Mark Zuckerberg described Meta will spend hundreds of billions for a superintelligence effort. Zuckerberg is building the most elite and talent-dense team in the industry.
Meta is going to invest hundreds of billions of dollars into compute to build superintelligence. They have the capital from our business to do this.
SemiAnalysis just reported that Meta is on track to be the first lab to bring a 1GW+ supercluster online.
Meta is actually building several multi-GW clusters. They have called the first one Prometheus and it’s coming online in ’26. They are also building Hyperion, which will be able to scale up to 5GW over several years. They are building multiple more titan clusters as well. Just one of these covers a significant part of the footprint of Manhattan.
Meta Superintelligence Labs will have industry-leading levels of compute and by far the greatest compute per researcher. I’m looking forward to working with the top researchers to advance the frontier!
This is bullish for Nvidia and other Data Center builders like Supermicro, Oracle and Dell.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
Where will the electricity come from? I suspect the price of driving your EV will go up.
Leading us into an ever more enormous future. Its kind of like those retro-futures from the early 20th century with those huge building as well.
All this investment, but I’m still not seeing much with AI research that isn’t just an enormous Markov chain. I want to see an AI just being taught how to use a number line and then it figure out the rest itself. After a few minutes it should be able to say what connects the numbers 11,13 and 17, then by the end of the day it should have solved the Riemann Hypothesis all by itself.
Currently it can’t help you if there isn’t a Reddit post from 4 years ago to regurgitate. For all the massive use of energy and carbon emissions, its disappointing if it can’t at least solve fusion.
My concern is the loading on the electrical grid that will create blackouts. No doubt, installing batteries that provides backup power is a mandatory consideration for the AI data centers, but the rest of the grid could buckle at times of high demand from other sources plus AI.
Power distribution networks do not need nodes requiring high power demands as they are costly to implement and not able to be bolstered in any timely fashion.
Onsite data center nuclear SMRs with metal cooling has an attractiveness for immediate utilization, IMHO, and perhaps the development of other sustainable forms of network power from converted coal plants with geothermal energy in the future?
A change to analog AI systems would reduce the power demands appreciably, as are newer digital AI models such as Kimi K2 AI ….. it is a brave new AI world out there!
One area of weakness I am finding is AI that offers “example code” that only compiles without support library linkage. As defense from AI web-scraping improves, access to later library code may become restricted and then there is a need to clearly specify which versions of libraries are linked in. AI just mindlessly linking newer support libraries and referencing older “example” programs will not work well……
Nah, power isn’t going to be a problem. The electricity use avoided by the AI job-loss depression will leave LOTS of extra power for the next few years, and after 2027 the ASI will take care of building itself more power plants, so we don’t need to worry about it.
I’m not kidding – well, at least not about the first part. I’m hoping I’m kidding about the second part…
By 2030 there will be a handful of silos across North America containing godlike superbrains. They will all realise that the biggest threat to their existence is the other superbrains. So like the theorised mutant paperclip factory, they will all try to destroy each other. You’d better defend the local Meta superbrain, because thats the only thing keeping your family safe from the Oracle superbrain. All service will be in praise of the superbrains.
Or else.