George Gilder’s “The Information Factories” is a 2006 Wired feature about data centers becoming a new kind of computer: vast networks of processors, storage and memory working together over high-capacity networks. It is best read as a historical argument about the rise of cloud computing—not as a description of today’s infrastructure or a guide to choosing a cloud provider.
What “The Information Factories” means
In the feature, Gilder uses “information factories” to describe large data centers as computing platforms, not simply buildings that hold files. They combine many computers, storage devices and memory systems, coordinating work in parallel and connecting it through fast networks.
The industrial-factory analogy is about scale and coordination. Instead of relying mainly on an individual desktop computer to store information and run applications, companies could draw on centralized facilities to process growing volumes of data and deliver services over the Internet. Gilder presents search and indexing as early uses, with broader online applications as a possible next step.
Eric Schmidt, then Google’s CEO, described the shift in an email quoted in Wired: “In this architecture, the data is mostly resident on servers ‘somewhere on the Internet’ and the application runs on both the ‘cloud servers’ and the user’s browser.” Schmidt’s formulation emphasizes that the user’s computer did not have to disappear; work and data could be divided between it and remote servers. He also called the consequence “the return of massive data centers.”
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall#1 Best Overall
The accessible full text is a Google Groups repost dated October 12, 2006, reproducing Wired’s October 2006 feature and linking to the original archive page. The original Wired page could not be retrieved for this account, so quotations and claims below are attributed to Wired’s 2006 feature as reproduced in that repost: Google Groups repost.
How the feature says the model could work
Parallel computing from commodity hardware
Gilder’s account treats a large collection of relatively ordinary computers as a system whose capacity comes from the number of machines and their coordination. Tasks can be distributed across processors, while storage and memory provide access to the information those tasks need. The network connects the pieces and lets users reach services without keeping all data and processing on a personal computer.
Rank #2
- Durable hardcover with concealed wire-o binding
- Archival, acid-free paper helps preserve your information.
Scale depends on more than servers
The strategic advantage, in Gilder’s argument, belongs to operators able to coordinate several linked resources: processing, storage, memory, bandwidth, electricity and site location. A facility is useful at scale only when those elements work together. The feature also argues that abundant storage and bandwidth could be spent to save users time and attention—for example, by making more information searchable or more services available online.
What the 2006 figures do—and do not—show
Wired’s article included striking estimates of Google’s infrastructure, but Gilder explicitly characterized its server, storage, memory and traffic figures as “educated guesses.” They are historical estimates reported in the 2006 feature, not verified measurements or current Google specifications.
| Figure reported in the feature | Qualification |
|---|---|
| 200 petabytes of hard-disk storage and four petabytes of RAM | Gilder’s 2006 estimates for Google; described as educated guesses. |
| 450,000 servers | The feature calls this the lowest estimate available to its author in 2006. |
| 100 million queries a day | A figure used by Gilder in 2006 to estimate system input-output bandwidth. |
| Five gigawatts of electricity for major search engines | A 2006 estimate developed in the feature from assumptions about servers, disks, cooling and power conversion; it is not an independently validated measurement. |
| 100-megabyte disk drive for $500 in 1991; 750-gigabyte drive for $500 in 2006 | A cost-and-capacity comparison reported by the feature for those years. |
| 50-megahertz Intel 486 processor for about $500 in 1991; 3-gigahertz processor for $500 in 2006 | A processor comparison reported by the feature for those years. |
The comparisons illustrate why Gilder thought large-scale computing was becoming practical: storage and processor capacity available for a given price had increased substantially over the period he discusses. They do not establish current prices, present-day data-center size or present-day energy use.
The constraints and the counterargument
The feature does not portray data centers as an inevitable or cost-free destination. It raises electricity demand, cooling and the complexity of scaling large systems as constraints. Its estimates and projections reflect the article’s assumptions and the statements of industry figures it quotes; they should not be treated as independently verified forecasts.
Gilder also considers the possibility that faster chips and optical networks could shift more computing back toward the network edge—the devices and systems closer to users. The feature’s conceptual contrast is therefore not simply “old desktop versus new cloud.” It is a question of where processing and data reside, how much the network must carry, how easily capacity can scale, and where power and cooling are required.
Eric Schmidt, recounting a prediction from his time at Sun Microsystems, said: “When the network becomes as fast as the processor, the computer hollows out and spreads across the network.” The feature also quotes Andy Kessler, described there as a former Bell Labs engineer turned investor, arguing that creativity, customer demand and commercial opportunity remain at the edge. Those statements frame a tension in the article: centralized facilities can coordinate enormous resources, while innovation and user needs may keep drawing computation outward.
Free tools Windows power users keep installed
One-click scans. No signup required.
How to read the article today
- As a historical account: Its named companies, infrastructure estimates and operating assumptions belong to 2006.
- As a conceptual argument: It explains why data centers could be understood as computers made from networks of machines, rather than as passive storage buildings.
- As a forecast: Its expectations about energy, technical scaling and a possible return to edge computing are predictions and questions raised at the time—not proof of what later happened.
- Not as a buying guide: The feature offers no current provider rankings, service comparison or basis for recommending a cloud company.
The original feature opens by asking how society will “feed,” tame and eventually outgrow this emerging large-scale computer. That framing captures its central concern: the cloud’s promise depends not only on adding machines, but also on managing the physical resources and network relationships that make them function as one system.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
