At its GTC conference, Nvidia projected a significant growth opportunity for its investors, estimating that data center infrastructure capital expenditure could surpass $1 trillion by 2028. CEO Jensen Huang highlighted this forecast, suggesting a considerable potential increase in Nvidia’s stock value if these projections materialize. Despite this optimistic outlook, the investors’ response to the announcement remained muted.
The projection of $1 trillion in data center infrastructure spending represents a potential acceleration in the sector, which would be advantageous for Nvidia. Nvidia’s graphics processing units (GPUs) have become integral to building artificial intelligence (AI) infrastructure due to their efficient data processing capabilities.
Nvidia outlined its expectations for 2024, estimating that data center infrastructure spending would reach approximately $400 billion. The company reported total revenue of $130.5 billion for its past fiscal year, with $115.2 billion attributed to its data center segment. Research firm Dell’Oro Group recently estimated that data center infrastructure spending in 2024 reached $455 billion, positioning Nvidia to capture around 25% to 30% of the expenditure.
If Nvidia maintains its current market share, this could translate to $250 billion to $300 billion in data center infrastructure revenue alone by 2028. Nvidia plans to sustain its leadership with advancements in both its chips and software. It introduced the new Blackwell Ultra GPU, expected to begin shipping in the year’s second half. The Blackwell chips are noted for their increased power, catering to time-sensitive services. Nvidia predicts revenue from Blackwell will surpass that of its previous Hopper architecture.
In its continued innovation, Nvidia plans to release its new Vera Rubin chip, featuring a combination of a GPU and its next-generation Rubin architecture with a custom-designed central processing unit (CPU) using Arm’s technology. The new CPU is expected to be twice as fast as the existing one in the Grace Blackwell chips. Furthermore, Nvidia aims to quadruple the number of GPU dies in its Blackwell chips by developing the “Rubin Next” chip, targeted for launch in the latter half of 2027.
Beyond hardware, Nvidia also launched Nvidia Dynamo, an open-source software system designed to enhance inference throughput and reduce costs. This system is meant to facilitate and expedite inference communication across numerous GPUs, functioning as an operating system for AI operations.
Nvidia is not limiting its scope to data centers. The company is also exploring opportunities in robotics and autonomous driving markets. Huang announced the advent of “the age of generalist robotics” with Isaac GROOT N1, described as the world’s first “open Humanoid Robot foundation model.” This model is trainable on both real and synthetic data to assist robots in task mastery, addressing a global job shortage.
Additionally, Nvidia is collaborating with General Motors (GM) to develop an autonomous driving system for the automaker, despite GM’s previous challenges in the sector. Nvidia will not only provide GPUs but will also aid GM in creating custom AI systems, utilizing Nvidia’s GPUs and software to train AI manufacturing models for next-generation factory robotics. This collaboration follows a recent agreement with Toyota, where Nvidia supplies chips and software for advanced driver-assistance features.
Considering these developments, Nvidia maintains a substantial opportunity for expansion. AI infrastructure investment is on the rise, and Nvidia continues to innovate, targeting dominance in AI inference and branching into other potential markets. Currently, Nvidia’s stock is seen as attractively valued post-market sell-off, trading at a forward price-to-earnings (P/E) ratio below 26 times this year’s analyst estimates, and a price/earnings-to-growth (PEG) ratio under 0.5, indicating an undervaluation.
Thus, Nvidia presents a compelling long-term investment proposition at its current valuation levels.
