Last Monday night, Nvidia (NVDA 0.37%) CEO Jensen Huang gave the opening keynote speech to kick off CES 2025, which ran until Friday in Las Vegas.
Nvidia is the leader in providing chips -- primarily graphics processing units (GPUs) -- and related technology to enable artificial intelligence (AI) capabilities. So, Huang naturally spent much of his approximately 1.5-hour presentation on the topic of AI. He covered Nvidia's new AI-related products and partnerships along with how he sees the AI industry evolving.
There was much great content in Huang's speech, but one of the most exciting things for Nvidia stock investors was this comment: "AI agents [are] likely to be a multitrillion-dollar opportunity."
What are AI agents?
A group of AI agents is a "digital workforce," as Huang said last Tuesday at Nvidia's CES financial analyst conference.
The development of AI agents is now possible due to generative AI, a relatively new technology that greatly increases the possible use cases of AI. Generative AI's amazing capabilities were first demonstrated to consumers in late 2022 with OpenAI's release of its ChatGPT chatbot.
What differentiates a chatbot, which can be extremely useful in some situations, from an AI agent is the degree of autonomy. A chatbot can do things like answer questions, generate text, and help solve problems. But it is not capable of working independently or taking initiative, as an AI agent can do.
How is Nvidia involved in AI agents?
On last quarter's earnings call, CFO Colette Kress summarized Nvidia's involvement in AI agents:
Nvidia AI Enterprise, which includes Nvidia NeMo and NIM microservices, is an operating platform of agentic AI. Industry leaders are using Nvidia AI to build copilots and agents.
Working with Nvidia, Cadence [Design Systems], Cloudera, Cohesity, NetApp, Salesforce, SAP, and ServiceNow are racing to accelerate development of these applications with the potential for billions of agents to be deployed in the coming years. Consulting leaders like Accenture and Deloitte are taking Nvidia AI to the world's enterprises.
A couple of comments about this quote: First, while Nvidia AI Enterprise is the company's operating platform to create AI agents, it's not exclusively devoted to agentic AI. Second, the list of companies using Nvidia's technology to develop AI agents is not meant to be all-inclusive.
The process varies, but how it generally works is that large enterprises use Nvidia AI Enterprise to create AI agents of their tool or tools. The enterprises will then rent out these agents to their customers via the cloud. Moreover, companies across various industries are expected to use AI agents to improve their own operations.
Nvidia top management expects fast adoption of AI agents
Management expects "Nvidia AI Enterprise full-year revenue to increase over 2x from last year," Kress said on last quarter's earnings call. This fast growth is being driven partly by the rush among enterprises to develop AI agents.
There is indeed a rush because top execs in many industries know that their businesses will suffer if they are slower than their competitors to use AI agents to improve their own businesses, and -- in the case of cloud-based software companies -- to offer AI agents to their customers.
At Nvidia's CES financial analyst conference, Huang made this comment that highlights how fast he expects AI agents to be used in certain fields: "There are 30 million software engineers [globally]. Starting next year, if a software engineer in your company is not assisted with an AI [agent] you are losing already fast."
Along with software engineers, Huang expects that folks who develop marketing campaigns will be among early adopters of AI agents. Moreover, he added that all knowledge workers -- a number he pegged at about 1 billion worldwide -- will eventually be assisted by AI agents.
NASDAQ: NVDA
Key Data Points
Putting a "multitrillion dollar opportunity" in context
Huang didn't expand on what he meant by "multitrillion" dollars, but "multi" is generally defined as more than two. So, multitrillion should mean at least $3 trillion. Moreover, he didn't share how long he thought it would take the agentic AI market to reach that sum in annual revenue. But we know from his remarks that he expects this market to grow very rapidly.
Let's go with "multitrillion" referring to $3 trillion, which is a conservative assumption. Here is some data to help put this massive number in context:
- The global car market was worth an estimated $3.4 trillion in 2023.
- 2023 gross domestic product (GDP) for Germany, Japan, India, the U.K., and France -- the third through seventh largest economies in the world -- was about $4.5 trillion, $4.2 trillion, $3.6 trillion, $3.4 trillion, and $3.1 trillion, respectively, according to the International Monetary Fund (IMF).
Let's assume that Nvidia captures just 5% of what I'm estimating to be a $3 trillion agentic AI market. This is another conservative estimate, given Nvidia's dominance of AI-enabling chip and related technology. Five percent of $3 trillion is $150 billion. This would be nearly all new revenue since Nvidia is in the very early stages of making money from AI agents.
In its third quarter of fiscal year 2025 (ended Oct. 27, 2024), Nvidia generated revenue of $35.1 billion, and Wall Street expects its fiscal year 2025 (ends late January) revenue will grow 112% year over year to $129.1 billion.
These numbers illustrate how the AI agent market has the potential to absolutely turbocharge Nvidia's revenue growth, which is already impressive. Much higher revenue should lead to much higher earnings, which in turn, should help power Nvidia stock higher over the long term.
