AI Capabilities Double Every 7 Months - Faster Than Moore's Law

Remember when we thought Moore's Law was impressive? There's a new exponential growth curve in town, and it's leaving traditional computing power in the dust.

According to research from METR , AI systems' ability to complete coding tasks successfully has been doubling every 7 months since 2019.

Let that sink in.

While computer hardware following Moore's Law improves 100% every 24 months, AI coding abilities improve 100% every 7 months – more than three times faster!

What Makes This Growth So Remarkable?

This isn't just about AI getting marginally better at existing skills. We're witnessing a fundamental expansion in what AI can accomplish.

The key metric here is the length and complexity of coding tasks these systems can successfully tackle.

In practical terms, this means an AI that could only write a simple function to calculate factorials in 2019 might now be capable of building an entire e-commerce platform with authentication, payment processing, and inventory management.

Beyond Moore's Law: A New Paradigm

Moore's Law has been an impressive driver of technological advancement, but it requires physical changes to computer chips – cramming more transistors onto silicon wafers. It's bound by the constraints of the physical world.

AI capability growth, however, stems from improvements in algorithms, training methods, and data usage – areas not directly bound by traditional manufacturing constraints. While there are potential limiters – like GPU availability and power requirements for massive data centers – the current trajectory shows eye-popping rates of improvement that dramatically outpace conventional hardware evolution.

What This Means For Our Future

If this exponential growth continues, we're heading into uncharted territory. Tasks that seem impossibly complex for AI today could become trivial in just a few years. The implications span every industry:

• Software Development: AI could soon handle increasingly complex development projects autonomously, as demonstrated by advances in RAG (Retrieval Augmented Generation) technology
• Healthcare: Medical diagnostic systems might tackle cases of unprecedented complexity
• Scientific Research: AI could help solve problems that have stumped human researchers for decades, particularly in breaking down knowledge silos that often hinder research progress
• Mega Projects: As highlighted on the "All In" podcast, David Friedberg notes that AI could dramatically simplify planning for massive infrastructure projects, allowing teams to focus more on execution rather than getting lost in years of preparation

The Human Element

Despite this explosive growth, human expertise remains crucial. As AI tackles bigger challenges, we need skilled professionals to:

• Define meaningful problems for AI to solve
• Verify and validate solutions
• Integrate AI capabilities into meaningful products and services
• Navigate the ethical implications of increasingly capable systems

What's Next?

The METR research hints at a future where the bottleneck isn't computing power but our imagination. As AI capability growth continues to accelerate, our biggest challenge may be figuring out the most beneficial ways to apply these rapidly evolving tools. For a practical example of how this rapid AI advancement is being applied, see our guide on implementing HR chatbots .

The next few years promise to be a wild ride as we navigate this new exponential frontier. One thing is clear: we're no longer in the era of incremental improvements. We've entered the age of compounding AI capabilities, and the pace shows no signs of slowing down.