Today, Nvidia announced its financial results for the first fiscal quarter of fiscal year 2025: revenue of US$26.044 billion, up 262% year on year; net profit of US$14.881 billion, up 628% year on year, up 21% month on month; non-GAAP net profit of US$15.238 billion, up 462% year on year and 19% month on month (Note: Nvidia's fiscal year is out of sync with the natural year; the end of January 2024 to the end of January 2025 is the 2025 fiscal year).
After the financial report was released, Nvidia's founder, president and CEO Huang Renxun, and executive vice president and chief financial officer Colette Kress attended the subsequent earnings conference call to interpret the highlights of the financial report and answer questions from analysts.
The following is the main content of the analysis question and answer session:
Bernstein Research Analyst Stacy Rasgon: I have a question about the Blackwell platform. The chip is now fully in production, what does this mean for shipping and delivery times? What does this mean for customers?
Huang Renxun: We will ship after a period of production, but our productive shipments will begin in the second quarter and gradually increase in the third quarter. Customer data centers should be able to use the Blackwell platform from the fourth quarter.
Stacy Rasgon: So the Blackwell platform can start contributing revenue this year?
Huang Renxun: We will see the Blackwell platform contribute a lot of revenue to the company this year.
UBS analyst Timothy Arcuri: I'd like to ask Hwang In-hoon how the company's deployment of the Blackwell platform will be different. How would a Blackwell deployment be different from that of the Hopper platform, given the nature of its system and the huge demand for performance in the market? I'm asking this because large-scale liquid cooling hasn't been implemented yet, and there are some engineering challenges both at the node level and within the data center. Will these factors of complexity prolong the transition to Blackwell? How does management see all of this progressing?
Huang Renxun: Blackwell has a variety of configurations. Blackwell is a platform, not a GPU. The platform supports air cooling, liquid cooling, X86 architecture, Grace processors, as well as the InfiniBand computer network communication standard, Spectrum X networking platform, and NVLink communication protocol, all of which I demonstrated at GTC (Nvidia Global Artificial Intelligence Conference for Developers). As a result, for some customers, they can upgrade their existing Hopper deployed data center, making it easy to transition from H100 to H200 to B100.
Whether electrically or mechanically, the Blackwell system is designed to be backwards compatible, and of course, the software stack running on Harper works well on Blackwell. We've also been pushing the entire ecosystem to prepare for liquid cooling. Over the past period, we have discussed Blackwell with ecosystem partners, including long-term communication with CSPs (cloud service providers), data center service providers, ODMs (original design manufacturers), system manufacturers, and supply chains, as well as communication with the cooling supply chain, liquid cooling supply chain, and data center supply chain. Everyone is ready for Blackwell's upcoming launch, and they are all looking forward to the extraordinary performance Grace Blackwell 2 can achieve.
Bank of America Merrill Lynch Analyst Vivek Arya: How can the company ensure that the market uses Nvidia products sufficiently, rather than early buying or hoarding due to tight supply, competition, or other factors? What systemic factors can convince investors that they can maintain good profits in the face of very large shipments?
Huang Renxun: I'll talk about the big aspects first, then answer your questions directly in detail. The market demand for GPUs and data centers is amazing. We are racing against time every day because of apps like ChatGPT and GPT 4.0, and the multi-modal trend is becoming more and more obvious, including growing AI models such as Gemini and Anthropic. All CSP operations consume huge amounts of GPUs.
In addition, there are thousands of generative artificial intelligence startups, distributed in various fields from multimedia to digital characters, various design tools, productivity applications, digital biology, audiovisual industry migration to video, etc., to help them train end-to-end models, expand the field of operation of autonomous vehicles, etc. The customer list was really long, and it put a lot of pressure on us to get the system delivered and up and running as quickly as possible. Of course, the ones I mentioned earlier do not include all sovereign artificial intelligence. They want to use their own natural resources, that is, data, to train their own regional models, and they are also under great pressure to establish these systems. Overall, demand is very high and has exceeded our supply capacity.
From a longer-term perspective, we're completely redesigning the way computers work; this is a platform shift. Of course, we've seen many people compare this transformation to several platform changes in the past, but time will clearly tell, and this transformation is far more profound than before. The reason is that today a computer is no longer just an instruction-driven device; it's a device that understands intention, not only how we interact with it, but also what we ask it to do. It has the ability to reason and iteratively infer, and is used to process plans and come up with solutions.
As a result, every aspect of computer operation is changing. Not only is it able to retrieve pre-recorded files, but it can instantly generate context-related intelligent answers. This will also change the world's computing stack, and even the computing stack of personal computers will revolutionize it. And this is just the beginning. Compared to our work in the lab and the results of our collaboration with startups, big companies, and developers around the world, what people see today is just the tip of the iceberg; future applications will be huge and extraordinary.
Morgan Stanley Analyst Joseph Moore: I understand the strong demand you just mentioned, the huge demand for H200 and Blackwell products. So as sales of these products gradually increase, does management expect the growth in sales of Hopper and H100 products to stop? Will sales of Hopper and H100 products slow down as people wait for new products to be released, or do they think there is enough demand for the H100 to maintain growth?
Huang Renxun: We have seen market demand for Hopper continue to increase this quarter, and we expect demand for Hopper to exceed supply for some time as the company switches to H200 and Blackwell. All companies are rushing to bring their infrastructure online; these actions can help them reduce costs and increase revenue, and they hope to do so soon.
(Continuously updated...)