Thinking Machines Corporation was founded in 1983 to commercialize Danny Hillis’s MIT doctoral work on massively parallel computing. Its product, the Connection Machine, was the most glamorous hardware of the symbolic-AI era - a black cube studded with blinking red LEDs, designed to attack artificial intelligence and symbolic processing by spreading a problem across tens of thousands of tiny processors at once. The Computer History Museum’s catalog entry for a CM-1 records the original design: 65,536 one-bit processors, sixteen to a chip, wired together in a twelve-dimensional hypercube.
The machine was a cultural object as much as a computer. Richard Feynman worked on its router; designer Tamiko Thiel gave it its striking cubic form; and by 1993 the four fastest computers in the world were Connection Machines. The company also drew much of its revenue from a single source - DARPA’s Strategic Computing program - which paid for advanced supercomputers in the name of military AI.
That dependence was fatal. As the second AI winter set in and government funding for exotic parallel AI hardware dried up, Thinking Machines could not find enough commercial buyers for million-dollar machines. The architecture that made it special - massive, special-purpose parallelism for AI - also made it expensive and narrow. In August 1994 the company filed for Chapter 11 bankruptcy. Its hardware business was sold to Sun Microsystems, and the surviving software remnant drifted into data mining before being absorbed by Oracle in 1999.
Why business readers should care: Thinking Machines is the classic case of a company whose technology was admired, whose machines genuinely led the world, and which still failed - because it sold a costly, specialized product into a market propped up by one customer’s research budget. When the subsidy stopped, the brilliance did not save it.