AI Boom Drives Nvidia's Revenue Growth with Core Business Revenue Surging 409%
-
Nvidia (NVDA.US), the chip giant hailed as the 'strongest shovel seller' in the AI field, has once again reported exceptionally strong quarterly results and outlook that far exceeded market expectations. With the groundbreaking emergence of generative AI like ChatGPT, the world is gradually stepping into a new AI era. Since then, not only the tech industry but various sectors globally have seen surging demand for Nvidia's AI chips—specifically the A100/H100 chips used in AI training and inference—leading the tech giant to announce yet another set of astonishingly strong financial results following the previous three quarters.
In the fourth quarter of fiscal year 2024, ending January 28, Nvidia's total revenue more than tripled to $22.1 billion. Excluding certain items, non-GAAP earnings per share were $5.16, both significantly surpassing Wall Street analysts' consensus estimates of $20.4 billion in revenue and $4.60 per share. More importantly, Nvidia expects another substantial revenue increase in the current quarter, helping justify its soaring stock price and maintaining its position as one of the world's most valuable companies.
The total revenue highlights Nvidia's consecutive growth scale: back in fiscal year 2021, the company's entire annual revenue didn't even reach this figure. Additionally, Nvidia's core business segment—the data center division providing A100/H100 chips globally—achieved Q4 revenue of approximately $18.4 billion, a staggering 409% year-over-year increase.
After a 240% stock surge in 2023, Nvidia's shares have risen another 40% so far in 2024. The company's market capitalization has grown by over $400 billion this year, reaching a total of $1.67 trillion, as investors bet it will remain the primary beneficiary of the AI computing boom. NVIDIA CEO Jensen Huang stated: "GPU-accelerated computing and generative AI have reached a 'tipping point.' Demand from companies, industries, and even nations worldwide is surging."
During an earnings call with Wall Street analysts, Huang noted that NVIDIA's latest products would remain in short supply for the rest of the year. He emphasized that while supply is increasing, demand shows no signs of slowing down. "Generative AI has kicked off a whole new investment cycle," Huang said. He projected that "data center infrastructure will double in scale over the next five years, representing a market opportunity worth hundreds of billions of dollars annually."
Regarding the Chinese market, Huang mentioned that the company has begun shipping samples of new chips compliant with export restrictions to Chinese clients. This move is expected to help revive NVIDIA's business in China. "We will do our best to compete and succeed in this market," Huang stressed.
Following NVIDIA's explosive earnings report, its stock surged over 11% in after-hours trading, lifting the broader semiconductor and tech sectors. This rebound came after a week of weakness in AI-related tech and chip stocks due to cautious investor sentiment ahead of NVIDIA's earnings release. The company's performance has reignited global tech investors' "AI faith," potentially sparking significant waves in global stock markets.
Chris Caso, an analyst at Wolfe Research, noted in a report: "Global equity markets were focused on this report, so expectations were high. However, NVIDIA's outlook was strong enough to justify the stock's rise and leave room for further gains in the second half of the year." However, there is no doubt that competition in the AI chip field will become increasingly fierce. NVIDIA's strongest competitor, AMD (AMD.US), has recently begun selling its MI300 series AI GPU accelerators. AMD expects to generate $3.5 billion in revenue from this series this year, up from the previously forecasted $2 billion. AI chip startups will also become formidable challengers to NVIDIA. For instance, Groq recently launched its self-developed LPU, which boasts text generation speeds faster than a blink and inference performance 10 times faster than NVIDIA's GPUs. But NVIDIA is not standing still. Analysts predict the company will soon mass-produce the more powerful H200 AI chip and the highly anticipated B100.
NVIDIA - The Undisputed 'Shovel Seller' in the AI Field
The latest performance proves that NVIDIA remains the undisputed 'strongest shovel seller' in the global AI field. With a 90% share in the AI training market, it is capitalizing on the unprecedented global AI adoption wave. For example, while Groq's LPU is currently more suitable for inference, training large language models still requires purchasing NVIDIA GPUs in bulk.
NVIDIA recognized the potential of GPUs in AI and deep learning early on, investing heavily in related R&D and successfully building a robust ecosystem around its GPU hardware.
NVIDIA has been a leader in global high-performance computing for years, particularly with its CUDA computing platform, which is the preferred choice for AI training/inference and other high-performance computing applications. NVIDIA's current hottest AI chip, the H100 GPU accelerator, is based on its groundbreaking Hopper GPU architecture, delivering unparalleled computing power, especially in floating-point operations, tensor core performance, and AI-specific acceleration. ChatGPT developer OpenAI, US tech giants Amazon.com Inc., Meta Platforms (parent of Facebook and Instagram), Tesla, Microsoft, and Alphabet (Google's parent) are NVIDIA's largest customers, accounting for nearly 50% of its total revenue. These companies are now investing heavily in AI computing hardware, such as NVIDIA's AI chips.
