Thank
you, Stewart. Q3 was another record quarter. We continue to deliver
incredible growth. Revenue of $35.1 billion was up 17% sequentially and
up 94% year on year and well above our outlook of $32.5 billion.
All
market platforms posted strong sequential and year-over-year growth,
fueled by the adoption of NVIDIA accelerated computing and AI. Starting
with data center. Another record was achieved in data center. Revenue of
$30.8 billion, up 17% sequential and up 112% year on year.
NVIDIA
Hopper demand is exceptional, and sequentially, NVIDIA H200 sales
increased significantly to double-digit billions, the fastest prod ramp
in our company's history. The H200 delivers up to 2x faster inference
performance and up to 50% improved TCO. Cloud service providers were
approximately half of our data center sales with revenue increasing more
than 2x year on year. CSPs deployed NVIDIA H200 infrastructure and
high-speed networking with installations scaling to tens of thousands of
DPUs to grow their business and serve rapidly rising demand for AI
training and inference workloads.
NVIDIA H200-powered cloud
instances are now available from AWS, CoreWeave, and Microsoft Azure,
with Google Cloud and OCI coming soon. Alongside significant growth from
our large CSPs, NVIDIA GPU regional cloud revenue jumped year on year
as North America, India, and Asia Pacific regions ramped NVIDIA Cloud
instances and sovereign cloud build-outs. Consumer Internet revenue more
than doubled year on year as companies scaled their NVIDIA Hopper
infrastructure to support next-generation AI models training,
multimodal, and agentic AI, deep learning recommender engines, and
generative AI inference and content creation workloads. NVIDIA Ampere
and Hopper infrastructures are fueling inference revenue growth for
customers.
NVIDIA is the largest inference platform in the world.
Our large installed base and rich software ecosystem encourage
developers to optimize for NVIDIA and deliver continued performance and
TCO improvements. Rapid advancements in NVIDIA software algorithms
boosted Hopper inference throughput by an incredible 5x in one year and
cut time to first token by 5x. Our upcoming release of NVIDIA NIM will
boost Hopper inference performance by an additional 2.4x.....
....GAAP
gross margin was 74.6% and non-GAAP gross margin was 75%, down
sequentially, primarily driven by a mix shift of the H100 systems to
more complex and higher-cost systems within data center. Sequentially,
GAAP operating expenses and non-GAAP operating expenses were up 9% due
to higher compute, infrastructure, and engineering development costs for
new product introductions. In Q3, we returned $11.2 billion to
shareholders in the form of share repurchases and cash dividends. Well,
let me turn to the outlook for the fourth quarter.
Total
revenue is expected to be $37.5 billion, plus or minus 2%, which
incorporates continued demand for Hopper architecture and the initial
ramp of our Blackwell products. While demand greatly exceed supply, we
are on track to exceed our previous Blackwell revenue estimate of
several billion dollars as our visibility into supply continues to
increase. On Gaming, although sell-through was strong in Q3, we expect
fourth-quarter revenue to decline sequentially due to supply
constraints. GAAP and non-GAAP gross margins are expected to be 73% and
73.5%, respectively, plus or minus 50 basis points....
****
....Questions & Answers:
Operator
[Operator instructions] We'll pause for just a moment to compile the Q&A roster. As a reminder, please limit yourself to one question. Your first question comes from the line of C.J. Muse of Cantor Fitzgerald.
Your line is open.
C.J. Muse -- Analyst
Yeah.
Good afternoon. Thank you for taking the question. I guess just a
question for you on the debate around whether scaling for large language
models have stalled.
Obviously, we're very early here, but would
love to hear your thoughts on this front. How are you helping your
customers as they work through these issues? And then obviously, part of
the context here is we're discussing clusters that have yet to benefit
from Blackwell. So, is this driving even greater demand for Blackwell?
Thank you.
Jensen Huang -- President and Chief Executive Officer
Our
foundation model pretraining scaling is intact, and it's continuing. As
you know, this is an empirical law, not a fundamental physical law. But
the evidence is that it continues to scale. What we're learning,
however, is that it's not enough, that we've now discovered two other
ways to scale.
One is post-training scaling. Of course, the first
generation of post-training was reinforcement learning human feedback,
but now we have reinforcement learning AI feedback, and all forms of
synthetic data generated data that assists in post-training scaling. And
one of the biggest events and one of the most exciting developments is
Strawberry, ChatGPT o1, OpenAI's o1, which does inference time scaling,
what is called test time scaling. The longer it thinks, the better and
higher-quality answer it produces.
And it considers approaches
like chain of thought and multi-path planning and all kinds of
techniques necessary to reflect and so on and so forth. And it's --
intuitively, it's a little bit like us doing thinking in our head before
we answer your question. And so, we now have three ways of scaling, and
we're seeing all three ways of scaling. And as a result of that, the
demand for our infrastructure is really great.
You see now that
at the tail end of the last generation of foundation models were at
about 100,000 Hoppers. The next generation starts at 100,000 Blackwells.
And so, that kind of gives you a sense of where the industry is moving
with respect to pretraining scaling, post-training scaling, and then now
very importantly, inference time scaling. And so, the demand is really
great for all of those reasons.
But remember, simultaneously,
we're seeing inference really starting to scale up for our company. We
are the largest inference platform in the world today because our
installed base is so large. And everything that was trained on Amperes
and Hoppers inference incredibly on Amperes and Hoppers. And as we move
to Blackwells for training foundation models, it leads behind it a large
installed base of extraordinary infrastructure for inference.
And
so, we're seeing inference demand go up. We're seeing inference time
scaling go up. We see the number of AI native companies continue to
grow. And of course, we're starting to see enterprise adoption of
agentic AI really is the latest rage.
And so, we're seeing a lot of demand coming from a lot of different places.
Operator
Your next question comes from the line of Toshiya Hari of Goldman Sachs. Your line is open.
Toshiya Hari -- Analyst
Hi. Good afternoon. Thank you so much for taking the question. Jensen, you executed the mass change earlier this year.
There
were some reports over the weekend about some heating issues. On the
back of this, we've had investors ask about your ability to execute to
the road map you presented at GTC this year with Ultra coming out next
year and the transition to Rubin in '26. Can you sort of speak to that?
And some investors are questioning that, so if you can sort of speak to
your ability to execute on time, that would be super helpful. And then a
quick part B....
