NVIDIA is redefining the future of artificial intelligence (AI) and computing through its rapid advancements in data centers and the revolutionary Blackwell architecture. With AI workloads becoming more demanding, NVIDIA’s cutting-edge innovations are accelerating the transformation of data centers into AI factories, powering the next era of intelligence and automation.
Unprecedented Growth in NVIDIA’s Data Center Revenue
In fiscal year 2025, NVIDIA’s data center revenue soared to $115.2 billion—more than double the previous year. The fourth quarter alone set a record $35.6 billion, marking a 16% sequential increase and an astonishing 93% year-over-year growth. This meteoric rise is driven by the ramp-up of Blackwell and continued demand for Hopper 200, underscoring the industry’s shift toward accelerated computing.
Key Drivers of Revenue Expansion
- Cloud Service Providers (CSPs): Large CSPs accounted for nearly half of NVIDIA’s data center revenue, with sales almost doubling year-over-year.
- Consumer Internet Growth: A threefold increase in revenue was fueled by generative AI and deep learning applications.
- Enterprise AI: The enterprise sector saw its revenue nearly double due to growing demand for model fine-tuning and AI-powered workflows.
Geographic and Networking Expansion
The U.S. led the way in sequential growth for data center revenue, while China’s sales remained below pre-export control levels. Alongside revenue growth, NVIDIA is making a significant transition in networking technology. Moving from NVLink 8 with InfiniBand to NVLink 72 with Spectrum X, the company is enhancing AI infrastructure. Microsoft Azure and other major players are leveraging Spectrum X to construct AI factories, pushing AI scalability to new heights.
Blackwell: The Fastest Ramp in NVIDIA’s History
Blackwell is NVIDIA’s most advanced AI computing architecture, setting records with $11 billion in revenue in its first quarter of sales. It is designed for reasoning AI models, delivering up to 25x higher token throughput and 20x lower cost compared to Hopper. Remarkably, many early deployments of the GB200 are for inference tasks—a first for a new NVIDIA architecture.
Blackwell’s Cutting-Edge Innovations
- FP4 Transformer Engine: Enables 25x faster reasoning AI model execution than Hopper.
- NVLink 72 Scale-Up Fabric: Enhances system connectivity and efficiency.
- New Software Technologies: Optimized for the next generation of AI workloads.
- Engineering Feats: Each Grace Blackwell NVLink 72 rack incorporates millions of components sourced from hundreds of manufacturing sites.
AI Scaling Laws and the Future of AI Computing
The demand for AI computing is being driven by three key scaling laws:
- Traditional Scaling: Increasing computational power continues to drive AI improvements.
- Post-Training Scaling: Requires significantly more compute resources than pretraining.
- Inference Time Scaling: The demand for AI models to perform complex real-time reasoning is skyrocketing.
As AI transitions into reasoning-based models, enterprises and sovereign entities are investing heavily in AI-driven computing. This expansion includes enterprise AI, robotics (physical AI), and sovereign AI initiatives worldwide.
Data Centers Evolving into AI Factories
Data centers are no longer just storage and processing hubs; they are transforming into AI factories. The majority of future data center capital expenditures will focus on accelerated computing and AI, enabling organizations to either rent or operate their own AI factories to stay competitive in the AI-driven economy.
What’s Next? Blackwell Ultra and Beyond
NVIDIA is preparing to launch Blackwell Ultra in the second half of the year. Featuring enhanced networking, memory, and processors, this new iteration retains Blackwell’s system architecture, simplifying the transition compared to the Hopper-to-Blackwell upgrade.
Financial Outlook: Strong Gross Margins
As Blackwell ramps up, NVIDIA expects gross margins to be in the low 70s, potentially improving to the mid-70s later in the fiscal year, reflecting strong efficiency gains and demand.
Future Looks Bright
With Blackwell at the forefront, NVIDIA is not just driving AI innovation, it is reshaping the entire data center landscape. As AI computing demand accelerates, NVIDIA’s advancements in accelerated computing, networking, and software will continue to fuel the next wave of AI breakthroughs. Businesses and institutions must prepare for an AI-centric future, where data centers evolve into the engines of intelligence, automation, and discovery.