“The computer age is over.”

That is the first sentence of George Gilder’s 2000 book Telecosm: How Infinite Bandwidth Will Revolutionize Our World. It landed at the peak of the dot-com boom, when the computer industry was the most profitable and celebrated sector of the global economy — Intel was making $10 billion a year, Microsoft was fighting antitrust battles over its desktop monopoly, and every teenager in America wanted a computer science degree.

Gilder said the whole thing was already a sideshow.

The real action, he argued, was shifting from the microprocessor to the network — from processing power to communications power, from the microcosm of the chip to the telecosm of the fiber. The defining abundance of the next era would be bandwidth, not transistors. The prophet of the free economy — the man who had handed Reagan a microchip a few years earlier — was now telling the world the chip was no longer where the money was. The rest of the Guru Files follows the arc that kept moving.

People laughed at him. Then they stopped laughing.

The Telecosm Thesis

Gordon Moore’s famous observation from 1965 — that transistor density on a chip doubles roughly every 18 months — had been the governing law of the technology industry for thirty-five years. Every product roadmap, every stock thesis, every venture capital bet was built on the assumption that computing power was the scarce resource and the one that mattered most.

Gilder argued this was backwards.

Bandwidth, he said, was growing ten times faster than computing power. Fiber optic technology was advancing at a pace that made Moore’s Law look slow. A single fiber-optic thread, as thick as a human hair, could carry more information than all the wireless spectrum then in use. A cable with 864 such fibers, running at the speeds Wavelength Division Multiplexing made possible, could carry petahertz of data — a million times the bandwidth of all AM radio to satellite TV combined.

The implication was both technical and economic. When bandwidth becomes the defining abundance, the entire architecture of computing inverts — computers no longer need to be smart at the center, they can be dumb at the center and smart at the edges, and the network becomes the computer.

Eric Schmidt, then CTO of Sun Microsystems, had been saying something similar: “The network is the computer.” Gilder gave the argument its full theoretical weight. If bandwidth expands faster than processing power, the value in the system shifts from the CPU to the connection. Moore’s Law still matters. But Gilder’s Law — that bandwidth doubles every six months — matters more.

The Dumb Network Wins

The most controversial part of the Telecosm thesis was what Gilder called the “dumb network.”

In the 1990s, the telecom industry was betting heavily on intelligent networks. ATM switches that could prioritize traffic. SONET rings that could self-heal. Complex protocols that managed every packet like an air traffic controller manages planes. The assumption was that bandwidth was scarce and precious, so the network had to be smart about using it.

Gilder said that was backward. If bandwidth becomes infinite and cheap, the network does not need intelligence — it just needs to move bits as fast as possible. The intelligence moves to the edges, to the computers, phones, and devices connected to it. The network becomes a pipe, a very fast, very dumb, very cheap pipe.

This was heretical to the telecom establishment. Lucent, Nortel, and Alcatel were selling billions of dollars of smart networking equipment, and the entire SONET ecosystem was built on the opposite assumption. But Gilder was right — the dumb network is how the internet actually works. The intelligence is in your phone and in the server at the other end. The network in between just moves packets.

Bob Metcalfe, the inventor of Ethernet, had predicted in 1996 that the internet would collapse under its own weight — a “gigalapse” that would bring the web to its knees. By 2000, he had conceded Gilder was right, because the bandwidth explosion made the collapse impossible. There was too much fiber, too much capacity, too much dumb pipe.

The Scorecard

The Telecosm record is better than most technology predictions from 2000.

What Gilder got right.

Fiber optics dominance. Gilder argued that fiber would become the backbone of global communications. Today, 99% of intercontinental data traffic runs through undersea fiber cables. The fibersphere he described in 2000 is reality.

Qualcomm and CDMA. Gilder was early on Qualcomm, writing about CDMA cellular technology as a spectrum-reuse revolution. Qualcomm’s stock had already surged 27x in 1999. It went on to become a $150 billion company that collects licensing fees on every 3G, 4G, and 5G phone sold. His fundamental insight — that software-defined radio and spectrum reuse would dominate — was correct.

The decline of SONET and ATM. Both are essentially dead. Ethernet and IP ate them. The dumb network won.

