At the end of May, I noted five high-flying artificial intelligence stocks that seemed primed for a stock split.Bangalore Stock Exchange

Sure enough, in a span of weeks, three of those five went ahead and announced splits.

That of course leaves two others with sky-high stock prices and continued AI momentum. Here's why one of them, Super Micro Computer , may be the next to split.Jaipur Stock

It wasn't long ago that Supermicro was a relatively unknown stock, with a market cap that relegated it to the small-cap Russell 2000 index. However, with a massive 246.2% gain in 2023 and then another 188.2% gain in the first half of 2024, Supermicro elbowed its way into the S&P 500 index in the first half, making it the biggest gainer in the index, even outpacing NvidiaPune Investment. In July, the accolades kept rolling in, as it was just announced Supermicro will be admitted to the Nasdaq-100 index on July 22.

After the tremendous run in the stock, Supermicro touched a high of $1,229 back in February, before pulling back to $879 per share as of this writing.

Could a stock split be coming? While stock splits don't change the value of a company at all, they do usually mark a vote of confidence on the part of management. Supermicro's stock is definitely in the territory where other companies recently split their stock. But how does management plan to maintain and grow this valuation after its tremendous run?

Obviously, Supermicro caught lightning in a bottle with AI, as its building-block server architecture, fast time to market, and power-efficient designs were perfectly suited for the AI era. As such, Supermicro has seen its revenues and earnings explode over the past two years, as it has grabbed far more share of the AI server market than it had of the traditional server market.

But after the skyrocketing growth over the past two years, is there really opportunity for further growth? Supermicro now trades at 48.8 times trailing earnings and 24.8 times its 2025 earnings estimates.

Supermicro has generally followed the trajectory of Nvidia, which at some level makes sense. After all, Supermicro's AI servers include Nvidia's chips, which it then marks up between 14% and 17% -- Supermicro's gross margin target -- for the value-add that Supermicro provides with its server architecture.

So, one leg of growth will have to do with the AI accelerator market, which is mostly Nvidia but will also include Nvidia's competitors as they ramp their competitive offerings.

While some have cautioned that AI investment may soon peter out, given the strength of the current boom, others are more optimistic. It's true that the most enthusiastic commentators are semiconductor executives perhaps "talking their book," such as Advanced Micro Devices CEO Lisa Su, who forecasts a $400 billion AI chip market by 2027 and Taiwan Semiconductor Manufacturing's management, which forecasts a 50% average growth rate for AI chips through that time. But these executives have a strong track record.

Supermicro does seem confident in the market's growth. Not only will the company bring its Malaysia factory online very soon, but management just announced it would be building three new manufacturing facilities right in the heart of Silicon Valley, which will soon double the company's capacity for liquid-cooled "supercluster" capacity.

One thing critics have pointed to as a limiting factor on AI growth is the tremendous amount of electricity required to power it all. To that end, Supermicro CEO Charles Liang thinks the near future of AI supercomputing will depend on liquid cooling technology.

To date, liquid cooling has only garnered about 1% of the data center market, according to Liang. This has traditionally been due to the difficulty of managing liquid cooling systems in data centers, plus the extra cost.

However, now that AI chips require so much electricity, and liquid-cooled racks greatly reduce the amount of heat generated and the need for large air conditioning systems, liquid cooling now makes much more sense. According to Liang, a liquid-cooled data center can cut operational expenses for AI data center operators by 40%.

And Supermicro has been preparing for this moment, having invested in liquid-cooling technology over the recent past. Last quarter, the company's inventory greatly increased, which Liang noted was due to accumulating parts for a big liquid-cooled push in the back half of the year.

As Supermicro is just on the cusp of bringing liquid-cooled racks to the server market, that should bode well for growth and margins. The more differentiated its offerings and the greater the operational costs it can cut for customers, the more it will be able to maintain margins on typically high-dollar but low-margin AI servers.Nagpur Investment