Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers
https://apnews.com/article/trump-ai-openai-oracle-softbank-son-altman-ellison-be261f8a8ee07a0623d4170397348c41maybe i am getting to old or to friendly to humans, but it's staggering to me how the priorities are for such things.
Instead we gave a small number of people all of this money for a moonshot in a state where they squabble over who’s allowed to use which bathroom and if I need an abortion I might die.
I don't know how to make sense of this level of investment. I feel that I lack the proper conceptual framework to make sense of the purchasing power of half a trillion USD in this context.
Maybe your calibration isn't poor. Maybe they really are all wrong but there's a tendency here to these these people behind the scenes are all charlatans, fueling hype without equal substance hoping to make a quick buck before it all comes crashing down, but i don't think that's true at all. I think these people really genuinely believe they're going to get there. And if you genuinely think that, them this kind of investment isn't so crazy.
1) reasoning capabilities in latest models are rapidly approaching superhuman levels and continue to scale with compute.
2) intelligence at a certain level is easier to achieve algorithmically when the hardware improves. There's also a larger path to intelligence and often simpler mechanisms
3) most current generation reasoning AI models leverage test time compute and RL in training--both of which can make use of more compute readily. For example RL on coding against compilers proofs against verifiers.
All of this points to compute now being basically the only bottleneck to massively superhuman AIs in domains like math and coding--rest no comment (idk what superhuman is in a domain with no objective evals)
My sense anecdotally from within the space is yes people are feeling like we most likely have a "straight shot" to AGI now. Progress has been insane over the last few years but there's been this lurking worry around signs that the pre-training scaling paradigm has diminishing returns.
What recent outputs like o1, o3, DeepSeek-R1 are showing is that that's fine, we now have a new paradigm around test-time compute. For various reasons people think this is going to be more scalable and not run into the kind of data issues you'd get with a pre-training paradigm.
You can definitely debate on whether that's true or not but this is the first time I've been really seeing people think we've cracked "it", and the rest is scaling, better training etc.
This sort of $100-500B budget doesn't sound like training cluster money, more like anticipating massive industry uptake and multiple datacenters running inference (with all of corporate America's data sitting in the cloud).
I think what's been going on is compute/$ has been exponentially rising for decades in a steady way and has recently passed the point that you can get human brain level compute for modest money. The tendency has been once the compute is there lots of bright PhDs get hired to figure algorithms to use it so that bit gets sorted in a few years. (as written about by Kurzweil, Wait But Why and similar).
So it's not so much brute forcing AGI so much that exponential growth makes it inevitable at some point and that point is probably quite soon. At least that seems to be what they are betting.
The annual global spend on human labour is ~$100tn so if you either replace that with AGI or just add $100tn AGI and double GDP output, it's quite a lot of money.
The thing about investments, specifically in the world of tech startups and VC money, is that speculation is not something you merely capitalize on as an investor, it's also something you capitalize on as a business. Investors desperately want to speculate (gamble) on AI to scratch that itch, to the tune of $500 billion, apparently.
So this says less about, 'Are we close to AGI?' or, 'Is it worth it?' and more about, 'Are people really willing to gamble this much?'. Collectively, yes, they are.
Can't answer that question, but, if the only thing to change in the next four years was that generation got cheaper and cheaper, we haven't even begun to understand the transformative power of what we have available today. I think we've felt like 5-10% of the effects that integrating today's technology can bring, especially if generation costs come down to maybe 1% of what they currently are, and latency of the big models becomes close to instantaneous.
Remember Trump's BIG WIN of Foxconn investing $10B to build a factory in Wisconsin, creating 13000 jobs?
That was in 2017. 7 years later, it's employing about 1000 people if that. Not really clear what, if anything, is being made at the partially-built factory. [0]
And everyone's forgotten about it by now.
I expect this to be something along those lines.
[0] https://www.jsonline.com/story/money/business/2023/03/23/wha...
Or they think the odds are high enough that the gamble makes sense. Even if they think it's a 20% chance, their competitors are investing at this scale, their only real options are keep up or drop out.
“My NEW Official Trump Meme is HERE! It's time to celebrate everything we stand for: WINNING! Join my very special Trump Community. GET YOUR $TRUMP NOW.”
Your calibration is probably fine, stargate is not a means to achieve AGI, it’s a means to start construction on a few million square feet of datacenters thereby “reindustrializing America”
It rather means that they see their only chance for substantial progress in Moar Power!
for instance: https://en.wikipedia.org/wiki/2024_United_States_telecommuni...
Didn't go well for South America in the 60s and 70s but perhaps, as economists are prone to saying, "this time will be different".
South America didn't have a mix of domestic and foreign investors deploying massive quantities of private money into capital assets in the 60s and 70s. They had governments borrowing to fund their citizens' consumption. Massive difference on multiple levels.
I'm not saying "this time will be different". I'm saying this is business as usual.
The fact that a handful of individuals have half a trillion dollars to throw at something that may or may not work while working people can pay the price of a decent used car each year, every year to their health insurance company only to have claims denied is insane.
This is disputed [1]. In reality, a handful of individuals have the capital to seed a half-a-trillion dollar megaproject, which then entails the project to raise capital from more people.
