ChatGPT Designs Cancer Vaccine

2026-03-14 · Source: There's An AI For That · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Robotics & Autonomous Systems · Depth: Advanced, extended

Summary

This content provides a comprehensive overview of the AI infrastructure landscape, highlighting critical bottlenecks and market dynamics. It details Google's demonstration of AI smart glasses at MWC, xAI's need for a "full rebuild" as stated by Elon Musk, and Uber's former CEO launching Atoms, a robotics company. A key focus is the story of an Australian tech executive using ChatGPT and AlphaFold to design a custom mRNA cancer vaccine for his dog, Rose, which led to a 75% tumor reduction. The analysis also delves into the semiconductor supply chain, identifying ASML's EUV machines as the ultimate bottleneck for AI compute scaling by 2028-2029, with only about 100 units projected annually by 2030. It further discusses the escalating memory crunch, projecting significant price increases for consumer electronics as DRAM shifts to AI, and explores the challenges and potential solutions for power and labor constraints in data center expansion.

Key takeaway

For CTOs and VPs of Engineering planning long-term AI infrastructure, your strategy must account for the semiconductor supply chain's limitations. Prioritize securing long-term contracts for advanced logic and memory, as these will be the primary constraints on compute capacity through 2030. Diversify your compute providers beyond traditional hyperscalers and explore modular data center designs to mitigate future labor and power bottlenecks, ensuring your AI initiatives can scale effectively.

Key insights

Semiconductor manufacturing, particularly EUV lithography, is the ultimate bottleneck for AI compute scaling, not power or data centers.

Principles

• Compute efficiency gains from research outweigh development allocation.
• Smaller models with faster RL cycles accelerate research feedback loops.
• Market inefficiencies in compute capacity can be arbitraged by early commitment.

Method

An Australian tech executive used ChatGPT to sequence DNA mutations from a tumor and AlphaFold to predict 3D protein structures, enabling the manufacture of a custom mRNA cancer vaccine.

In practice

• Consider behind-the-meter power solutions for data centers.
• Explore modular data center designs to mitigate labor constraints.
• Prioritize smaller models for faster RL and research iteration.

Topics

• AI Compute Bottlenecks
• EUV Lithography
• HBM Memory
• Data Center Infrastructure
• AI Labs Strategy

Code references

• LMCache/LMCache
• browserbase/stagehand
• lightpanda-io/browser

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, Research Scientist, Investor

Related on AIssential

• Intel CEO on Elon Musk — The semiconductor infrastructure lags AI growth, creating a critical bottleneck for future development.
• GTC preview: Inside the AI factory — The $1T infrastructure war under the hood of the AI economy — The AI industry is shifting from software to industrial-scale intelligence manufacturing, driven by massive infrastructure investments and supply chain constraints.
• Google's in a CRAZY position — Proactive infrastructure investment and manufacturing-based data center deployment mitigate AI compute constraints.
• Two Events Illustrate How AI is Rewriting the Memory Book — AI demand is driving unprecedented HBM shortages, reshaping the global memory industry landscape.
• The $1 Trillion Data Center Crisis: Why America’s AI Ambitions Are Built on Crumbling… — Current data center infrastructure is unsustainable, demanding a radical redesign for AI and quantum computing's resource needs.
• How DeepSeek V4 Connects to the US Power Grid — Energy infrastructure is the new frontline in the US-China AI competition, impacting compute availability and geopolitical strategy.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by There's An AI For That.