March 22 2024

Full Stack Computing

AI is eating the world. Hearing this for the umpteenth time, the words “artificial intelligence” are fast becoming meaningless buzz words, or at best, the latest ploy for companies to instantly boost their enterprise value.

Yet, it is real, and it is fast becoming ubiquitous. In a nutshell, accelerated computing powers a right-in-time “spare intelligence”, which comes with the promise of giving a competitive edge to any player adopting it. Faster, smarter, in real time.

First, OpenAI’s ChatGPT shocked the world by its lightening-fast response time, and its ability to answer, decode, format, generate, render pretty much everything. It brutally invited itself as table stakes to our day-to-day, more than any Midjourney had done previously in the realm of images. Other AI models quickly followed to build on the hype. This triggered the gold rush.

As in any gold rush, picks and shovels are the surest bet. To that end, Jensen’s Nvidia is certainly the quintessential example. It has never been more apparent than in Nvidia’s keynote this week, also dubbed as the “Woodstock of AI”, gathering 11,000 afficionados.

From GPUs designer (the “chips”), Nvidia is rapidly integrating its proprietary software (CUDA), Cloud Services, as well as Applications (pre-trained AI models called NIMs). Nvidia is offering a full computing stack required for AI, from processors to hardware to software to any company that would rather prefer a packaged offer, supposing it readies hundreds of millions to do it. No wonder its share price ascent has become unstoppable at it reaches across infrastructure, and software. As its total addressable market expands, so does its price-earning ratio at 38x, closing in on Amazon’s 43x based on FY24 estimates.

Now, the battle for applications rages on. Microsoft recruiting Mustafa Suleyman, the founder of Inflection AI, this week to head a new consumer-AI division is only the latest sign of battle formation.

Who would ultimately find gold? Those industrial players who already sit on gold, are going to be in no rush. AI would probably be first implemented in the darkest corners of their company’s processes, in compliance or HR. More nimble smaller companies stand out to adopt it most, thus cutting through costs and lead times.

Use cases are plentiful, and would command large investments: Biotech/pharma > molecular research, Logistics > bottlenecks modelization, Automotive > driving assistance, Robotics > space & movement synchronization. Energy > new infrastructure to respond to higher AI’s energy intensity and cooling requirements.

Such acceleration of everything would certainly leave some in the dust. This would pit accelerationists against neo-luddites, and pose a rather philosophical question: does it work in favor of human progress? I leave you to ponder this one.

Price of gold – Bloomberg

The actual price of gold went from $20/ounce in the 1920s to $2170/ounce today. The first pioneers in 1848 certainly had in mind gold’s enduring value, both practical and symbolic. Little did they know that gold would now be a hedge against inflation, and systemic risk. Yet, they dug in search of a use case.

Stay safe out there !

About –

360 Advisory LLC is a Boston-based RIA managing investments