ComputePublished Top storyAll news from that day
Ongoing case: What the AI buildout costs and needs 6 stories
Huang says Nvidia is probably running about 40% of cloud providers' capex
On the Bg2 Pod, published 15 October 2026, Nvidia chief executive Jensen Huang gave his own figures for per-gigawatt supply, capex share and startups' open-model use.
Check our sources · 9 facts from 2 sources
Video: Bg2 Pod on YouTube
Key points
- Huang says what Nvidia offers into one gigawatt rose "to 20 somewhat billion per gigawatt to now 40 plus per gigawatt" (54:11).
- Huang says Nvidia's share of the world's capex was "in single digits" about six years earlier (54:39) and that with cloud providers "we're probably running about 40%" (54:46).
- Huang says of the startups that rent Nvidia compute that "80% of them use open models" (1:06:15).
What happened
Jensen Huang, Nvidia's founder and chief executive, was interviewed on the Bg2 Pod. The video was published on the programme's YouTube channel on 15 October 2026.
Asked about Nvidia's growth, Huang said that what Nvidia offers into one gigawatt of data-centre capacity has risen "to 20 somewhat billion per gigawatt to now 40 plus per gigawatt" (54:11). The figure is his own, given in conversation.
On Nvidia's share of spending he said "Our percentage of the world's capex was in single digits" about six years earlier (54:39), and that with the world's cloud providers "we're probably running about 40%" (54:46).
Answering a question on whether the labs can keep growing their revenue, Huang began "People misunderstand what's going on" (1:01:58) and said "Nvidia represents call it 80% of the revenue share of AI infrastructure in the world today" (1:02:03).
On open models he said of the startups that rent Nvidia compute that "80% of them use open models" (1:06:15), and that "many of them taking Neotron" (1:06:35), which is the automatic captions' rendering of Nemotron, Nvidia's model family.
What it means for you
Our viewThese are Huang's own figures, given in conversation, and for the per-gigawatt number the transcript carries no unit beyond "billion" for the earlier values. A supplier's estimate of its share of customers' spending shows how it frames its position, but it is not an audited number.
His remark that startups renting Nvidia compute mostly use open models is likewise his description, not a survey. If you plan capacity or negotiate GPU pricing, ask your cloud provider how much of your quoted price reflects accelerator costs, and compare that with what Huang says.
It adds no new facts.
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against these 2 sources.
1 Jensen Huang, Founder & CEO
Open the source archived copy-
Jensen Huang's title on NVIDIA's own newsroom biography is Founder & CEO.
cite
2 Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more
Open the source archived copy-
The Bg2 Pod video with Jensen Huang, titled "Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more", was published on the Bg2 Pod YouTube channel on 15 October 2026.
citeYouTube page: title Jensen Huang | Does AI Revenue Justify Capex, ’27 AI Bottlenecks, Singularity, Open vs Closed & more; channel Bg2 Pod; uploadDate 2026-10-15T03:00:25-07:00
-
Jensen Huang said, in the Bg2 Pod interview, that what Nvidia offers into one gigawatt "went from" about 18 billion or so per gigawatt "to 20 somewhat billion per gigawatt to now 40 plus per gigawatt" (54:11). Spoken words, no unit stated beyond "billion" for the earlier figures.
citethe the offering that we we offer into one gigawatt went from went from uh uh I guess about you know 18 billion or so per gigawatt to 20 somewhat billion per gigawatt to now 40 plus per gigawatt (54:03-54:16)
-
Jensen Huang said "Our percentage of the world's capex was in single digits" about six years earlier (54:39).
citeOur percentage of the world’s capex was in single digits. (54:39), after: just go back in just uh six years ago (54:34)
-
Jensen Huang said that with the world's cloud providers "we're probably running about 40%", speaking of Nvidia's share of their capex (54:46).
citeright now with the world’s CSPs we’re probably running about 40% (54:43-54:48)
-
Asked whether the labs can keep growing revenue two to three times a year, Jensen Huang began his answer "People misunderstand what's going on" (1:01:58).
citeQ (1:01:35): the labs can continue to two to 3x their revenues every year ... A (1:01:58): Oh, yeah. People misunderstand what’s going on.
-
Jensen Huang said "Nvidia represents call it 80% of the revenue share of AI infrastructure in the world today" (1:02:03).
citeNvidia represents call it 80% of the revenue share of AI infrastructure in the world today (1:02:03-1:02:08)
-
Jensen Huang said of the startups that rent Nvidia compute that "80% of them use open models" (1:06:15).
citeif you look at all these startup companies, 80% of them use open models (1:06:14-1:06:17); the startups are those he described at 1:05:25-1:05:29: thousands of them. And what do they do? They rent NVIDIA compute.
-
Jensen Huang said "many of them taking Neotron" (1:06:35); the automatic captions render Nemotron, Nvidia's model family, as "Neotron".
citethey’re taking um these open models and and many of them taking Neotron (1:06:31-1:06:35)
Every sentence above is checked against these 2 sources. Each fact has its own address you can cite.
Source 1 · NVIDIA Newsroom
Jensen Huang, Founder & CEO
Source 2 · Bg2 Pod · 15 Oct 2026
Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more
The Bg2 Pod video with Jensen Huang, titled "Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more", was published on the Bg2 Pod YouTube channel on 15 October 2026.
YouTube page: title Jensen Huang | Does AI Revenue Justify Capex, ’27 AI Bottlenecks, Singularity, Open vs Closed & more; channel Bg2 Pod; uploadDate 2026-10-15T03:00:25-07:00
Jensen Huang said, in the Bg2 Pod interview, that what Nvidia offers into one gigawatt "went from" about 18 billion or so per gigawatt "to 20 somewhat billion per gigawatt to now 40 plus per gigawatt" (54:11). Spoken words, no unit stated beyond "billion" for the earlier figures.
the the offering that we we offer into one gigawatt went from went from uh uh I guess about you know 18 billion or so per gigawatt to 20 somewhat billion per gigawatt to now 40 plus per gigawatt (54:03-54:16)
Jensen Huang said "Our percentage of the world's capex was in single digits" about six years earlier (54:39).
Our percentage of the world’s capex was in single digits. (54:39), after: just go back in just uh six years ago (54:34)
Show all 8 factsShow fewer facts
Jensen Huang said that with the world's cloud providers "we're probably running about 40%", speaking of Nvidia's share of their capex (54:46).
right now with the world’s CSPs we’re probably running about 40% (54:43-54:48)
Asked whether the labs can keep growing revenue two to three times a year, Jensen Huang began his answer "People misunderstand what's going on" (1:01:58).
Q (1:01:35): the labs can continue to two to 3x their revenues every year ... A (1:01:58): Oh, yeah. People misunderstand what’s going on.
Jensen Huang said "Nvidia represents call it 80% of the revenue share of AI infrastructure in the world today" (1:02:03).
Nvidia represents call it 80% of the revenue share of AI infrastructure in the world today (1:02:03-1:02:08)
Jensen Huang said of the startups that rent Nvidia compute that "80% of them use open models" (1:06:15).
if you look at all these startup companies, 80% of them use open models (1:06:14-1:06:17); the startups are those he described at 1:05:25-1:05:29: thousands of them. And what do they do? They rent NVIDIA compute.
Jensen Huang said "many of them taking Neotron" (1:06:35); the automatic captions render Nemotron, Nvidia's model family, as "Neotron".
they’re taking um these open models and and many of them taking Neotron (1:06:31-1:06:35)
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep