7 months apart
5 Feb 2026
“That $50,000,000,000 of COGS needs to burn on infra, which cost roughly with if a five if you're talking about five year depreciation, call it $250,000,000,000, right, of infra Yeah. For a $100,000,000,000 of revenue.”
spoken ·The MAD Podcast with Matt Turck·source
25 Aug 2026
“As we go forward into the future, the the numbers for computer ballooning, right, we're at, you know, you know, a little bit over a trillion dollars of CapEx this year.”
spoken ·Dwarkesh Patel·source
7 months apart
5 Feb 2026
“That $50,000,000,000 of COGS needs to burn on infra, which cost roughly with if a five if you're talking about five year depreciation, call it $250,000,000,000,”
spoken ·The MAD Podcast with Matt Turck·source
25 Aug 2026
“As we go forward into the future, the the numbers for computer ballooning, right, we're at, you know, you know, a little bit over a trillion dollars of CapEx this year.”
spoken ·Dwarkesh Patel·source
Patel argues AI software will have lower R&D costs but much higher cost of goods sold from operating services.
“Like the R and D cost is much lower in terms of people, but the cost of goods sold in terms of actually operating the service, I think will be much higher.”
Patel predicts OpenAI will raise 50 to 100 billion dollars by early next year to fund planned cluster builds.
“there's no fucking way you can pay for the scale of clusters that are being planned to be built next year for OpenAI until unless they raise, like, 50 to $100,000,000,000, which I think they will raise that, like, end of this year, early next year.”
Patel claims China could build a gigawatt data center in six months near Three Gorges Dam due to power infrastructure.
“China could just build it in six months, I think, around the 3 Gorges Dam or many other places. Right? Because they have they have the ability to do the substations.”
Patel says Microsoft and OpenAI have 500,000 GB200 GPUs totaling one gigawatt coming online next year.
“Microsoft slash OpenAI slash their partners for them. And then and then potentially even more, 500 k GB two hundreds, right, is a gigawatt. Right? And that's, like, online next year.”
Patel claims China could build a gigawatt data center in six months near the Three Gorges Dam but hasn't.
“China could just build it in six months, I think, around the 3 Gorges Dam or many other places. Right?”
Patel reports Microsoft plans 1.5 gigawatts in one site within two and a half years across four 300 megawatt buildings.
“And so the scale that they wanna reach to, 1.5 gigawatts in two and a half years in one site.”
Patel says ordering a substation today takes three years to build due to stagnant grid infrastructure.
“if you were to order a substation today, would take three years to be built. And so it just takes so long because The US power grid has been like this for decades.”
Patel says in some Midwest regions, grid transmission costs exceed power generation costs, constraining data center buildouts.
“It's like, what the flip? Right? Like, it's like so, like, the grid needs huge investments.”
Patel forecasts data center power will grow from 150 megawatts today to gigawatt-scale clusters by 2026.
“We're seeing, like, you know, 400, 500 megawatts cluster sort of being built out now. Right? Like you know? And and then, you know, in 2026, I'd imagine it'll be, like, gigawatt scale clusters.”
Patel breaks down full H100 GPU cost at $40,000 to $45,000 including networking and infrastructure versus $24,000 for chip alone.
“each GPU, right, once you include networking, building, all this sort of stuff maybe is you know, the GPU itself of h 100 is, like, 24,000, but once you add everything else up, it's, like, forty, forty five thousand dollars per GPU all in of everything.”
Patel identifies a 40 gigawatt power shortfall for US data centers even with all planned capacity additions.
“There is an over 40 gigawatt hole in power for The US. Even if we don't turn off any more coal, in fact we restart some, we take all the natural gas dual combined cycle reactor deliveries that we're going to have, and you add in renewables, you have a 40 gigawatt hole of power production in The US.”
Patel identifies a 40 gigawatt power deficit in US even with all planned gas and restarted coal.
“Even if we don't turn off any more coal, in fact we restart some, we take all the natural gas dual combined cycle reactor deliveries that we're going to have,”
Patel calculates networking for 65,000 GPU cluster costs $500 million total, or $55,000 per server.
“You know the total cost of that networking for that cluster is is close to $500,000,000. Right? It's $55,000 per server.”
Patel says certain states will see over half their power consumed by AI within a few years.
“When we look at power consumption by state, something pretty simple, there are certain states where more than half their power is going be consumed by AI just in the next few years.”
Patel argues that efficiency gains alone without capability increases cannot justify massive AI infrastructure investments.
“That would not pay for all of these build outs. Right? AI is useful today, but it's not capable of doing a lot of things.”
Patel says CoreWeave can build $10 billion clusters from start to completion much faster than competitors.
“it lets you get to these, you know, call it $10,000,000,000 clusters, right, in in time scales, from project start to completion that are much much shorter than anyone else.”
Patel notes CoreWeave has fully liquid cooled data centers while hyperscalers are still working on designs.
“They're doing full liquid cooling as well and they're working on those designs, but, you guys have fully liquid cooled data centers.”
Patel describes customer with over 10,000 GPUs whose engineers demanded CoreWeave infrastructure after acquisition.
“What's what's funny is there was a customer that had more than 10,000 GPUs with you. They got acquired.”
Patel describes former Yandex workers paid bonuses for speed who use drugs to complete data center buildouts faster.
“Like and they get paid bonuses for being faster, and therefore, do, like, certain drugs to be able to finish the build outs faster.”
Patel reveals Meta is building temporary tent structures to house GPUs because permanent buildings take too long to construct.
“Meta is literally building these, like, temporary, like, tent structures to put GPUs in because building the building takes too long, and it takes too much labor, as you mentioned labor.”
Patel reports new Trump tax bill allows year-one GPU depreciation, worth $10 billion annually to Meta alone.
“the new the new Trump, you know, tax bill institutes something really incredible, which is that you can depreciate all of the GPU cluster cost in year one, which we put out, like, a note about how, like, the tax implications to, like, meta are, like, $10,000,000,000 a year. And across each of the major hyperscalers, it's, like, massive.”
Patel states 80% of GPU data center cost is capital equipment, only 20% is land, power, and cooling.
“It's the it's the physical data center conversion power conversion equipment. All of this stuff is, like, 80% of the cost. And then 20% is gonna be your land and your power and your cooling”
Patel reports 10 gigawatts of AI infrastructure being deployed collectively in the US over next two years.
“In The US collectively there's like 10 gigawatts going up over the next year and a half, two years.”
Patel says GB 200 servers from Nvidia cost well over $3 million, preventing individual user time slicing.
“Know GB 200 from Nvidia cost well over $3,000,000 so the cost of these things is huge and individual users can't splice these servers.”
Patel describes Meta's temporary tent-like data centers designed for single GPU generation lifecycles instead of 30-year buildings.
“Meta has got a data center design that's only built to last for a single generation. They put them in these tent buildings.”
Patel contrasts traditional data centers with minimal retrofit costs against modern GPU deployments requiring major upgrades each generation.
“it used to be you'd build a data center and there wouldn't really be much retrofits cost, right? You would slide servers in and out, you would rent racks to people.”
Patel quantifies InferenceMax uses tens of millions of dollars in GPU hardware with daily software updates.
“we have tens of millions of dollars of hardware of the current and last generation GPUs. As I mentioned before, the software updates every single day.”
Patel says Microsoft was on pace to beat Amazon as largest infrastructure provider by 2026-27 but paused expansion.
“So last year Microsoft was on path to be the largest infrastructure provider by far.”
Patel explains agentic training only needs to sync tokens every few minutes versus weights every seconds in pre-training.
“When you're doing these rollouts and especially as things get more and more agentic and training, you might not only need to send not the entire weights but just the tokens that are relevant.”
Patel notes semiconductor industry doubles transistors yearly but energy industry wasn't prepared for similar growth.
“Whereas the energy industry in America wasn't. And and so, like, initially, people were, like, not creative. They're like, let's do let's do these kinds of gas plants.”
Patel calculates $100B AI revenue requires $250B infrastructure with five-year depreciation at 50% margins.
“That $50,000,000,000 of COGS needs to burn on infra, which cost roughly with if a five if you're talking about five year depreciation, call it $250,000,000,000,”
Patel calculates $100B AI revenue requires $250B infrastructure spend at five-year depreciation, double current capex.
“That $50,000,000,000 of COGS needs to burn on infra, which cost roughly with if a five if you're talking about five year depreciation, call it $250,000,000,000, right, of infra Yeah. For a $100,000,000,000 of revenue.”
Patel says Google spending $180B and Amazon $200B on AI infrastructure represents 4x increase from recent years.
“If we start looking at like, hey, this year, Google spending 2 Amazon spending $200,000,000,000, Google spending a $180,000,000,000 on on AI infrastructure primarily. Right?”
Patel forecasts data center capacity will reach 30 gigawatts in the next year.
“This year, we're deploying 20 gigawatts of data centers. Next year, that number goes up 50 gigawatts or 50%, sorry. 50%, sorry.”
Patel says a data center is physically redoing doors to accommodate the double-width Helios racks.
“There's a there's a data center that I know of specifically that is going to be deploying Helios, and they are redoing the doors so that they can reel the Helios in.”
Patel asks Douglas if he can use the model without privacy protections
“@_sholtodouglas Can I use your model with none of my privacy.”
Patel forecasts AI CapEx will exceed $2 trillion by 2028, up from just over $1 trillion in 2025.
“As we go forward into the future, the the numbers for computer ballooning, right, we're at, you know, you know, a little bit over a trillion dollars of CapEx this year.”
Patel questions where $3-4 trillion in CapEx will come from for 2028 AI infrastructure buildout.
“So if you're at 3 or $4,000,000,000,000 of CapEx, where does all this cash come from?”
Patel estimates $3-4 trillion total CapEx needed by 2028 across compute, data centers, and energy infrastructure.
“So to enable, let's say, that 100 gigawatts by 2030 or let's even like let's even like pare it down to 2028 where it's like 3 or $4,000,000,000,000 of CapEx across all of these items.”