The AI Shortcut

Welcome back!

Here are three ideas that stood out this week.

Most discussions about AI focus on models.

But AI doesn't run on hype.

It runs on infrastructure.

As models become more powerful, companies will need enormous amounts of:

• Chips

• Memory

• Storage

• Electricity

The next phase of the AI race may be determined as much by physical infrastructure as by software.

The winners won't necessarily be the companies with the smartest models.

They may be the companies that can support them at scale.

AI assistants and image generators receive most of the attention.

But one of the most exciting applications of AI may be scientific discovery.

Researchers are already using AI to help:

• Discover new materials

• Accelerate drug development

• Predict protein structures

• Analyze enormous datasets

The long-term impact of AI on science could end up being even larger than its impact on productivity.

Today's AI mostly lives on screens.

But over time, AI may increasingly move into the physical world.

Robots powered by increasingly capable AI systems could eventually perform tasks in:

• Warehouses

• Manufacturing

• Healthcare

• Agriculture

• Homes

This transition will likely take years.

But it could become one of the most important technological shifts of the next decade.

• AI agents

• Robotics

• Scientific AI

• AI video generation

• Infrastructure and chips

• OpenAI, Anthropic and Google competition

Thanks for reading.

See you next week.

— The AI Shortcut