Nvidia Economics: Make $5-$7 for Every $1 Spent on GPUs

By Agam Shah

June 30, 2024

Nvidia is saying that companies could make $5 to $7 for every $1 invested in GPUs over a four-year period. Customers are investing billions in new Nvidia hardware to keep up with newer AI models to drive revenue and productivity, said Ian Buck, vice president and general manager of Nvidia’s hyperscale and HPC business, this month at the Bank of America Securities 2024 Global Technology Conference.

Companies racing to build huge data centers will especially benefit and see huge returns over the data center’s four-to-five-year lifespan.

“Every dollar a cloud provider spends on buying a GPU, they make it back at $5 over four years,” Buck said.

Buck said inferencing is even more profitable.

“Here the economics are even better: for every dollar spent, $7 is turned over in that same time period and growing,” Buck said.

AI inference around Llama, Mistral, or Gemma is growing and is served by tokens. Nvidia is packaging open-source AI models in containers called Nvidia Inference Microservices, or NIMs.

Nvidia said its upcoming Blackwell GPU, announced earlier this year, was optimized for inferencing. The GPU supports the FP4 and FP6 data types, which adds more energy efficiency when running low-intensity AI workloads.

Preparing For Rubin

Cloud providers start planning data centers two years in advance and want to know what GPU architectures will look like in the future.

Nvidia has shared plans for Rubin — a new GPU announced at Computex — so cloud providers can prepare data centers for the GPUs. Rubin will ship in 2026 and succeed Blackwell and Blackwell Ultra, which is coming in 2025.

“It’s really important for us to do that — data centers don’t drop out of the sky, they’re big construction projects. They need to understand ‘what is a Blackwell data center going to look like and how is it going to be different than Hopper?'” Buck said.

Blackwell provides an opportunity to move to a denser form of computing and use technologies like liquid cooling, as air cooling isn’t as efficient.

Nvidia introduces a new GPU every year, which helps the company keep pace with AI growth and, in turn, help customers plan products and AI strategies.

“Rubin has been in conversation for quite some time with those biggest customers — they know where we’re going and the timescales,” Buck said.

The speed and capabilities of AI are tied directly to the hardware. The more money spent on GPUs, the more companies can train larger models, which in turn delivers more revenue.

Microsoft and Google have pinned their future on AI and are in a race to develop more capable large-language models. Microsoft relies heavily on new GPUs to prop up its GPT-4 backend, while Google relies on its TPUs to run its AI infrastructure.

Blackwell in Short Supply

Nvidia is now producing Blackwell GPUs, and samples will be released soon. But customers can expect the initial batch of GPUs — which will ship by the end of the year — to be in short supply.

“With every new technology transition comes … a mix of challenges of supply and demand. We experienced that certainly with Hopper, and there’ll be similar kinds of supply-demand constraints in the ramp of Blackwell … at the end of this year and going into next year,” Buck said.

NVIDIA founder and CEO Jensen Huang display production version of Blackwell. (Source Nvidia)

Buck said data center companies are deprecating CPU infrastructure to make space for more GPUs. Hopper GPUs are being retained, while older GPUs based on Ampere and Volta architectures are being resold.

Nvidia will maintain multiple tiers of GPUs, with Hopper becoming its mainstream AI GPU as Blackwell ramps. Nvidia has already made multiple hardware and software improvements to boost the Hopper’s performance.

All the cloud providers will offer Blackwell GPUs and servers.

Brain-Sized Modes

The GPT-4 model has about 1.8 trillion parameters, and the number of parameters will continue to grow as AI scaling hasn’t reached its limits, Buck said.

“The human brain is 100 billion to 150 trillion, depending on who you are, with the neurons and connections in your head. We’re at about 2 trillion now in AI… we haven’t gotten to reasoning yet,” Buck said.

There will be one large model with trillions of parameters around which smaller and more specialized models will be built. The number of parameters works in Nvidia’s favor, as it helps sell more GPUs.

Mixture of Experts

Nvidia is adapting its GPU architecture with a shift to a mixture-of-experts model from the original foundational model approach.