Handheld computing. Gilder predicted that computers would “hollow out” as their components migrated to the network and to handheld devices. This was written in 2000, seven years before the iPhone. He described a world where your phone was your computer, where the network did the heavy lifting, where “Cray on a tray” processing power could sit in a cable box. It took longer than he expected. But the direction was right.

Bandwidth abundance. This was the core prediction, and it turned out to be the most prescient. Gilder said bandwidth would become so cheap and plentiful that it would be treated as a free resource. In 2000, a T1 line (1.5 Mbps) cost hundreds of dollars a month. Today, a gigabit fiber connection costs less than a dinner out. The price of bandwidth has dropped by a factor of roughly 10,000 in twenty-five years.

What he got early.

Gilder was not right on everything. He was early on a lot.

The bandwidth glut of 2001-2003. Gilder argued through the dot-com bust that there was no bandwidth glut — only a connectivity bottleneck between backbones and enterprises. The market disagreed violently. Global Crossing, Level 3, Metromedia Fiber, and Williams Communications all went bankrupt or near-bankrupt. More than a trillion dollars of telecom market cap evaporated. Gilder was directionally right that the demand would eventually fill the pipes. But he was early by about five years, and being early in a leveraged sector is the same as being wrong.

His stock picks. Gilder’s Gilder Technology Report recommended specific companies, and the track record is mixed. Qualcomm and Broadcom worked. JDS Uniphase and Corning did not. The “Nine Stars of the Telecosm” list from the book includes companies that mostly got acquired or went to zero. The thesis was better than the stock selection.

Timing of the broadband consumer revolution. Gilder predicted in 1994 that the world would have “infinite bandwidth in a decade.” It took closer to two decades. The last-mile problem was harder than he assumed. Municipal broadband, cable DOCSIS upgrades, and fiber-to-the-home all took longer to deploy than his timeline projected.

The Deeper Argument

The surface read of Telecosm is about technology. The deeper read is about how scarcity and abundance shape economic eras.

Gilder’s framework is consistent across all his books — every era has a defining abundance. In the industrial age it was steam power, in the computer age it was transistors, and in the telecosm it is bandwidth. Each abundance solves the scarcity of the prior era and creates a new scarcity that the next era will solve.

This is why Gilder never stayed in one lane. He followed the abundance. When microchips were the defining abundance (1980s-90s), he wrote about chips and the microcosm. When bandwidth became the defining abundance (late 1990s onward), he wrote about fiber and the telecosm. When data center compute became the new bottleneck (2010s), he wrote about blockchain and decentralized architecture in Life After Google. When AI hype suggested intelligence itself was the new abundance (2020s), he pushed back with Gaming AI — arguing that creativity, not computation, is the truly scarce resource.

The thread through all of it is the same idea — follow the scarcity, understand what is becoming abundant and cheap, and bet on the architecture that exploits the abundance. That is Gilder’s method, and Telecosm was where he fully articulated it.

Why It Matters Now

Telecosm is twenty-six years old. But its framework is more relevant today than it was in 2000.

The AI boom is built on bandwidth. Every large language model, every cloud service, every video stream depends on the dumb network Gilder described. The hyperscale data centers that run the internet’s intelligence are connected by fiber backbones that follow the telecosm architecture. The “exaflood” of data that Metcalfe predicted in the 1990s is reality — global internet traffic runs in the zettabytes per year.

Gilder’s current focus on wafer-scale computing is the same framework applied to a different layer. The microchip is reaching its physical limits, and the next abundance, he argues, will come from wafer-scale systems that replace hundreds of chips with a single silicon wafer. Gilder identified this scarcity-abundance pattern back in 1994 in his essay “The Bandwidth Tidal Wave” — the defining scarcity of one era creates the defining abundance of the next.

The book that opened with “the computer age is over” was announcing a pivot, not an ending. The computer age ended because computers became so abundant and cheap that the action moved elsewhere. That is Gilder’s signature insight — the real money is in the thing that is still scarce, not the thing that is becoming abundant.

Telecosm was his most complete statement of that idea. Twenty-six years later, it is worth reading for anyone who wants to understand where the next pivot might land.