Learning Library

← Back to Library

Microsoft Reaffirms $80 B AI Spending

Key Points

  • Microsoft reaffirmed its commitment to an $80 billion AI spend for the year, but said it may “readjust” allocations as its largest Azure AI tenant, OpenAI, contemplates moving to a SoftBank‑Oracle stack and certain data‑center projects (e.g., Kenosha, WI and Atlanta, GA) could be delayed.
  • Nvidia announced that its latest Blackwell chip architecture is already fully booked through 2025 and quickly filling orders for 2026, signaling that demand for AI compute hardware remains robust despite rumors of a slowdown.
  • The speaker emphasized that scaling AI inevitably requires significant compute investment—there is no “magic key” that bypasses hardware spending, even with China’s announced multibillion‑dollar AI infrastructure plans.
  • Satya Nadella’s call for AI to drive major GDP growth was critiqued as overlooking the immediate productivity gains already seen when workers can use AI tools, suggesting the focus should be on augmenting rather than replacing labor.
  • Current consensus is that AI’s most tangible impact lies in enhancing team and individual effectiveness now, and the notion that AI can fully replace jobs at scale remains premature.

Full Transcript

# Microsoft Reaffirms $80 B AI Spending **Source:** [https://www.youtube.com/watch?v=-T-WtXgVlXQ](https://www.youtube.com/watch?v=-T-WtXgVlXQ) **Duration:** 00:05:31 ## Summary - Microsoft reaffirmed its commitment to an $80 billion AI spend for the year, but said it may “readjust” allocations as its largest Azure AI tenant, OpenAI, contemplates moving to a SoftBank‑Oracle stack and certain data‑center projects (e.g., Kenosha, WI and Atlanta, GA) could be delayed. - Nvidia announced that its latest Blackwell chip architecture is already fully booked through 2025 and quickly filling orders for 2026, signaling that demand for AI compute hardware remains robust despite rumors of a slowdown. - The speaker emphasized that scaling AI inevitably requires significant compute investment—there is no “magic key” that bypasses hardware spending, even with China’s announced multibillion‑dollar AI infrastructure plans. - Satya Nadella’s call for AI to drive major GDP growth was critiqued as overlooking the immediate productivity gains already seen when workers can use AI tools, suggesting the focus should be on augmenting rather than replacing labor. - Current consensus is that AI’s most tangible impact lies in enhancing team and individual effectiveness now, and the notion that AI can fully replace jobs at scale remains premature. ## Sections - [00:00:00](https://www.youtube.com/watch?v=-T-WtXgVlXQ&t=0s) **Microsoft Reaffirms $80B AI Spend Amid Rumors** - Microsoft clarified it will maintain its $80 billion AI data‑center investment despite speculation about OpenAI relocating, while Nvidia announced its latest Blackwell chips are fully ordered. ## Full Transcript
0:00so the internet was full over the 0:01weekend of reports that Microsoft was 0:04somehow delaying data center spend uh in 0:06other words that their 80 billion was 0:08not going to be the 80 billion they said 0:10they'd spend on AI there were rumors 0:12that a couple of data centers were 0:13pushed back Etc well this morning right 0:17when the markets open Microsoft got very 0:19clear and they said we're not changing 0:21our overall 80 billion spend Target you 0:23shouldn't interpret it that 0:25way we may be quote unquote readjusting 0:28things now readjusting would make sense 0:30because open AI is rumored to be moving 0:35to the soft Bank Oracle stack and not 0:38really rumored at this point I think if 0:39you have a presidential press conference 0:41it's more than a rumor uh and at the end 0:45of the day they're the largest tenant 0:47for Microsoft's Azure AI stack right now 0:50and so if they start to move that's 0:52going to change and readjust investment 0:54plans in a private research note I saw 0:58that it was Kenosha Wisconsin and 1:00Atlanta Georgia that are the ones that 1:02they're thinking about moving delaying 1:04adjusting Etc we will have to see how 1:06that plays out but the point is 1:07Microsoft was very keen to put uh paid 1:11to the rumors and say no no no no we're 1:13still spending $80 billion on AI this 1:15year which is adjusting it right 1:17adjusting it as our largest tenant 1:19leaves now at the same time over the 1:22weekend Nvidia