The mixture-of-experts involves multiple neural networks validating answers by referring to each other.

“The 1.8 trillion parameter GPT model has 16 different neural networks, all trying to answer their part of the layer and then confer and meet up and decide what the right answer is,” Buck said.

The upcoming GB200 NVL72 rack-scale server, which has 72 Blackwell GPUs and 36 Grace CPUs, is designed for the mixture-of-experts model. The multiple GPUs and CPUs are interconnected, which allows mixture-of-expert models.

“These guys can all be communicating with each other and not get blocked on IO. This evolution is constantly happening in model architectures,” Buck said.

Techniques to Lock Down Customers

Nvidia’s CEO, Jensen Huang, made some wild comments at HPE’s Discover this month, urging people to buy more of the company’s hardware and software.

Nvidia and HPE announced a new portfolio of products with the uncomplicated title “Nvidia AI Computing by HPE.”

“We’ve made it so that you could have small, medium, large, extra-large. And as you know, the more you buy, the more you save,” Huang said during a stage appearance at Discover.

Huang’s other questionable comments were earlier this year, when he said future programmers wouldn’t need to learn how to code. But loading AI models on Nvidia GPUs requires knowledge of the command line and scripts to create and run AI environments.

Nvidia’s proprietary rhetoric and complete dominance in the AI market have made it a target of an antitrust investigation.

Buck had to be careful when he tried to tone down the concerns over CUDA, saying that “moat is a complicated word.”

Both executives have explained that CUDA is a must-have for its GPUs- to max out a GPU, you need CUDA. Open-source software works with Nvidia GPUs but doesn’t pack the punch of CUDA’s libraries and runtime.

Backward compatibility and continuity are Nvidia’s distinctive advantages. Support for Nvidia’s AI models and software can carry forward to the next generation GPU. The same isn’t true for ASICs such as Intel’s Gaudi, which must be tuned again for every new model.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

ARM, Fujitsu Targeting Open-source Software for Power Efficiency in 2-nm Chip

July 19, 2024

Fujitsu and ARM are relying on open-source software to bring power efficiency to an air-cooled supercomputing chip that will ship in 2027. Monaka chip, which will be made using the 2-nanometer process, is based on the Read more…

SCALEing the CUDA Castle

July 18, 2024

In a previous article, HPCwire has reported on a way in which AMD can get across the CUDA moat that protects the Nvidia CUDA castle (at least for PyTorch AI projects.). Other tools have joined the CUDA castle siege. AMD Read more…

Quantum Watchers – Terrific Interview with Caltech’s John Preskill by CERN

July 17, 2024

In case you missed it, there's a fascinating interview with John Preskill, the prominent Caltech physicist and pioneering quantum computing researcher that was recently posted by CERN’s department of experimental physi Read more…

Aurora AI-Driven Atmosphere Model is 5,000x Faster Than Traditional Systems

July 16, 2024

While the onset of human-driven climate change brings with it many horrors, the increase in the frequency and strength of storms poses an enormous threat to communities across the globe. As climate change is warming ocea Read more…

Researchers Say Memory Bandwidth and NVLink Speeds in Hopper Not So Simple

July 15, 2024

Researchers measured the real-world bandwidth of Nvidia's Grace Hopper superchip, with the chip-to-chip interconnect results falling well short of theoretical claims. A paper published on July 10 by researchers in the U. Read more…

Belt-Tightening in Store for Most Federal FY25 Science Budets

July 15, 2024

If it’s summer, it’s federal budgeting time, not to mention an election year as well. There’s an excellent summary of the curent state of FY25 efforts reported in AIP’s policy FYI: Science Policy News. Belt-tight Read more…

SCALEing the CUDA Castle

July 18, 2024

In a previous article, HPCwire has reported on a way in which AMD can get across the CUDA moat that protects the Nvidia CUDA castle (at least for PyTorch AI pro Read more…

Aurora AI-Driven Atmosphere Model is 5,000x Faster Than Traditional Systems

July 16, 2024

While the onset of human-driven climate change brings with it many horrors, the increase in the frequency and strength of storms poses an enormous threat to com Read more…