Tesla CEO Elon Musk compares the tech industry's AI arms race to a high-stakes "poker game," where companies must invest billions annually in AI hardware to remain competitive. The billionaire revealed that Tesla alone will spend over $500 million on NVIDIA's AI chips in 2024, but warned that "billions more" in hardware investments will be needed to compete with larger rivals.
As the world enters the AI era, NVIDIA's data center business has become its core operation, surpassing its historically dominant gaming graphics card segment. NVIDIA's Data Center division—which supplies A100/H100 chips globally—has transformed from a "side business" (gaming was NVIDIA's primary focus since founding) into the company's most powerful revenue driver.
NVIDIA's Data Center division has consistently outperformed other business segments for multiple quarters, with Q4 revenue surging 409% YoY to $18.4 billion. The company predicts global data center infrastructure will double in scale within five years. Meanwhile, NVIDIA's gaming division saw revenue grow 56% YoY to $2.9 billion, benefiting from the global chip demand recovery. NVIDIA is currently focused on expanding its artificial intelligence hardware and software ecosystem beyond large data centers. The 61-year-old Jensen Huang has recently traveled the world, advocating that governments need sovereign-level AI systems to both protect data and gain competitive advantages in AI.
Huang first introduced the concept of 'sovereign AI capabilities,' indicating a surge in national-level demand for AI hardware. He stated that countries worldwide now intend to build and operate their own AI infrastructure domestically, which will significantly increase demand for NVIDIA's hardware products. In recent interviews, Huang mentioned that nations including India, Japan, France, and Canada are discussing the importance of investing in 'sovereign AI capabilities.'
In terms of financial outlook, NVIDIA, the world's most valuable chip company, projected total Q4 2024 revenue (ending April 2024) to reach approximately $24 billion. This figure substantially exceeds Wall Street analysts' average forecast of $21.9 billion. This exceptionally strong performance outlook highlights NVIDIA's position as a prime beneficiary of the global AI boom, effectively serving as the leading provider of core AI infrastructure.
Facing surging consumer demand for generative AI products like ChatGPT and Google Bard, along with other enterprise AI software tools, tech giants and data center operators worldwide are scrambling to stock up on NVIDIA's H100 GPU accelerators. The H100 excels at handling the intensive workloads required for AI training and inference.
GPUs: One of the Core Infrastructures of the AI Era With the world stepping into the AI era and the acceleration of ubiquitous connectivity, global computing demands are witnessing explosive growth. This is particularly true in AI training domains, where various specialized tasks involve massive matrix operations, forward and backward propagation in neural networks, and other computation-intensive operations that place extremely high demands on hardware performance. However, these challenges are far beyond what CPUs, which have long benefited from Moore's Law, can handle. Even large numbers of CPUs cannot solve this issue, as CPUs were originally designed for general-purpose computing across various conventional tasks, not for processing massive-scale parallel computing patterns and high-density matrix operations.
More importantly, as innovation and development in the global chip industry enter the "Post-Moore Era," CPUs, once the driving force behind human societal progress, can no longer achieve rapid breakthroughs like the "nm-level" advancements seen from 22nm to 10nm within less than five years. Subsequent nm-level breakthroughs face numerous obstacles, such as quantum tunneling and massive investment requirements, which severely limit further CPU performance upgrades and optimizations.
Consequently, GPUs, with their numerous computing cores, ability to execute multiple high-intensity AI tasks simultaneously, and exceptional proficiency in parallel computing, have emerged as the most critical hardware in the chip industry in recent years. GPUs hold unparalleled advantages in high-performance computing fields like AI training and inference, which are crucial for extremely complex AI tasks such as image recognition, natural language processing, and large-scale matrix operations. Modern GPU architectures are further optimized for AI-specific tasks like deep learning. For example, NVIDIA's Tensor Cores can accelerate highly critical, intensive operations such as matrix multiplication and convolution calculations, enabling parallel processing of large-scale floating-point and integer matrix computations to enhance computational efficiency.
Since the advent of ChatGPT, as AI's impact on the global high-tech industry and technological development grows increasingly significant, CPUs—which focus on single-threaded performance and general-purpose computing—remain an indispensable part of the chip industry. However, their status and importance in the field have been far surpassed by GPUs. From a theoretical perspective, the exponential performance growth predicted by Moore's Law has not disappeared in recent years but has shifted from CPUs to GPUs based on numerous cores. GPU performance continues to follow the exponential growth trend, doubling approximately every 2.2 years. In contrast, Intel CPU GFLOPs still show growth but appear as a flat line compared to GPU GFLOPs.
