Nvidia (NVDA -0.01%) has been flirting with the $1 trillion valuation milestone, putting it in the rarefied company of Apple, Microsoft, Alphabet, and Amazon

All of these companies, Nvidia included, have at least one thing in common: They're platform businesses that generate revenue from enabling interactions and transactions between people and things, rather than simply selling a product. In other words, besides selling a product or service themselves, platforms facilitate the building of other businesses on top of their technology.

How do you build a platform business? One key ingredient is a tight integration between hard assets (in Nvidia's case semiconductors) and software. Nvidia's recent success thus exposes a major flaw in Advanced Micro Devices (AMD 2.26%): a lack of software to transform it from a peddler of chips into something much more. 

Given its shortcoming, is AMD stock still a buy for the long term?

How Nvidia went from seller of chips to enabler of AI

Generative artificial intelligence (AI) services like ChatGPT have gone viral, but AI has actually already been around for decades. To power an AI service, you have to start with powerful cutting-edge chips that can handle the demands of computing lots of data.

But to enable the step-up in capabilities that generative AI unlocks, something more than just developing a powerful chip was needed. Nvidia CEO Jensen Huang has long called AI a "full-stack problem." That means going beyond the chip, and the ultimate computing system itself, and also building a library of application frameworks (think of it as building blocks for software developers) and pre-built apps available via subscription. 

After well over a decade of work across hardware and an ever-expanding library of software, the public is suddenly aware of this massive "video game" company called Nvidia that has quietly extended its reach into all corners of the economy, from cloud computing services to healthcare development to modern vehicle design.

A chart showing the layers of Nvidia's business, starting with chips, software frameworks on top, and software application services above that.

Image source: Nvidia.

AMD's response will have limits ... for now

Nvidia currently has something of a monopoly on high-end chips used in training generative AI systems, but AMD makes advanced data center circuitry too. Its upcoming chip system, the Instinct MI300, has been teased for months and is set to make its debut this summer. Given all the hype around AI these days, and the outpouring of orders Nvidia has received for its latest systems, AMD is almost sure to score some victories in this department as well. 

However, AMD's Achilles' heel (and that of nearly all other chip companies right now) is a lack of software ecosystem like Nvidia has been building. 

Of course, all chip designers are also software developers, to a certain extent. A basic instruction set is needed so that these little pieces of silicon actually do something useful. But AMD's current tools lack the depth and specificity that enable researchers and developers from many fields to quickly deploy the hardware and start building their businesses on top of it. And AMD certainly lacks a growing library of subscription software like Nvidia has built. At this stage, Nvidia's end-to-end solutions are unique and have given it the ability to scale up into a true platform business. 

AMD is making strides in this direction. The acquisition of Xilinx in early 2022 was accompanied by the smaller designer's software development and AI tools. And upstart cloud processor company Pensando, which was purchased by AMD shortly after Xilinx, also touts an extensive software stack with its data center processors.

At the end of the day, though, AMD is still ultimately a seller of chips. Until it creates more of a platform -- complete with a deep software library and optional subscription services -- it will lack some of the ability Nvidia has to scale toward that $1 trillion milestone.

This doesn't mean AMD stock isn't a buy for the long term. I own shares of AMD too. But given their similar valuations at the moment (Nvidia trades for 51 times expected current year earnings, compared to 45 times for AMD) and Nvidia's greater ability to scale to a larger size, I'm not so sure AMD is the top chip stock to buy right at the moment.