NVIDIA HGX B300 on CoreWeave offers the performance at scale you need to solve the challenges of building and deploying agentic AI. CoreWeave Staff Operations Engineer Zach Merendino discusses the rollout from the floor of our purpose-built AI data center, and how CoreWeave Cloud delivers proven performance trusted by AI pioneers, at scale.
1
00:00:07,273 --> 00:00:09,042
Hey there. I'm Zach Merendino
2
00:00:09,042 --> 00:00:11,077
staff operations engineer at CoreWeave.
3
00:00:11,077 --> 00:00:13,179
And today I'm in one of Switch’s data centers
4
00:00:13,179 --> 00:00:16,116
where we're rolling out NVIDIA HGX B300
5
00:00:16,116 --> 00:00:17,717
for general availability.
6
00:00:17,717 --> 00:00:18,618
CoreWeave unlocks
7
00:00:18,618 --> 00:00:20,387
NVIDIA HGX B300 performance
8
00:00:20,387 --> 00:00:21,988
at scale for Agentic AI
9
00:00:21,988 --> 00:00:24,791
by delivering partnership, pace and performance.
10
00:00:24,791 --> 00:00:27,394
By combining the NVIDIA Blackwell Ultra innovations
11
00:00:27,394 --> 00:00:29,162
with our purpose built AI cloud,
12
00:00:29,162 --> 00:00:31,398
CoreWeave delivers what your team needs
13
00:00:31,398 --> 00:00:32,465
to stay one step ahead.
14
00:00:32,465 --> 00:00:34,367
We provide AI-native infrastructure
15
00:00:34,367 --> 00:00:36,569
co-designed and optimized with NVIDIA
16
00:00:36,569 --> 00:00:39,406
to support Agentic AI at production scale
17
00:00:39,406 --> 00:00:41,541
backed by direct-expert support
18
00:00:41,541 --> 00:00:43,743
that's trusted by our leading AI pioneers
19
00:00:43,743 --> 00:00:46,246
like Mistral, Cohere and Meta.
20
00:00:46,246 --> 00:00:47,881
CoreWeave's observability platform
21
00:00:47,881 --> 00:00:49,516
and operating standard Mission Control
22
00:00:49,516 --> 00:00:51,217
reduces iteration cycles,
23
00:00:51,217 --> 00:00:53,086
surfaces issues faster
24
00:00:53,086 --> 00:00:55,288
and allows you to move from run to results
25
00:00:55,288 --> 00:00:56,923
with predictable speed at scale
26
00:00:56,923 --> 00:00:58,691
and sustain performance that pairs
27
00:00:58,691 --> 00:01:01,461
the NVIDIA HGX B300’s expanded memory
28
00:01:01,461 --> 00:01:02,562
with non-blocking
29
00:01:02,562 --> 00:01:04,964
NVIDIA Quantum X 800 InfiniBand networking
30
00:01:04,964 --> 00:01:07,367
to deliver twice the node to node bandwidth
31
00:01:07,367 --> 00:01:09,502
across multi thousand GPU node clusters.
32
00:01:09,502 --> 00:01:11,871
With NVIDIA HGX B300 on CoreWeave,
33
00:01:11,871 --> 00:01:14,274
you can run Agentic AI at production scale
34
00:01:14,274 --> 00:01:17,110
on a proven platform trusted by AI pioneers.
35
00:01:17,110 --> 00:01:18,244
CoreWeave cloud unlocks
36
00:01:18,244 --> 00:01:20,747
NVIDIA HGX B300 performance at scale
37
00:01:20,747 --> 00:01:22,048
so you can be ready for anything.
