At the Nvidia GPU Technology Conference in San Jose, California, on March 18, 2024, Jensen Huang, the co-founder and CEO of Nvidia, proudly revealed the company’s latest creation, the Blackwell GPU chip. The tech world is eagerly looking forward to Tuesday’s unveiling, when Huang will introduce Rubin, Nvidia’s latest AI graphics processor, at its annual GTC conference.
While many tech companies opt for alphanumeric codes that can seem cryptic, Nvidia has often charted its own course by naming its GPU architectures in honor of legendary women scientists. The next major AI platform will carry the name of Vera Rubin, an American astronomer known for groundbreaking work on dark matter.
Although Nvidia has not officially explained its naming choices or heavily spotlighted their diversity implications, the company’s consistent strategy illustrates a notable commitment to celebrating underrepresented scientists in the tech industry. This stands out even more as many DEI initiatives face reductions under the Trump administration.
Vera Rubin’s contributions to our understanding of dark matter—an elusive form of matter believed to account for a quarter of the universe—are significant. She was a staunch advocate for women in science throughout her career.
Since 1998, when Nvidia introduced its “Fahrenheit” microarchitecture, the tradition of naming architectures after renowned scientists has been an integral part of its identity. This cultural element once even inspired an exclusive employee t-shirt featuring cartoons of these famous figures.
As Nvidia ascended to become one of the top three most valuable tech companies globally, its nod to these scientists has attracted more attention. The firm is a crucial supplier for big names like Google, Microsoft, Amazon, OpenAI, Tesla, and Meta.
On Tuesday, investors are keen to learn more about the Rubin chip’s speed, configurations, and shipping timeline. Huang generally shares a succinct bio of the scientist behind the name before unveiling a new architecture. For instance, at the previous GTC conference, he mentioned David Blackwell, an eminent mathematician and game theorist, when introducing a major chip.
Naming the new chip platform Rubin aligns perfectly with Nvidia’s bid to cement its foothold as the leader in AI hardware. “Vera” will be the moniker for Nvidia’s upcoming central processor, while “Rubin” will denote the new GPU.
Vera Rubin, who was born in Philadelphia in 1928, made pivotal discoveries by studying deep space, often collaborating with other scientists to enhance telescopic technology. Her work in the late 60s, especially her observations of the Andromeda galaxy, significantly redefined the scientific comprehension of our universe. In the 1980s, she noted unexpectedly fast movements of stars on the galaxy’s fringes, which substantiated the existence of dark matter.
During her lifetime, Rubin published over 100 papers and achieved a trio of advanced degrees. Despite her accomplishments, she encountered challenges due to her gender, at times being restricted from collecting her own observational data or barred from certain observatories. In 2016, she passed away, and by 2019, the state-of-the-art Vera C. Rubin Observatory in Chile was named in her honor. This year, details about her advocacy were edited out of a biography on its official site.
Beyond Rubin, Huang previously named chips after other prominent figures. Nvidia’s “Blackwell” architecture paid tribute to David Blackwell, the first Black American inducted into the National Academy of Sciences. It succeeded the “Hopper” chips, named after Grace Hopper, a computing pioneer, and the “Ada Lovelace” architecture, reflecting the legacy of the 19th-century mathematician.
Originally, Nvidia’s scientist names played second fiddle to product names. However, last year, perceptions shifted as “Blackwell” started symbolizing an all-encompassing technology platform for advanced AI products.
For Nvidia, establishing the Rubin chip’s name recognition akin to Hopper and Blackwell is imperative. Following a record fiscal year where sales more than doubled to $124.62 billion, largely buoyed by Hopper and early Blackwell demand, Nvidia’s challenge is to sustain this trajectory. The company targets Vera chip releases by 2026, with expectations also centered around updated Blackwell Ultra chips arriving later this year.
Huang hinted at forthcoming technology, suggesting a future architecture inspired by another scientist. Melius Research’s Ben Reitzes foresees enthusiasm among partners ranging from Microsoft to sovereign entities driving investor confidence.
Tuesday’s keynote marks a critical evaluation of Nvidia’s acceleration in product releases, a strategy geared to maintain its status at the cutting edge of technology. With Blackwell entering the market last year, anticipation is high for what the Rubin line will bring.
