Just a few months ago, I floated the idea that small cloud-computing infrastructure company DigitalOcean (DOCN 3.30%) could make a move with artificial intelligence (AI). For years, developers and small businesses have been asking if the company would add graphics processing units (GPUs) -- the chips favored for AI algorithm training -- to its ecosystem.  

The time has finally arrived. DigitalOcean just announced it is indeed giving its growing user base what it wants: GPUs! But this is no organic addition to the existing global DigitalOcean data center platform.

The company said it acquired start-up Paperspace to accelerate its jump into the AI era and keep pace with data center giants like Amazon's (NASDAQ: AMZN) AWS, Microsoft (NASDAQ: MSFT) Azure, Alphabet's (NASDAQ: GOOGL)(NASDAQ: GOOG) Google Cloud, and Oracle (NYSE: ORCL) Cloud. 

I like DigitalOcean's latest move on the chessboard.  

Why buy new when you can buy slightly used?

AI, and specifically generative AI like the viral app ChatGPT, are all the rage right now. The promise of turning simple natural-language text prompts into multi-paragraph prose, pictures, video, and even software code holds obvious and tremendous appeal. 

But training this new breed of generative AI algorithm doesn't come cheap. It requires massive amounts of data, and lots of expensive GPUs -- especially those being sold by Nvidia (NVDA 6.18%). Its latest-and-greatest H100 GPU systems for data centers reportedly can cost as much as $40,000 a pop!  

These monsters are the domain of the cloud titans, though, as they slug it out for big customer acquisition and retention (holding on to existing customers, or getting them to increase spending) by racing to be the first to offer Nvidia's semiconductor AI ammunition.

Not all AI tasks and other high-performance computing require that kind of firepower, though. Indeed, DigitalOcean has made plenty of hay over the years offering very affordable cloud infrastructure for tinkerers, entrepreneurs, and small businesses. Slugging it out with the tech giants simply wouldn't make sense. 

Which is why acquiring Paperspace makes a lot of sense. Paperspace already offers numerous Nvidia GPUs via its three data centers in California, New York, and the Netherlands (via co-location, when data center space is rented from, for example, a big real estate developer like Equinix).

Paperspace exclusively offers Nvidia GPU configurations as powerful as the A100, the previous flagship before the H100's commercial release in late 2022.

DigitalOcean said it got its hands on Paperspace for $111 million in cash.  

Does this purchase even make sense?

Here's a potential rub, though: DigitalOcean says that Paperspace won't have any material impact on its own revenue for 2023. Given that DigitalOcean itself still isn't all that big (with expected revenue of up to $720 million this year), did the company just blow a significant amount of cash on an AI pipe dream?

After all, DigitalOcean reported having just $613 million in cash and short-term investments, offset by debt of nearly $1.5 billion, at the end of March 2023.

The reason I like the Paperspace acquisition anyway is that, first and foremost, DigitalOcean generates positive free cash flow. It doesn't need that cash stockpile to fund its current growth initiatives, and it can fund its current stock repurchase plan (a return of cash to shareholders) with that free-cash-flow generation.  

And second, Paperspace looks like a good use of cash to promote future growth. The start-up's business model isn't so different from DigitalOcean's, but has a far more narrow focus on AI and machine learning. By contrast, DigitalOcean is more broadly focused on helping small businesses solve their future computing needs, and now has some specialized cloud-based Nvidia hardware to offer them, too. Next year could see a ramp-up in DigitalOcean's revenue growth as a result if the current AI hype cycle matures and comes to the small business world, too.  

Only time will tell what kind of return on investment DigitalOcean will get on that $111 million for Paperspace, but AI is a big market that's already been around for decades. Helping underserved small businesses put AI to good use sounds like a great bet to me, and a great way to get a bunch of AI infrastructure up and running quickly. 

DigitalOcean stock trades for a meager 21 times analysts' expected 2024 earnings per share and free cash flow. I believe this company just became a very attractively priced AI stock, especially compared to its heavy-hitting peers in the cloud infrastructure space.