Remembering Bob Dennard, inventor of the chip that changed the world

When Robert “Bob” Dennard joined IBM in 1958, he didn’t know he was going to change the world. Then, the state of the art for random access memory (RAM) was solid core, a bulky piece of circuitry with paltry memory capacity. He knew something better was possible. In 1966, Dennard cemented his place in history when he and his colleagues at IBM Research invented dynamic random-access memory (DRAM), the chip that changed the world by shrinking memory storage while boosting its capacity.

Dennard passed away on April 23 at the age of 91. Decades later, DRAM remains central to computing, as IBM Senior Vice President and Director of Research Darío Gil remarked at a memorial gathering for the esteemed engineer at the Thomas J. Watson Research Center in Yorktown Heights, New York. The memorial on June 7 was a celebration of Dennard’s life and work, with several of his family members, along with prominent IBMers, like former Research director John Kelly, in attendance.

“Without DRAM, there is no modern semiconductor, and the world looks a lot different — a lot slower, a lot less connected,” Gil said. “Today, DRAM chips power more than 14 billion smartphones, 3.6 billion computers and just about every consumer electronic device.”

But Dennard, who retired from IBM in 2014, will also be remembered for his inventiveness, his mentorship, and his humility. He held his relationships with other IBMers at the same level as his professional achievements. In 1978, he told IBM Research that the earnest and enthusiastic recognition he received from his peers was as important as the award money he received from IBM for the DRAM discovery. “He laid the genetic code for what we consider the DNA of today’s IBM researcher,” said Gil. “He was brilliant. He was restless. He was passionate.”

Born in 1932 in Terrell, Texas, Dennard was educated in a one-room schoolhouse, excelling in math and science but showing little interest in engineering. “I was interested in climbing trees and shooting birds with my BB gun and taking the cat for a swim in the pond,” he once told IBM historians. At the urging of a guidance counselor, Dennard pursued an education in electrical engineering. He received his B.S. and M.S. from Southern Methodist University, then earned his doctorate from Carnegie Institute of Technology (now Carnegie Mellon) in 1958 and joined IBM Research that year.

Right around this time, metal-oxide-semiconductors were entering the scene. He and his colleagues were working on a replacement for solid core memory. They were taking advantage of the new possibilities offered by silicon-based microelectronics, but their first design was too slow and too complicated, and Dennard felt downcast after watching a presentation from a rival IBM team working on thin-film magnetic memory. He channeled that feeling into creativity.

“I went home that evening discouraged because their approach looked very simple compared to the complex six transistor memory cell which my team was using for each bit of data,” he said. “That evening I started exploring the possibility of storing data in a simpler way as a charge level on a capacitor. Within a few hours I had gotten the basic ideas for the creation of DRAM ironed out in my mind.”

His team soon unveiled a small memory cell that used a single transistor and single capacitor. Within a year, IBM and Dennard had been awarded the patent for the development. And in the 1970s, DRAM was supplying the memory needs for most home and office computers. The invention of DRAM encapsulated Dennard’s way of solving problems: Once he wrapped his head around an issue and envisioned a solution, the inventions themselves were obvious. And the results speak for themselves: Compared to a DRAM chip, solid core memory looks like an abacus.

Beyond the hardware, Dennard left a legacy in the form of the road map for ever more sophisticated transistors. If Moore’s Law described the size and number of transistors, Dennard’s theory of scaling described how they would work as they got smaller and more numerous, predicting that as each chip contained more transistors, the power used in a given area would remain constant.

Ever the optimist, Dennard was known to say: “Yes, there is an end to scaling. But there is no end to creativity.”