Jensen Huang as CEO of NVIDIA: 33 years of unbeatable tech legacy
Jensen Huang transformed NVIDIA from gaming pioneer to AI powerhouse
From GeForce 256 to CUDA, Huang helped revolutionise modern computing
His extraordinary 33-year leadership leaves an unmatched technological legacy
NVIDIA has a special place in my heart. My very first Pentium III PC ran on a Riva TNT graphics card – it had 16MB of memory and needed an AGP slot on the motherboard. Stuff of ancient history now, obviously, but it makes me nostalgic. In many ways, it wouldn’t have been possible without the company’s outgoing CEO.
Jensen Huang didn't just build one of the most valuable companies ever. He helped change the fundamental architecture of computing, by not just co-inventing and coining the term GPU (with Chris Malachowsky and Curtis Priem) over three decades ago, but through NVIDIA he helped transform graphics processors designed for video games into the bedrock and engine powering the AI age we’re currently living through.
Before Jensen Huang the poster-boy of AI infrastructure, under whose stewardship NVIDIA became the highest valued company in the world with over $5 trillion market cap, I remember a Jen-Hsun Huang – a co-inventor who sparked a computing revolution and an entrepreneur who dared to dream. And someone’s name I routinely misspelled before realising it had conveniently changed to Jensen – when did that happen exactly? Never mind, that’s not the point.
His 33-year stewardship of NVIDIA as co-founder and CEO is one of the longest – longer than Jeff Bezos at Amazon or Reed Hastings at Netflix – and arguably the most consequential among modern companies anywhere in the world.
After starting NVIDIA in 1993, Huang made an error as CEO, betting on quadrilaterals as the atomic unit of computer graphics. It was a costly mistake, as NVIDIA’s early chips failed to make an impact and the company was facing financial disaster by 1995-1996 – before being rescued by SEGA, a famous story that Jensen Huang himself has recounted multiple times. What many don’t remember from that time, as the future of graphics processing lay in the balance, NVIDIA quietly pivoted to making triangles as the smallest unit of 3D graphics rendering. And the rest, as they say, was history.
The rewards of that pivot weren’t long in coming. In 1999, NVIDIA launched GeForce 256, the world’s first ever GPU. It made 3D graphics more lifelike and the CPU redundant for all things gaming. It was genuine progress in a big way, like a vision-impaired person seeing clearly for the very first time – that was the jump from the RIVA TNT/2 in my view.
Then came another groundbreaking moment, when NVIDIA silicon was announced to power the original Xbox in 2001, helping Microsoft’s entry into console gaming. And suddenly, out of nowhere, NVIDIA was now helping define how millions of gamers played worldwide.
But all this arguably pales in comparison to Huang’s most consequential gamble with CUDA, which was released to developers in 2007. More than the graphics hardware, CUDA is the serious software pedigree that’s NVIDIA’s crown jewels. And it began with an absurd idea 30 years ago – why waste all the GPU’s parallel processing for just drawing polygons on screen when it could help crunch scientific calculations as well? This was another of Huang’s generational bets at NVIDIA.
Five years later in 2012, AlexNet was born. Alex Krizhevsky, Ilya Sutskever (co-founder of OpenAI) and Geoffrey Hinton – the man regarded as The Godfather of AI – built a breakthrough neural network trained on two NVIDIA GeForce GTX 580 graphics cards to kickstart the deep-learning revolution that led to GenAI.
By 2017, NVIDIA’s Volta architecture introduced Tensor Cores to AI researchers, which had specialised circuitry tuned for the mathematics of neural networks. Coupled with transformer paper, the AI genie was well and truly out of the bottle at this point, and there was no looking back for NVIDIA as the key foundational block of the AI infrastructure.
Of course, in public and in front of the world, Jensen Huang was unmistakable in his black leather jacket – just like Steve Jobs made the black turtleneck his own. Whenever he graced the keynote stage, Huang had an almost mischievous delight in showing off NVIDIA’s tech wizardry. While there was theatre to his presentations, Huang could easily disappear into the architectural nuances of Hopper, Blackwell or Vera Rubin without losing the room. He seemed happiest when explaining why the next NVIDIA chip mattered.
Behind the rockstar persona was a famously demanding manager, as people have reported over the years. Jensen Huang famously had 60 direct reports at NVIDIA and avoided one-on-one meetings to avoid miscommunication. Not exactly Management 101 material, if you think about it. Nor was his belief that hardship makes people stronger, shaped by his childhood experience of cleaning bathrooms in boarding school.
In many ways, Huang’s story is larger than NVIDIA because his influence escaped graphics cards into geopolitics, from glass-door boardrooms to grass-roots agitations against AI datacentres in a way we have never seen before. The GPUs I once begged my parents to buy became essential tools for researchers studying diseases or scientists simulating the physical world and engineers building autonomous machines to LLMs. And before we knew it, they transformed and became the heartbeat of all things GenAI.
That’s why Jensen Huang’s departure feels different from an ordinary CEO leaving a company. Huang’s resignation as CEO of NVIDIA is understandable given his admission of infidelity and the serious allegations against him. While NVIDIA’s engineers, partners and customers will carry his work forward, I doubt they will ever truly replace Jensen Huang and what he means for the industry.
The next permanent CEO of NVIDIA has big shoes to fill, not unlike Sir Alex Ferguson's after his reign at Manchester United – and we know how that story's unfolding. NVIDIA shareholders would no doubt wish for a Tim Cook taking over at Apple after Steve Jobs. Only time will tell, of course. But for someone who remembers that little 16MB RIVA TNT card inside a Pentium III while marvelling at AI factories and the DGX and RTX Spark, it’s difficult not to marvel at how far the journey has travelled – and the man at the centre of it all.