[1] https://www.wsj.com/tech/musk-pours-cold-water-on-trump-back...
This money is managed by small amounts of people but it is aggregated from millions of investors, most of these are public companies. The US spends over 10x that amount on healthcare each year.
au·tar·ky /ˈôˌtärkē/ noun economic independence or self-sufficiency. "rural community autarchy is a Utopian dream" a country, state, or society which is economically independent. plural noun: autarkies; plural noun: autarchies
This is first-mover industrial development being funded by private actors looking out for a return on their investments. South America saw nothing similar--it was duplicating others' industrialisation with state capital (often borrowed from overseas) while spending massively on handouts.
'you should have spent all this time and money fighting climate change'
2. What are they doing? AGI/ASI is a neat trick, but then what? I’m not asking because I don’t think there is an answer; I’m asking because I want the REAL answer. Larry Ellison was talking about RNA cancer vaccines. Well, I was the one that made the neural network model for the company with the US patent on this technique, and that pitch makes little sense. As the problem is understood today, the computational problems are 99% solved with laptop-class hardware. There are some remaining problems that are not solved by neural networks, but by molecular dynamics, which are done in FP64. Even if FP8 neural structure approximation speeds it up 100x, FP64 will be 99% of the computation. So what we today call “AI infrastructure” is not appropriate for the task they talk about. What is it appropriate for? Well, I know that Sam is a bit uncreative, so I assume he’s just going to keep following the “HER” timeline and make a massive playground for LLMs to talk to each other and leave humanity behind. I don’t think that is necessarily unworthy of our Apollo-scale commitment, but there are serious questions about the honest of the project, and what we should demand for transparency. We’re obviously headed toward a symbiotic merger where LLMs and GenAI are completely in control of our understanding of the world. There is a difference between watching a high-production movie for two hours, and then going back to reality, versus a never-ending stream of false sensory information engineered individually to specifically control your behavior. The only question is whether we will be able to see behind the curtain of the great Oz. That’s what I mean by transparency. Not financial or organizational, but actual code, data, model, and prompt transparency. Is this a fundamental right worth fighting for?
This 5GW data centre capacity very roughly equates to 350000x NVIDIA DGX B200 (with 14.3kW maximum power consumption[4] and USD$500k price tag[5]) which if NVIDIA were selected would result in a very approximate total procurement of USD$175b from NVIDIA.
On top of the empty data centres and DGX B200's and in the remaining (potential) USD$265b we have to add:
* Networking equipment / fibre network builds between data centres.
* Engineering / software development / research and development across 4 years to design, build and be able to use the newly built infrastructure. This was estimated in mid 2024 to cost OpenAI US$1.5b/yr for retaining 1500 employees, or USD$1m/yr/employee[7]. Obviously this is a fraction of the total workforce needed to design and build out all the additional infrastructure that Microsoft, Oracle, etc would have to deliver.
* Electricity supply costs for current/initial operation. As an aside, these costs seemingly not be competitive with other global competitors if the USA decides to avoid the cheapest method of generation (renewables) and instead prefer the more expensive generation methods (nuclear, fossil fuels). It is however worth noting that China currently has ~80% of solar PV module manufacturing capacity and ~95% of wafer manufacturing capacity.[10]
* Costs for obtaining training data.
* Obsolescence management (4 years is a long time after which equipment will likely need to be completely replaced due to obsolescence).
* Any other current and ongoing costs of Microsoft, Oracle and OpenAI that they'll likely roll into the total announced amount to make it sound more impressive. As an example this could include R&D and sustainment costs in corporate ICT infrastructure and shared services such as authentication and security monitoring systems.
The question we can then turn to is whether this rate of spend can actually be achieved in 4 years?
Microsoft is planning to spend USD$80bn building data centres in 2025[7] with 1.5GW of new capacity to be added in the first six months of 2025[3]. This USD$80bn planned spend is for more than "Stargate" and would include all their other business units that require data centres to be built, so the total required spend of USD$45b-$75b to add 5GW data centre capacity is unlikely to be achieved quickly by Microsoft alone, hence the apparent reason for Oracle's involvement. However, Oracle are only planning a US$10b capital expenditure in 2025 equating to ~0.8GW capacity expansion[9]. The data centre builds will be schedule critical for the "Stargate" project because equipment can't be installed and turned on and large models trained (a lengthy activity) until data centres exist. And data centre builds are heavily dependent on electricity generation and transmission expansion which is slow to expand.
[1] https://news.ycombinator.com/item?id=39869158
[2] https://www.datacenterdynamics.com/en/news/microsoft-openai-...
[3] https://www.datacenterdynamics.com/en/news/microsoft-to-doub...
[4] https://resources.nvidia.com/en-us-dgx-systems/dgx-b200-data...
[5] https://wccftech.com/nvidia-blackwell-dgx-b200-price-half-a-...
[6] https://www.cushmanwakefield.com/en/united-states/insights/d...
[7] https://blogs.microsoft.com/on-the-issues/2025/01/03/the-gol...
[8] https://www.datacenterdynamics.com/en/news/openai-training-a...
[9] https://www.crn.com.au/news/oracle-q3-2024-ellison-says-ai-i...
[10] https://www.iea.org/reports/advancing-clean-technology-manuf...