ANALYSIS: Elon Musk Wants to Make Chips in Seven Days — and No One Should Believe Him

ANALYSIS: Elon Musk Wants to Make Chips in Seven Days — and No One Should Believe Him
Credit: Adobe Stock

The Official Promise: U.S. Independence in Semiconductors

Behind the spectacle lies a strategic argument that isn’t entirely absurd. The United States now manufactures less than 12% of the world’s semiconductors, down from 37% in 1990. Virtually all advanced chips—those powering artificial intelligence, autonomous vehicles, and military satellites—are manufactured in Taiwan, at TSMC’s factories. An island that China considers a rebel province. An island that accounts for more than 90% of global production of sub-7-nanometer chips.

The CHIPS Act, signed in 2022, injected $52.7 billion into the U.S. semiconductor industry. Intel received billions. TSMC is building a facility in Arizona. Samsung is investing in Texas. But Musk looks at this landscape and says: not enough. Not fast enough. Not on the scale needed to power xAI, Tesla, Neuralink, and whatever comes next.

The industrial reality: a chasm between announcement and chip

Building a state-of-the-art semiconductor factory—a “fab,” in the jargon—is no ordinary engineering project. It is the most complex engineering project in existence. TSMC’s P1 fab in Arizona cost over $40 billion and is still not fully operational after years of construction. Intel has spent $20 billion on its factories in Ohio—the first chips aren’t expected until 2027 at the earliest.

A modern fab requires EUV lithography machines manufactured exclusively by ASML in the Netherlands. Each machine costs between $150 million and $380 million. There’s a waiting list that stretches for several years. There are only a few thousand engineers worldwide capable of operating these machines—and most of them already work for TSMC, Samsung, or Intel. And yet, Musk is announcing a launch in seven days. Seven days.

Share this post

facebook icon twitter icon linkedin icon
Copied!