At the just-concluded GTC conference, Huang Renxun announced Nvidia's GPU roadmap for the future AI factory. From Ampere, Hopper, to Blackwell, Rubin and Feynman, each generation of GPU architecture is named after outstanding scientists in history. Behind these names not only embody the pinnacle achievements of human scientific wisdom, but also embodies Nvidia's tribute and inheritance of scientific and technological innovation.

At the GTC conference that ended last week, Huang spent more than two hours introducing us to Nvidia's progress over the past year.

Among them, the most concerned thing is the "one update every year" GPU chip architecture roadmap for future AI factories.

In 2024, Blackwell GPU will be released;

In the first half of 2025, Blackwell GPUs have been fully shipped;

In the second half of 2025, the Blackwell Ultra will be launched;

Vera Rubin will be released in the second half of 2026;

In the second half of 2027, the Rubin Ultra will be launched;

In the future, there is Feynman, which will be launched in 2028.

Have you ever wondered why these AI chip architectures are named so?

In fact, Nvidia's chip name comes from famous mathematicians and scientists in history.

Nvidia gave these scientific masters the highest tribute and was inspired by them.

These GPU architectures are inspired by history, and history has also given AI chips a soul.

Ampere. Hopper. Blackwell. Rubin. Feynman

Prior to Blackwell and Rubin, the Nvidia GPU architecture was named after Hopper and Ampere.

The "graphics cards" we are familiar with are named after, such as A100, H100, and H800. The numbers behind A and H represent the performance level.

André-Marie Ampère

André-Marie Ampère (French: André-Marie Ampère, 1775-1836) is a French physicist and mathematician, and one of the founders of classical electromagnetics.

The unit of current "ampere" is named after his last name.

André-Marie Ampère was born on January 20, 1775 in Lyon, France's second largest city, where both parents had been engaged in the silk trade for generations.

His most comprehensive scientific research is in the field of electricity. In 1820, Ampere developed electrodynamics by applying infinitesimal computing to electricity. The right-hand rule we are most familiar with is also named after Ampere.

Amper died on June 10, 1836. His body was moved to Montmartre Cemetery in 1869. His epitaph reads: "Tandem felix" (Finally happy).

The Nvidia Ampere architecture was launched at the 2020 GTC Conference.

Grace Hopper

Grace Murray Hopper (1906-1992) is a computer scientist and mathematician. He is one of the earliest programmers in the world and one of the earliest female programmers.

She was involved in the work of Universal Automatic Computer (UNIVAC), one of the earliest all-electronic digital computers. She was the first full-time programmer on Harvard Type 1 computers, creating the first modern compiler system, and the first advanced commercial computer programming language COBOL, known as the "Mother of COBOL".

She received her degree in mathematics from Vasa College, where she taught, and also received her master’s and doctorate in mathematics from Yale University. In 1943, she joined the Women’s Volunteer Emergency Services and was eventually promoted to Rear Admiral. She also predicts that computers will one day become widely used devices, just like today.

It is said that the word "Bug" to describe computer problems and faults was also proposed by her.

In 1973, Hopper was named a Distinguished Member by the British Computer Society, making her the first woman to receive the title. In 2016, she was posthumously awarded the Presidential Medal of Freedom.

Hopper died in 1992 at the age of 85.

Nvidia's Hopper chips drive the generative AI revolution in the ChatGPT era, costing about $40,000 per chip, quickly becoming a hot commodity for big tech giants and AI startups.

David Blackwell

David Harold Blackwell (1919-2010), an American statistician, emeritus professor of statistics at the University of California, Berkeley, and one of the proposers of the Rao-Blackwell theorem.

Blackwell is the first black academician of the National Academy of Sciences and the first black tenured faculty member at the University of California, Berkeley.

Blackwell is a mathematician and statistician who has made significant contributions to the fields of game theory, information theory and probability theory.

He attended the University of Illinois at Urbana-Champaign at the age of 16. One of his most famous contributions in the field is the Rao-Blackwell theorem used to improve estimators.

The Rao-Blackwell theorem is the cornerstone of statistical estimation theory, which provides a systematic approach to augmenting estimator by using sufficient statistical data.

Blackwell died in 2010 at the age of 91.

Nvidia's Blackwell chip is by far the most advanced chip. Nvidia is preparing the next generation of Blackwell Ultra chips.

Vera Rubin

Vera Rubin (1928-2016), an astronomer whose most famous work is to provide strong evidence of the existence of dark matter.

She received her bachelor’s degree in astronomy from Vassar College, her master’s degree from Cornell University, and her doctorate from Georgetown University. She studied many galaxies and their rotation rates.

Her work has been recognized for awards including the National Medal of Science and the Royal Astronomical Society Gold Medal.

She died in 2016 at the age of 88.

Nvidia's upcoming Rubin AI "Super Chip" platform is expected to debut in the second half of 2026.

Richard Feynman

Richard Phillips Feynman (1918-1988), an American theoretical physicist, won the Nobel Prize in Physics in 1965 for his contribution to quantum electrodynamics.

Feynman received his undergraduate degree from MIT and his PhD from Princeton University. He created the Feynman diagram, a graphical representation that helps calculate the probability of particle interactions.

He was recruited in 1941 for the Manhattan Project, the United States’ atomic bomb project, and then worked in the secret laboratory in Los Alamos, New Mexico.

Feynman later became a member of the committee investigating the explosion of the space shuttle Challenger.

He died in 1988 at the age of 69.

Nvidia's Feynman architecture is a GPU family that has not yet been fully announced.

Lovelace.Kepler.Truing.Lovelace

It can be said that the above "Ampere.Hopper. Blackwell. Rubin.Feynman" architecture is a product before and after the explosion of generative AI.

Previously, Nvidia's architectural name continued the tradition of naming scientists, such as Kepler, Turing and Lovelace.

Johannes Kepler

Johannes Kepler (German: Johannes Kepler, 1571-1630), a German astronomer and mathematician.

Kepler was a key figure in the scientific revolution in the seventeenth century.

His best-known achievements were Kepler's Law, which inspired Newton to come up with the law of universal gravitation.

Kepler GPUs are suitable for GeForce (Consumer Graphics), Quadro (Business/Professional Graphics) and Tesla (Computer/Enterprise/Data Center) series GPUs.

Alan Turing

Alan Mathison Turing (1912-1954).

British computer scientist, mathematician, logician, cryptographic analyst and theoretical biologist, he is known as the father of computer science and artificial intelligence.

Turing's greatness needs no more words.

The famous Turing machine model proposed by Turing laid the foundation for the logical working of modern computers.

Nvidia Turing architecture has made breakthroughs in computer graphics, enabling real-time ray tracing technology in consumer-grade products for the first time, thanks to its dedicated RT Cores (Ray tracing core).

Ada Lovelace

Augusta Ada King, Countess of Lovelace (1815-1852), whose original surname was Byron, was the daughter of an English mathematician and writer, the famous poet Lord Byron and Anne Bella Milbank Byron.

She was the first person to advocate that computers can not only be used for calculation, but also published the first paragraph of algorithms used for analyzing machines.

Eda Loveries is recognized as the first programmer in history, the "mother of programming".

What she is best known for is the translation and annotation of her colleague Charles Babbage’s analytical machine.

An early programming language was also named after her, and the second Tuesday of October is designated Adda Loveries Day in honor of women in the STEM field.

She died in 1852 at the age of 36.

Nvidia's Lovelace GPU architecture supports its 40-series graphics cards.

Inspiration, inheritance and tribute

Nvidia's choice of these names seems to be well thought out, and there is a potential thematic connection between these scientists' research areas and key innovations in their respective architectures.

Perhaps the inspiration behind every upgrade of GPU technology comes from these scientific masters.

Hopper predicted that computers would one day become the most widely used device, and Nvidia Hopper pioneered the era of generative AI;

Blackwell's theory has become the cornerstone of statistical estimation theory, and the Blackwell architecture has made efforts in data-intensive AI and complex systems;

Rubin's research provides evidence for the existence of dark matter, and as the "super chip" of Nvidia's next generation, the Rubin architecture will be committed to pushing the boundaries of computing power.

Nvidia’s naming is not only a respect for scientific history, but also an incentive to connect the company’s technological progress with the legacy of human intelligence.