The Time Is Now To Get A GenAI Strategy

March 18, 2024 Timothy Prickett Morgan

I was born a little less than a year after IBM launched its venerable System/360 mainframe platform, arguably the first information technology platform ever created and sold at scale. I graduated from college six months after the AS/400 was launched and was the founding editor of this newsletter you are now reading a little more than a year after Silverlake came to market and showed what a real platform could look like.

I was present during the Unix revolution, the client/server revolution, the commercial Internet revolution, and the big data revolutions, and I was absolutely at the forefront of commercial supercomputing and the rise of the machine learning and then deep learning revolutions that have inevitably led to the Cambrian explosion of generative AI.

And I can tell you as a history buff in data processing and information technology and the money it takes and makes that I ain’t seen nothing like the business that Nvidia has built up since the first inklings of GPU compute in the middle 2000s. Frankly, I think even co-founder and chief executive officer Jensen Huang, who I have had extensive conversations with more than once, did not expect the rapid phase change that we are seeing with generative AI and the explosive effect it has had on Nvidia’s datacenter business and how nearly everything will be done in the future.

There is no question that Nvidia sought to increase the performance and efficiency of supercomputers when it started teaching its GPUs to do complex mathematical operations rather than just try to draw triangles and shade them with light and color and calculate ray tracings to make realistic 3D games. That is all that Huang and his Nvidia co-founders, Curtis Priem and Chris Malachowsky, were trying to do when they started Big Green.

But Huang saw the end of Moore’s Law coming and put together a team and a business plan to first accelerate traditional HPC modeling and simulation workloads, then relational databases and in-memory databases and analytics platforms (Spark being the most important one), and then finally machine learning, deep learning, and now the large language models that support generative AI. And this year in calendar 2024, unless something strange happens, Nvidia will be the fifth company in the datacenter to ever break the $100 billion revenue level, as I reported over at The Next Platform – my other day job – a month ago when Nvidia reported its financial results for its fiscal 2024 year ended in January.

Only three other IT suppliers that do business predominantly in the datacenter and to corporate clients have done this in the history of data processing and information technology. IBM was running at that $100 billion level from 2008 through 2012. The old Hewlett Packard, including its PC and printer businesses, did it in the late 2000s and the early 2010s. And Dell Technologies did it in its fiscal 2022 and 2023 years but is now declining a bit. Like IBM and HPE have done as they sold off consulting and outsourcing businesses, as did Dell, which also sold off server virtualization juggernaut VMware to try to pay down some of its debts.

Not only is Nvidia going to be big, but it is going to be profitable. Very profitable. About half of that money that it rakes in during 2024 will go straight to the bottom line, which means Nvidia will be sitting on a pile of cash $75 billion dollars high as it finishes this year.

And the commanding lead that Nvidia has in supporting generative AI, which is what is driving its business almost straight up to the sky, is going to be on full display this week at the GPU Technical Conference 2024 in San Jose, which my wife, Nicole, and I, who are co-founders of The Next Platform, would normally be attending but which we cannot do this year because of illness. (She is recovering, thank heavens, but it has been a challenging five months.) No matter, we will be covering GTC 2024 from afar, and watching with the rest of the world how the GenAI explosion will be further fueled and how it will change the world.

This may seem like a million miles away from application and database modernization for the 120,000 IBM i and OS/400 shops in the world, but it really is not. If you don’t have a GenAI strategy – and evidence suggests that only 18 percent of the active customers in the IBM i base are working on GenAI extensions to their applications, as we reported back in February – then you need to get one. This GenAI think is as big as any revolution we have seen or studied, and your business is under risk and stress even if you can’t feel it.

And this is not just an IT problem. This is a societal, cultural, and corporate issue. And given that, everyone has a say because it affects all of us. This is not an IT issue, and the technologists don’t get to make decisions about when and how to apply GenAI by themselves.

And to get some sort of grip on where GenAI is at, you could do worse than subscribing to The Next Platform and figuring out how to merge it with The Four Hundred.

RELATED STORIES

IBM i Shops Are Still Getting Their Generative AI Acts Together

Top Priorities in 2024: Security and AI

Remain Launches AI Chatbot to Assist with Development on RDi

How Many Programmers Does The US Economy Need?

It Is Time To Have A Group Chat About AI