Jensen Huang emphasized that the global shift toward AI is just beginning. He believes that accelerated computing, which breaks down specific tasks into smaller parts for parallel processing, is becoming dominant. According to Mordor Intelligence, the GPU market (covering PCs, servers, high-performance computing, autonomous driving, etc.) is expected to reach approximately $206.95 billion in the next five years, with a compound annual growth rate (CAGR) of 32.70% during the forecast period (2024-2029).
Mordor Intelligence noted that GPU hardware is not only used for rendering images, animations, and video games but also for general computing purposes, deployed in nearly all computational devices worldwide. The active deployment trends in personal computers, laptops, and emerging applications (e.g., AR/VR, high-performance computing, AI, machine learning, blockchain, cryptocurrency mining, autonomous driving, and high-precision navigation for vehicles and robots)—especially in the AI field—will significantly drive GPU demand in the future.
Ahead of the earnings announcement, Wall Street's top investors are bullish on Nvidia.
Before Nvidia's latest earnings and outlook were released, the most elite investment institutions on Wall Street were overwhelmingly optimistic about Nvidia's stock performance over the next 12 months. Among them, Wall Street giants Goldman Sachs raised their 12-month target price for Nvidia from $625 to $800, Bank of America increased it from $700 to $800, UBS adjusted their target from $580 to $850, and Mizuho Securities lifted their target from $625 to $825.
Some institutions even set Nvidia's 12-month target stock price above $1,000 before the earnings report was released. Rosenblatt reiterated a 'buy' rating for Nvidia with a target price of $1,100, while Loop Capital set the highest Wall Street target at $1,200.
Loop Capital analysts explained that such a high target price reflects extreme optimism about the demand prospects for Nvidia's AI chips and the ongoing shift of hyperscale computing centers and data centers toward GPU systems. They believe Nvidia is at the 'front end' of a multi-year cycle, comparing this phenomenon to the 'internet infrastructure boom' of the mid-to-late 1990s—when internet adoption rates expanded fivefold from 1995 to 2001.
Undoubtedly, competition in the AI chip sector will become increasingly fierce.
With Nvidia's strongest competitor AMD recently launching sales of its MI300 series AI GPU accelerators, the emergence of Groq's self-developed LPU AI chip, and cloud computing giants like Google continuing to focus on ASIC chips for AI acceleration, the competition in the AI chip field is set to intensify further. NVIDIA is not standing still, with analysts anticipating the company's imminent mass production of the more powerful AI chip, the B100. Wall Street giant Morgan Stanley believes that NVIDIA's future product revenue is also promising, expecting the B100 to be a game-changer in artificial intelligence, with an impact even more significant than the previous flagship AI chip, the H100.
Originally scheduled for release in late 2024, the NVIDIA B100 has been moved up to the first half of 2024 due to overwhelming demand for AI chips. According to insiders, it has already entered the supply chain certification phase.
The B100 will effortlessly handle large language models with 173 billion parameters, more than twice as powerful as the current model, the H200. Additionally, the B100 will adopt more advanced HBM high-bandwidth memory specifications, expected to break new ground in stacking capacity and bandwidth, surpassing the existing 4.8TB/s.
The B100 AI GPU will incorporate cutting-edge technology, including an estimated 178 billion transistors and the latest HBM3e memory technology, a significant upgrade over the H100's HBM3. This means substantial improvements in both data processing and storage capabilities. The launch of the B100 also signifies NVIDIA's push to dominate the AI inference field, where it hasn't yet established full superiority. Specifically, the B100 will be over four times faster than the current H100 AI GPU in AI inference performance, highlighting its significant advantage in inference tasks for trained models.
Furthermore, the B100 is expected to feature a chiplet design, marking NVIDIA's first use of this technology in its GPU products. According to NVIDIA's product roadmap, the X100 is slated for release in 2025, further expanding the GPU product lineup and solidifying NVIDIA's leadership in AI chips. A month later, on March 18 local time, NVIDIA will host the 2024 GTC event for AI developers. During the event, NVIDIA CEO Jensen Huang will deliver a keynote speech that may reveal more details about the upcoming B100. Vivek Arya, a Bank of America analyst who has consistently been bullish on NVIDIA, predicts that the upcoming B100 GPU will be priced at least 10% to 30% higher than the H100.