Growing Pineapples in North Dakota
Around the time I researched and wrote The Spirit of Enterprise (1985), chiefly about Micron Technology of Boise, Idaho, the great Peter Drucker made a famous declaration: “To make microchips in America is like growing pineapples in North Dakota.”
Peter Drucker was the most insightful business writer of all time (“pursue effectiveness not efficiency,” “don’t solve problems”). I quote him often.
But in the pungent statement above, Drucker revealed he didn’t really comprehend the significance of the technology of microchips.
As the chief writer of the (Ben) Rosen Electronics Letter at the time, I believed that the entire world economy would be transformed by microchips.
After all, in the 1950s, the first microchips had empowered the US to answer Soviet advances in liquid-fueled intercontinental missile propulsion by enabling MIRVs (multiple independently targetable reentry vehicles).
The Soviets could still deliver a bigger payload to the US or into space. But using minuscule microchip controllers, the US could deliver its smaller payload onto several different targets from one smaller solid-fueled rocket.
My chief source of wisdom on the subject, Carver Mead of Caltech, had told me that microchips of the same size would eventually contain not thousands but billions of transistors. I put him on the cover of Forbes, with a quote, “You ain’t seen nothing yet.”
In effect, Drucker was saying that the US could not make the most important technology of the time and was doomed to become a second-class power. In those years, the Japanese utterly dominated the memory chip business. Drucker assumed that they possessed an invincible advantage.
Personally, having visited many wafer fabs and foundries in the United States, I did not believe that Drucker was right.
“You Ain’t Seen Nothing Yet”
One year later in 1986, I was able to personally hand President Ronald Reagan an industry-leading 64 kilobit dynamic random-access memory chip made in Idaho by Micron on a silicon die just 14 millimeters square (as I recall). That made it the smallest DRAM chip and thus the cheapest to produce.
I told the president that this tiny device on a sliver of sand the size of a fingernail would change the world.
It would radically alter the relationship between offense and defense in warfare by enabling cost effective missile defense. It also would launch Micron on a path that eventually would make it the world’s leading maker of memory chips and from time to time one of Wall Street’s most lucrative stocks.
Semiconductors matter. When I want to grasp the future, I still investigate what is going on in semiconductors.
Using microchips, President Trump killed General Qassem Suleimani.
Using microchips, the Iranians have been trying to hack our power grid, while our climate changers attempt to hack it with windmills and solar panels, and Californian politicians seek to tear it down entirely in twenty years.
So today I have a number of observations and prophecies to make about microchips, found below.
Introducing: The International Microchip Leaders
The US is no longer in the lead. The Taiwanese are.
Taiwan Semiconductor Manufacturing Corp. (TSMC), started by Morris Chang of the once world-leading Texas Instruments, is the world’s most proficient maker of microchips.
At 7 nanometer geometries going down to 3 nanometer, TSMC is way ahead of US fabs on the semiconductor roadmap. Contrary to some apparent assumptions in Washington, Taiwan is not a county in California.
The world leader in analog and mixed signal devices, however, is TowerJazz of Migdal Haemek in Israel (though its Jazz unit is in Newport Beach, California, and it also makes chips in Texas and Japan).
The largest single asset of US technology is Israel. We would be nowhere without it.
Compared to Israel, all of our other interests in the Middle East are utterly trivial and picayune piles of sand and savagery. Passing his Israel Test, President Trump actually may almost grasp that fact.
The leading chip designers are not in the US. They’re in Israel and even China.
Intel remains a key player chiefly because it is primarily an Israeli company. Its two leading edge divisions are Israeli purchases Mobileye and Habana and its leading fab is in Kiryat Gat, Israel.
According to industry benchmark tests, though, China’s Alibaba produces the world’s paramount artificial intelligence (AI) chip.
I recently visited Alibaba in Beijing and I could make a case that among a fantastic array of Chinese innovators, from Tencent and Huawei to Ping An, Alibaba is the world’s most accomplished company.
That said, this prophecy is not about China or Israel. It’s not about new Micron memory architectures, either (I’ll get to them later).
It’s about that same old stretch of sand and silliness, denim and denial, invention and fatuity, palm trees and foodie fetishes, fraught with the same old fantasies of climate cataclysm, fears of AI and robopocalypse, revels of Warriors and Forty Niners and “quantum supremacy,” stretching south of San Francisco.
Silicon Valley is Transforming Microchip Technology
Emerging last month in Los Altos, California, in the very heart of Silicon Valley, down the hill from Dr. Nick Tredennick of the original Gilder Technology Report, was the most important development in the global semiconductor industry.
It was a watershed turning of the tide of technological history. A permanent shift in the balance of power signified by Micron’s tiny DRAM in 1984.
From now on, the most important measure of microchip achievement will not be how small the dies are ― but how big.
Gone are the days of microchips the size of a fingernail. To signal the changing of the guard since Reagan, I would like to hand President Trump a microchip the size of a Mar-a-Lago Club dinner plate.
This technology can change the world economy.
I have to confess that I’ve been contemplating this change for decades. Ever since IBM chip design titan Gene Amdahl launched Trilogy Systems Corp. to achieve what is called wafer-scale integration (WSI).
WSI involves putting an integrated computer system on a single silicon wafer.
Here’s why that is significant. Today, ninety percent of the space occupied by electronic systems is on printed circuit boards ― not on silicon chips.
Deployed on them are passive elements such as resistors, capacitors, inductors and power busses that condition power flows, transmit signals, and deliver currents to microchips with exquisite precision.
The key goal of the industry must ultimately be to integrate these passive elements that currently defy the nanoscale imperatives of microchips, onto wafer-scale devices.
No one has done this yet, though there are important stealth projects under way that promise to accomplish this crucial grail.
The epochal step forward, however, has already been taken.
In Los Altos, the startup Celebras Systems launched an AI chip on a wafer. It promises to accelerate the process of training a neural network for AI from days or weeks to a few hours.
Hey, it may not be “quantum supremacy.” But it has the power to change everything, as I will explain tomorrow.
Editor, Gilder’s Daily Prophecy