made sure that they 1:24leaked that they were full up on chips 1:26and in fact that the blackw chip 1:27architecture their most current one 1:30is full up on orders all the way through 1:322025 and they're also rapidly filling up 1:352026 and so they're broadcasting that 1:38this whole rumored deep seek effect is 1:40also nothing people still need compute 1:43to scale and that's really not 1:45surprising to me like at the end of the 1:47day if after deep seek is released China 1:52announces a massive massive 1:54multi-billion dollar infrastructure 1:56spend they don't have a magic secret key 2:00that enables them to scale AI without 2:02spending on computers it doesn't exist 2:04it's not a thing uh and so yeah of 2:07course people are going to spend on it 2:08and the reason they're going to spend on 2:09it is that the value is still there and 2:11I know that SAA nadela went on a podcast 2:14with dwares and said I he wants to see 2:17AI do economically useful 2:19work I I I think he set the bar really 2:22weird there because if he's asking for 2:25like 10% GDP growth he's basically 2:27ignoring the effects that we see today 2:29which is workers reporting significant 2:31improvements in productivity when they 2:33are allowed to use 2:35AI so why sleep on that right like why 2:39not go after that and that's where a lot 2:40of the people who are focused on 2:42building an AI are thinking about it's 2:44like how do we get teams effective with 2:45AI how do we get individuals effective 2:47with AI um and that's where we see a lot 2:50of the gains and I think that's 2:51interesting because if you're trying to 2:53build AI out right now it may be a 2:57little bit premature to assume that you 2:59can build AI out to the point where you 3:00can just replace someone's job haven't 3:03seen that go well in fact Clara had that 3:04go very very badly last week uh but I 3:09have seen lots of productivity gains 3:12from existing workers getting augmented 3:14with AI that is a very real thing and 3:17whether or not Microsoft wants to pay 3:19attention to it is a different thing 3:21they have different customers right they 3:23don't have consumers they don't have 3:24individuals or teams I realize the irony 3:26of that statement uh they have 3:28Enterprises ENT Rises are who they work 3:30with and so they're going to move at the 3:33pace of the Enterprise and if their 3:34biggest Enterprise client open AI walks 3:36away they they're going to adjust their 3:38data center spend that just kind of 3:39makes sense even if it's overall 80 3:41billion they're going to rearrange it a 3:43little bit it's like salary caps in the 3:46National Football League you rearrange 3:48things a little 3:49bit so that's the story and I think that 3:53I find it very interesting there's a 3:56whole class of people that whenever a 3:58CEO says something that is 4:01remotely I don't know thoughtfully 4:04critical of AI they assume that this 4:06means that the the top of the quote 4:08unquote bubble is in and AI is doomed 4:10etc 4:11etc and and I think you should be able 4:13to have a conversation where SAA can 4:15make a point and say hey I'm really 4:17focused on economically useful work and 4:20even if I disagree with sort of the way 4:22he's playing that set of cards that he 4:24has that's his call and he's fair to 4:28call out that economically useful work 4:30is a good bar and we can have the 4:32conversation without making big 4:33assumptions about it and I think that 4:35one of the things that Microsoft is 4:37realizing is that the hype in the system 4:39is really around 4:41arguments over whether or not we are in 4:44a bubble and that's not super relevant 4:49if you're actually talking about 4:50meaningful investment over time and 4:52that's why the serious players are 4:53talking very clearly in the markets 4:55today and saying we are investing over 4:57time stop it so 5:00by the way ironically if you're one of 5:02those people who thinks we're in a 5:03bubble you know what one of the classic 5:05signs of a bubble is it's not this it's 5:08not people arguing about whether we're 5:09in a bubble it's when everybody gives up 5:12and everybody agrees that whatever we 5:14have going is the greatest thing ever 5:16that's what happened during the famous 5:18tulip crash 400 years 5:20ago yeah so we are not in a bubble by 5:23definition because all of you are 5:25arguing about it cheers yes it's that 5:28cult