Discussion about this post

User's avatar
Aegis's avatar

Having “infinite” (or even just a few orders of magnitude more compute would be huge for AI. AI progress is currently bottlenecked by the number of experiments one can run on current compute budgets.

No one believes the transformer is the best possible architecture. With 10,000x more compute you could run neural architecture search to look for better transformers. With 10,000x more compute you have the resources to run to scale up the algorithms from hundreds of promising but compute starved papers.

AI progress is almost a direct function of compute. The reason we didn’t have the transformer revolution when the original paper came out in 2016 is because the compute available then was 5 or 6 orders of magnitude less than what we have today.

I’m not the quantum computing expert, but it’s my understand that quantum computers are only good for narrow domains and can never run any type of AI. If so then I agree with your thesis. But if it turns out quantum computing can be directly used in AI, I’d be very bullish about the technology.

Expand full comment
Qryptonic LLC's avatar

A Qryptonic, LLC Research Team Rebuttal

Head of Quantum Computing, Qryptonic

Martin Shkreli’s recent piece, “Quantum Computing: 2075?” raises an intriguing mix of skepticism and pragmatism about this emerging field. We at Qryptonic appreciate his willingness to challenge big ideas—after all, a little doubt can drive meaningful innovation. But in the words of A Tribe Called Quest, it’s time to “Check the Rhime” on quantum computing’s true potential.

1. Demand for Quantum Is Real

While classical HPC and GPUs remain critical, quantum computing addresses problems classical systems struggle with—like large-scale optimization, certain AI subroutines, and molecule-level simulations in pharmaceuticals. Major corporations and governments aren’t investing billions in quantum R&D for novelty; they see the path to practical applications and returns.

2. Concrete Use Cases

Drug Discovery: More accurate quantum simulations of molecular interactions could shorten R&D cycles and save billions—a significant incentive for pharma giants.

Post-Quantum Security: Even with new cryptographic standards like Dilithium and Kyber, “harvest-now, decrypt-later” threats make quantum-safe solutions essential.

3. Market Timelines Aren’t All or Nothing

“Quantum winters” may come and go, but history shows us that setbacks often precede breakthroughs. IBM, Google, Microsoft, and others have released tangible roadmaps. Incremental gains in error correction, qubit fidelity, and quantum software prove we’re not simply guessing—we’re building toward real milestones.

4. The Bigger Picture

Skepticism aside, quantum computing’s lure isn’t just raw speed; it’s the unique advantage of harnessing quantum effects for complex tasks. Even if full-scale, fault-tolerant machines are still on the horizon, intermediate systems are already finding niche applications.

In closing, we appreciate Martin Shkreli’s candor and contrarian insights—much like a rare hip-hop collectible, disruptive ideas can spark excitement (or controversy!). But if we “Check the Rhime” in the quantum community, it’s clear that breakthroughs will arrive far sooner than 2075.

At Qryptonic, we remain confident that quantum computing will prove its worth in cybersecurity, pharmaceuticals, AI, and beyond—offering more than just a speculative dream, but a practical reality that’s unfolding faster than many realize.

Qryptonic, LLC Research Team

www.qryptonic.com

“Securing Tomorrow, Today.”

Expand full comment
4 more comments...

No posts