Shutterstock 1886124835

Researchers Say Memory Bandwidth and NVLink Speeds in Hopper Not So Simple

July 15, 2024

Researchers measured the real-world bandwidth of Nvidia's Grace Hopper superchip, with the chip-to-chip interconnect results falling well short of theoretical c Read more…

Shutterstock 2203611339

NSF Issues Next Solicitation and More Detail on National Quantum Virtual Laboratory

July 10, 2024

After percolating for roughly a year, NSF has issued the next solicitation for the National Quantum Virtual Lab program — this one focused on design and imple Read more…

NCSA’s SEAS Team Keeps APACE of AlphaFold2

July 9, 2024

High-performance computing (HPC) can often be challenging for researchers to use because it requires expertise in working with large datasets, scaling the softw Read more…

Anders Jensen on Europe’s Plan for AI-optimized Supercomputers, Welcoming the UK, and More

July 8, 2024

The recent ISC24 conference in Hamburg showcased LUMI and other leadership-class supercomputers co-funded by the EuroHPC Joint Undertaking (JU), including three Read more…

Generative AI to Account for 1.5% of World’s Power Consumption by 2029

July 8, 2024

Generative AI will take on a larger chunk of the world's power consumption to keep up with the hefty hardware requirements to run applications. "AI chips repres Read more…

US Senators Propose $32 Billion in Annual AI Spending, but Critics Remain Unconvinced

July 5, 2024

Senate leader, Chuck Schumer, and three colleagues want the US government to spend at least $32 billion annually by 2026 for non-defense related AI systems.  T Read more…

Atos Outlines Plans to Get Acquired, and a Path Forward

May 21, 2024

Atos – via its subsidiary Eviden – is the second major supercomputer maker outside of HPE, while others have largely dropped out. The lack of integrators and Atos' financial turmoil have the HPC market worried. If Atos goes under, HPE will be the only major option for building large-scale systems. Read more…

Everyone Except Nvidia Forms Ultra Accelerator Link (UALink) Consortium

May 30, 2024

Consider the GPU. An island of SIMD greatness that makes light work of matrix math. Originally designed to rapidly paint dots on a computer monitor, it was then Read more…

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock_1687123447

Nvidia Economics: Make $5-$7 for Every $1 Spent on GPUs

June 30, 2024

Nvidia is saying that companies could make $5 to $7 for every $1 invested in GPUs over a four-year period. Customers are investing billions in new Nvidia hardwa Read more…

Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study

June 10, 2024

Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst fir Read more…

AMD Clears Up Messy GPU Roadmap, Upgrades Chips Annually

June 3, 2024

In the world of AI, there's a desperate search for an alternative to Nvidia's GPUs, and AMD is stepping up to the plate. AMD detailed its updated GPU roadmap, w Read more…

Some Reasons Why Aurora Didn’t Take First Place in the Top500 List

May 15, 2024

The makers of the Aurora supercomputer, which is housed at the Argonne National Laboratory, gave some reasons why the system didn't make the top spot on the Top Read more…

Intel’s Next-gen Falcon Shores Coming Out in Late 2025 

April 30, 2024

It's a long wait for customers hanging on for Intel's next-generation GPU, Falcon Shores, which will be released in late 2025.  "Then we have a rich, a very Read more…

Leading Solution Providers

Contributors

Google Announces Sixth-generation AI Chip, a TPU Called Trillium

May 17, 2024

On Tuesday May 14th, Google announced its sixth-generation TPU (tensor processing unit) called Trillium.  The chip, essentially a TPU v6, is the company's l Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

IonQ Plots Path to Commercial (Quantum) Advantage

July 2, 2024

IonQ, the trapped ion quantum computing specialist, delivered a progress report last week firming up 2024/25 product goals and reviewing its technology roadmap. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

The NASA Black Hole Plunge

May 7, 2024

We have all thought about it. No one has done it, but now, thanks to HPC, we see what it looks like. Hold on to your feet because NASA has released videos of wh Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire