Nvidia’s Jensen Huang Likes Talking About Chips, But Not AI

Many tech executives present as philosophers. The CEO of the most valuable company in the world just want make chips.

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:00):
Nvidia's Jensen Huang likes talking about chips but not AI
by Joshua Brustine read by Ramesh Metani. Last July, Meta
Platform's Inc. Chief executive Officer Mark Zuckerberg sat on stage
at a conference with Nvidia Korp CEO Jensen Huang, marveling

(00:21):
at the wonders of artificial intelligence. The current AI models
were so good, Zuckerberg said that even if they never
got any better, it take five years just to figure
out the best products to build with them. It's a
pretty wild time, he added, then talking over Huang as
he tried to get a question in, and it's all

(00:42):
you know, you kind of made this happen. Zuckerberg's compliment
called Huang off guard, and he took a second to
regain his composure, smiling bashfully and saying that CEOs can
use the little praise from time to time. He might
not have acted so surprised. After decades in the tra ventures,
Huang has suddenly become one of the most celebrated executives

(01:04):
in Silicon Valley. The current AI boom has been built
entirely on the graphics processing units that his company makes,
leaving Nvidia to reap the payoff from a long shot
bet Huang made far before the phrase large language model
LM meant anything to anyone. It only makes sense that
people like Zuckerberg, whose company is a major Nvidia customer,

(01:28):
would take the chance to flatter him in public. Modern
day Silicon Valley has helped cultivate the mythos of the
founder who puts a dent in the universe through a
combination of vision, ruthlessness, and sheer will. The sixty two
year old Huang, usually referred to simply as Jensen, has
joined the ranks. Two recent books, last December's The Nvidia

(01:50):
Way W. W. Norton, by Baron's writer and former Bloomberg
opinion columnist, Take Him and the Thinking Machine Viking April
a by the journalist Stephen Witt, tell the story of
Nvidia's rapid rise. In doing so, they tried to fill
out Huang's place alongside more prominent tech leaders such as

(02:11):
Steve jobs Elon Musk and Zuckerberg. Both authors have clearly
talked to many of the same people, and each book
hits the major points of Nvidia's corporate history and Huang's biography.
Huang was born in Taipei in nineteen sixty three, his
parents sent him and his brother to live with an
uncle in the US when Huang was ten. The brothers

(02:32):
went to boarding school in Kentucky, and Huang developed into
an accomplished competitive table tennis player and talented electrical engineer.
After graduating from Oregon State University, he landed a job
designing microchips in Silicon Valley. Huang was working at the
chip designer ELSI Logic when Chris Malachowski and Curtis Prim,

(02:54):
two engineers who worked at ELISI customer Sun Microsystems, suggested
it was time for all of them to found a
startup that would make graphics chips for consumer video games.
Huang ran the numbers and decided it was a plausible idea,
and the three men sealed a deal at a Denny's
and San Jose, officially starting Nvidia in nineteen ninety three.

(03:17):
Like many startups, Nvidia spent its early years bouncing between
near fatal crises. The company designed its first chip on
the assumption that developers would be willing to rewrite their
software to take advantage of its unique capabilities. Few developers did,
which meant that many games performed poorly on Nvidia chips,

(03:38):
including crucially the mega hit first person shooter Doom. Nvidia's
second chip didn't do so well either, and there were
several moments where collapse seemed imminent. That collapse never quite came,
allowing those early stumbles to be integrated into nvidia law.
They're now seen as a key reason the company's sped

(04:00):
up the pace at which it developed new products and
ingrained the efficient and hard charging culture that exists to
this day. The real turning point for Endvidia, though, was
Huang's decision to position its chips to reach beyond its
core consumers. Relatively early in his company's existence, Huang realized
that the same architecture that worked well for graphics processing

(04:22):
could have other users. He began pushing Nvidia to tailor
its physical chips to reduce those capabilities, while also building
software tools for scientists and non gaming applications in its
core gaming business, and Vidia faced intense competition, but it
had this new market basically to itself, mostly because the

(04:43):
market didn't exist. It was as if Wright's wit. Huang
was going to build a baseball diamond in a corn
field and wait for players to arrive. And Vidia was
a public company at this point, and many of its
customers and shareholders were irked by Huang's attitude to semiconductor design.
But Huang exerted substantial control of the company and stayed

(05:06):
the course, and eventually those new players arrived, bringing with
them a reward that surpassed what anyone could have reasonably
wished for. Without much prompting from Nvidia, the people who
were building the technology that would evolve into today's AI
models noticed that its GPUs were ideal for their purposes.

(05:27):
They began building their systems around Nvidia chips, first as
academics and then within commercial operations, with untold billions to spend.
By the time everyone else noticed what was going on,
and Vidio was so far ahead that it was too
late to do much about it. Gaming hardware now makes
up less than ten percent of the company's overall business.

(05:50):
Huang had done what basically every start up founder sets
out to do. He had made a long shot bet
on something no one else could see, and then carried
through on that vision with a combination of pathological self
confidence and feverish workaholism. That he done so with a
company already established in a different field only made the

(06:11):
feat that much more impressive. Both Kim and Wit are
open in their admiration for Huang as they seek to
explain his formula for success, even choosing some of the
same telling personal details. From Huang's affection for Clayton Christensen's
The Innovator's Dilemma, to his strategic temper to his attractive handwriting.
The takeaway from each book is that Huang is an

(06:33):
effective leader with significant personal charisma who has remained genuinely
popular with his employees even as he works them to
the bone. Still, their differing approaches are obvious from the
first paragraphs. Kim, who approaches Nvidia as a case study
in effective business leadership, starts with an extended metaphor in

(06:53):
which Huang's enthusiastic use of whiteboards explains his approach to management.
This tendency to kill him represents Huang's demand that his
employees approach problems from first principles and not get too
attached to any one idea at the whiteboard. He writes
later in the book, there's no place to hide, and

(07:13):
when you finish, no matter how brilliant your thoughts are,
you must always wipe them away and start anew This
rhapsodic attitude extends to more or less every aspect of
Huang's leadership. It has been well documented in these books
and elsewhere that Nvidia's internal culture tilts towards the brutal.
Kim Describeshuang's tendency to berate employees in front of audiences

(07:37):
instead of abuse, though this is interpreted as an act
of kindness, just Huang's way of, in his words, torturing
them into greatness. Kim does draw the line at Huang's
insistence on annihilating employees in ping pong, which he describes
as petty. The Thinking Machine, by contrast, begins by marveling
at the sheer unlikeliness of Nvidia's sob and rise. This

(08:01):
is the story of how niche vendor of video game
hardware became the most valuable company in the world, which
writes in its first sentence, As the technology and videos
enabling progresses, some obvious questions arise about the impact on
people outside the company. In large part, the story of
modern Silicon Valley has been about how companies respond to

(08:23):
such consequences. More than other industries, tech has earned a
reputation for seeing its work as more than simply commerce.
Venture capitalists regularly present as philosophers and start up founders
as not only building chatbots, but also developing plans for
how to implement universal basic income once that chatbots achieve

(08:46):
superhuman intelligence. The AI industry has always had a quasi
religious streak. It's not unheard of for employees to debate
whether their day jobs are an existential threat to the
human race. This is not Huang's or by extension and
vidious style. Technologists such as Elon Musk might see themselves

(09:08):
standing on Mars and then work backward from there, but
Wit Wrights Huang went in the opposite direction. He started
with the capabilities of the circuits sitting in front of him,
then projected forward as far as logic would allow. Huang
is certainly a step further removed from the public than
the men running the handful of other trillion dollar U

(09:31):
S tech companies, all of which make software applications directly
for consumers. WIT's book ends with the author attempting to
engage Huang on some of the headier issues surrounding artificial intelligence.
Huang first tells him that these are questions better posed
to someone like Musk, and then loses his temper before
shutting the conversation down completely. In contrast with other tech leaders,

(09:56):
many of whom were weaned on science fiction and draw
on it for inspiration, Huang is basically an engineer. It's
not only that he doesn't seem to believe that the
most alarmous scenarios about AI will come to pass. It's
that he doesn't think that he should have to discuss
it at all. That's someone else's job.