PeopleofSuperintelligence

Lab leader

William Dally

Also known as Bill Dally

William Dally is a computer architect whose career links parallel-computing systems with the GPUs used for neural-network learning. He studied electrical engineering at Virginia Tech and Stanford, then earned a computer-science PhD at Caltech before building parallel machines at MIT and Stanford. At NVIDIA, he recognized that GPUs could perform neural-network image recognition faster with a fraction of the processors; he is now chief scientist and senior vice president of research. He coauthored a paper on a deep-learning inference accelerator and shared the 2025 Queen Elizabeth Prize for Engineering for contributions to modern machine learning. In 2026, he described using reinforcement learning to adapt chip cell libraries to new fabrication processes.

Works
486
Citations
51,667
h-index
91

Publication numbers from OpenAlex, as of 14 October 2026.

Photo of William Dally

Timeline

  1. 1970s–1983

    A computer in junior high

    As a junior-high student in the Washington, D.C., region, he encountered his first computer in the early 1970s. He studied electrical engineering at Virginia Tech and Stanford before entering Caltech's computer-science PhD program in 1983.

  2. 1983–1986

    Designing parallel hardware

    At Caltech, he completed a computer-science PhD and designed the MOSSIM Simulation Engine and Torus Routing chip. The chip pioneered wormhole routing and virtual-channel flow control.

  3. 1986–1997

    Building parallel machines

    His MIT group built the J-Machine and M-Machine, experimental parallel computers with low-overhead synchronization and communication. The J-Machine used wormhole routing to speed messages between nodes.

  4. 1997–2009

    A lab and a desert drive

    At Stanford, he chaired computer science while his team developed architecture, networking, signaling, routing, and synchronization technologies used in large parallel computers. During the 2005 DARPA Grand Challenge, a team collected road data to train a statistical machine-learning model.

  5. Since 2009

    GPUs for visual learning

    He joined NVIDIA as chief scientist in January 2009. After hearing about Andrew Ng's Google neural network for recognizing cats, he saw that NVIDIA GPUs could perform similar image-recognition work faster with a fraction of the processors.

  6. 2023–2025

    A prize and an inference chip

    He coauthored a 2023 paper on a 95.6-TOPS/W deep-learning inference accelerator. In 2025, he shared the Queen Elizabeth Prize for Engineering with six other engineers for contributions to modern machine learning.

  7. As of October 2026

    Leading NVIDIA research

    In April 2026, he described NVCell, a reinforcement-learning program for porting standard-cell libraries to a new process; the task runs overnight on one GPU. As of October 2026, he is NVIDIA's chief scientist and senior vice president of research.

Sources and credits

Photo: The White House, via Wikimedia Commons. License: Public domain. Source: commons.wikimedia.org. Cropped to a square and resized. Ask us to remove or replace this photo.

  1. William Dally | Research research.nvidia.com. Checked 14 October 2026.
  2. “This Is Going to Be Really Big”: How Nvidia’s Bill Dally Shaped the AI Era alumni.caltech.edu. Checked 14 October 2026.
  3. 2025 Queen Elizabeth Prize for Engineering awarded for Modern Machine Learning qeprize.org. Checked 14 October 2026.
  4. Not that we should be at all surprised, but Nvidia leans on AI pretty hard for speeding up how it plans and designs its next generation of GPUs pcgamer.com. Checked 14 October 2026.
  5. The engine behind the AI revolution: NVIDIA’s chief scientist William Dally on what really drives computing forward news.nus.edu.sg. Checked 14 October 2026.