whole world made it wrong” Opened Conventional AI Scaling falls out of consumptionThe Chinese AI Lab DIPSEC, which is early this year Revealing a highly skilled “logic” model Has been called R 1Investors encourage fears that Nvidia Monster Chips may no longer be needed for competitive AI training.

However, Huang repeatedly emphasized that electricity-hungry rational models would actually reduce the demand for company chips, not less. That’s why in GTC, Huang showed The next line of Nvidia Vera Rubin claims the next line of Rubin GPUs claim that they will perform almost twice the rate of Nvidia’s current best blackwel chip (that is, the AI ​​models will run).

The threat to spending less time to address the Nvidia business Huang is on the occasion of the hardware and cloud suppliers of cerebra, grock and other lowly expensive estimates. Although almost every hypersscale training is developed a custom chip for estimation. The AWS contains Graviton and Inferntia (Which it is aggressively exempt), Google has TPU, and Microsoft has Cobalt 100The

Along with the same vein, Tech giants are currently very dependent on Nvidia chips with OpenAI and Meta, are trying to reduce these relationships In-house Hardware Attempts. If they – and the other rival to the aforesaid – it is almost definitely definitely weaken the NVIDIA Strangoldholdold in the AI ​​chips market.

This is why Nvidia shares prices have dropped Huang’s main statement followed by about 4%. Investors are probably “a last thing” – or probably hope for a accelerated launch window. In the end, they didn’t get it.

Tariff

Nvidia also sought to eliminate concerns about tariffs in GTC 2025.

The United States has imposed any tariff on Taiwan (where Nvidia gets most of its chips), and Huang Claim The tariffs will not “significant damage” in a short time. He stopped with the promise that Nvidia would be protected from long-term economic impact, but-the way they eventually form.

The Trump administration’s “America First” message with Nvidia Huang has received clearly GTC Spending several hundred billion dollars to produce in the United States when the company will help diversify its supply chains, it is also a huge cost for Nvidia, which depends on the multi-reliever-doller assessment of healthy profit margins.

New business

Since it looks at the seeds and development of the business other than its original chips line, Nvidia in the GTC attracted his new investment to Quantum, which is an industry that the company has neglected Histor. On the first Quantum Day of GTC, Huang apologized He suggested to CEOs of Major Quantum companies to crash a minor stock in January 2021 that the technology would not be very effective for the next 15 to 30 years.

Tuesday, Nvidia Declaration It will open a new center in Boston, NVAC, to advance quantum computing with the “top” hardware and software markers. The center must be equipped with Nvidia chips, which the company says that the researchers will enhance the quantum system and models needed to correct the quantum error.

More immediately, in the future, Nvidia sees what it is “Personal AI Super Computer” is being called As a potential new revenue builder.

In GTC, agency Opened DGX spark (previously called Project) And DGX stations, both users are designed to allow prototypes, delicate-sun and AI models to run on the edge of different sizes. Both are not cheap – they are retail for thousands of dollars – but Huang courageously announced that they represent the future of the private PC.

Huang said during his keynote, “This is a computer of the AI ​​era.” “This should be what computers see and the computers will run it in the future.”

We will soon see if the customers agree.