The AI race may be entering a new phase. The limiting factor is no longer simply creating more intelligence, but packaging, deploying, and integrating the intelligence we already possess. As the cost of compute and AI capability continues to fall, value is likely to migrate away from raw model performance and toward trusted deployment, reliable infrastructure, workflow integration, verification, and real-world execution. At the same time, governments are competing for scarce capital while financing record debt, and industry is investing hundreds of billions of dollars in data centers, power generation, and semiconductor infrastructure before the long-term technological architecture is settled. The winners of the next decade may therefore not be those with the largest models or the most compute, but those who most effectively combine abundant intelligence with dependable energy, adaptable infrastructure, disciplined capital allocation, and systems that people and institutions can confidently rely upon
Can’t help but think that some day many data centers will fall prey to the same fate as current shopping malls. Just a thought, but I barely made it out of a state school college.
What technology doesn't get lapped? And now with AGI and ASI coming jaggedly, who knows? It may be that earthbound data centers will become shopping malls. It may be that onprem compute is all that everyone needs. But that's a guess.
Spot on and ahead of the curve. @Dr. Alex Wissner-Gross and @Peter H. Diamandis MOONSHOTS are my go to source. Reminds me of Lewis Carroll's Through the Looking-Glass, the Red Queen explains the strange logic of her world to Alice with this famous idea: "Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that!"
The AI race may be entering a new phase. The limiting factor is no longer simply creating more intelligence, but packaging, deploying, and integrating the intelligence we already possess. As the cost of compute and AI capability continues to fall, value is likely to migrate away from raw model performance and toward trusted deployment, reliable infrastructure, workflow integration, verification, and real-world execution. At the same time, governments are competing for scarce capital while financing record debt, and industry is investing hundreds of billions of dollars in data centers, power generation, and semiconductor infrastructure before the long-term technological architecture is settled. The winners of the next decade may therefore not be those with the largest models or the most compute, but those who most effectively combine abundant intelligence with dependable energy, adaptable infrastructure, disciplined capital allocation, and systems that people and institutions can confidently rely upon
"Genius now ships with a unit price."
Can’t help but think that some day many data centers will fall prey to the same fate as current shopping malls. Just a thought, but I barely made it out of a state school college.
What technology doesn't get lapped? And now with AGI and ASI coming jaggedly, who knows? It may be that earthbound data centers will become shopping malls. It may be that onprem compute is all that everyone needs. But that's a guess.
Spot on and ahead of the curve. @Dr. Alex Wissner-Gross and @Peter H. Diamandis MOONSHOTS are my go to source. Reminds me of Lewis Carroll's Through the Looking-Glass, the Red Queen explains the strange logic of her world to Alice with this famous idea: "Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that!"
“Bitcoin as unproductive as ever” is spot on. 1.3T for a whole bunch of nothing