In the next 12 months, revenue from artificial-intelligence platforms will grow dramatically as businesses climb with AI or get left behind, according to Jensen Huang, Nvidia's founder and chief executive.
In a press conference Wednesday at Nvidia Corp.'s NVDA, annual GTC developer conference, Huang said that generative AI apps like OpenAI s ChatGPT backed by a multibillion dollar investment from Microsoft Corp. MSFT has only accounted for a tiny, tiny single-digit percentage of revenue over the past 12 months.
It is hard to say that Huang expects revenue from generative AI to be quite large over the course of twelve months. The previous day, Huang told analysts that AI and its infrastructure-as-a-service offering are going to expand our business model. Read: How Nvidia plans to fuel the AI surge and a new era of chipmaking.
Several companies are hopping on board the AI train with Nvidia hardware at the center. Adobe Inc. ADBE launched its AI co-pilot product, Firefly, on Tuesday, while Alphabet Inc. s Google GOOG, GOOGL, Cloud Services launched its own AI called Bard and has signed a partnership with Nvidia for the chip maker to offer services on the server infrastructure that Google buys.
As for that infrastructure, Huang said that cloud service providers, also referred to as hyperscalers, face a challenge because the size of AI models is estimated to increase in size 10 fold every year. Those models require a lot of server power, and Huang expects that demand to boost hardware sales.
Data centers, like all data centers around the world, are suffering from the end of Moore's Law, Huang said. They can't continue to expand their computing without some way of acceleration, so Nvidia's accelerated computing solution helps them take workload that used to take thousands and thousands of CPUs and reduce that down to a few GPUs. Huang went on to his previous contention that Moore's Law is dead, noting that the law of doubling performance in chips every 18 months at the same cost was a story of the past. execs like Intel Corp. INTC, CEO Pat Gelsinger, contend that Moore's Law is alive and well. As the demand for server power grows as the AI market develops, one of the biggest challenges for hyperscalers is to achieve carbon neutrality.
A solution Nvidia offers is a software library that can be used to improve computational lithography, a process of etching billions and billions of transistors onto a silicon wafer that later becomes a chip.
The software library not only saves money but also reduces power consumption, according to Huang. The software reduces a monthlong process called masking to an overnight affair, which could save $20 billion, according to the CEO. The fab can reduce its carbon footprint by investing the saved power somewhere else.
In February, Nvidia Chief Financial Officer Colette Kress said that software brings in hundreds of millions of dollars in revenue, and that it is getting stronger. In its most recent fiscal year, Nvidia had total revenue of $27 billion.
Concerning the future of AI and its ethical use, Huang said he would cut off services for any customer who builds a malignant AI.
He said that we only sell to companies that do good, which brings to mind the Don t be evil motto of its partner, Google.
Nvidia shares have gone up 5.3% for the week, compared to gains of 2% on the Dow Jones Industrial Average DJIA, the S&P 500 SPX, and the tech-heavy Nasdaq Composite COMP, the tech-heavy Nasdaq Composite COMP, and the Dow Jones Industrial Average DJIA.