The escalating power demands of tomorrow’s data centers present a significant challenge, prompting exploration of diverse energy sources. Initiatives focusing on safer nuclear power, such as Microsoft and Constellation’s revival of Three Mile Island and the development of smaller, localized nuclear plants, are gaining momentum. Concurrently, research into renewable energy solutions, including ambitious projects involving satellite-based power generation and AI operations, are underway. These innovative approaches aim to harness alternative power sources to meet the burgeoning energy needs of the digital age. Addressing this challenge requires a multifaceted approach, drawing inspiration from historical precedents of overcoming similar technological hurdles.
Drawing parallels from the history of electrical engineering offers valuable insights into tackling the AI power conundrum. Deborah Douglas, Senior Director of Collections and Curator of Science and Technology at the MIT Museum, highlights the evolution of the power grid from disparate networks, mirroring the development of the decentralized internet. She underscores the ingenuity of early electrical engineers who devised tools and techniques long before the advent of modern electrical systems. Focusing on the 1940s, a period of significant societal and technological transformation, Douglas showcases the story of Phyllis Fox, a pioneering figure who navigated gender barriers and contributed significantly to the field of computing.
Phyllis Fox’s journey began with her master’s thesis at MIT in 1949, titled “The solution of power network problems on large-scale digital computers.” Her story exemplifies the struggle against gender bias prevalent at the time, while highlighting her determination to reshape the technological landscape. Douglas recounts Fox’s experience working as an engineering assistant at GE, where she was tasked with tedious manual calculations using a Marchant calculator. Fox’s ingenuity led her to discover a differential analyzer on campus, which she surreptitiously used during her lunch breaks to complete a full day’s work, showcasing her resourcefulness and drive for efficiency. This echoes the concept of the “wait calculation,” where delaying laborious manual tasks becomes advantageous with the eventual emergence of automating tools.
Fox’s resourcefulness paid off, granting her access to the differential analyzer and allowing her to contribute to the advancement of the field. However, her progress was temporarily halted by post-World War II societal norms, leading to her dismissal upon the return of male colleagues from the war. Undeterred, Fox pursued her master’s degree and subsequently found employment with Jay Forrester, a forward-thinking individual who embraced diversity in his hiring practices. Forrester’s inclusive approach fostered an environment where Fox’s talent flourished, contributing to the development of groundbreaking technologies.
The pioneering work of Fox and her colleagues involved massive machines like the Rockefeller differential analyzer, which were instrumental in solving complex differential equations. These analog computing behemoths, akin to large appliances in size, represent a significant milestone in computational history. Fox’s journey continued with her involvement in teaching at MIT, where she elucidated process calculations using flowcharts and block diagrams, laying the groundwork for modern computing methodologies. Her contributions extended to the Atomic Energy Commission’s Computer Center, where she developed Dynamo, an early computer simulation language. Furthering her impact on the field, Fox authored the first LISP manual and achieved tenure at the Newark College of Engineering (now the New Jersey Institute of Technology) in 1972, solidifying her legacy as a computing pioneer.
Phyllis Fox’s story serves as a potent reminder of the transformative power of individual ingenuity and the importance of embracing diversity in technological pursuits. Douglas concludes her presentation with the hope that individuals like Fox will emerge to address current technological bottlenecks, echoing the spirit of innovation that drove earlier breakthroughs. Reflecting on the history of analog computing and the evolution of tools like the differential analyzer inspires confidence in our ability to develop innovative solutions for the AI power challenge. While the projected power demands of data centers might seem daunting, history demonstrates our capacity to adapt and create new tools to meet evolving needs. Drawing inspiration from these historical examples and appreciating the ingenuity of past innovations provides a framework for addressing contemporary challenges.
The seemingly insurmountable power demands of future data centers, measured in gigawatts and terawatts, may appear less intimidating when viewed through the lens of historical ingenuity. The rapid evolution of technology suggests that our current perspective may be drastically different in just a decade or two. By embracing the lessons learned from pioneering figures like Phyllis Fox and appreciating the ingenuity embedded in now-obsolete machines, we can foster a spirit of innovation to overcome the power challenges of the future. Just as the differential analyzer revolutionized computing in its time, new tools and approaches will undoubtedly emerge to address the power needs of the AI era. The key lies in fostering an environment that encourages creativity, embraces diversity, and learns from the successes and challenges of the past.
