MarketWise称多家AI数据中心公司筹备IPO，并提示可通过软银间接获得部分敞口

Key Points

In the past week or so, four AI data-center companies have filed for an initial public offering (“IPO”) or said they are exploring the possibility.

Data-center companies are looking to capitalize on massive investor demand for AI infrastructure, including the potential debut of Vantage in what could be the largest data-center company IPO ever.

Buying SoftBank shares could give investors exposure to two of these private data-center companies before they go public, though that exposure would be diluted across SoftBank’s broader portfolio.

A wave of artificial intelligence (“AI”) data-center companies is about to hit the market in the coming months, as companies look to cash in on massive investor interest. That surge may include the largest data-center company IPO ever – and investors can access it now through an already public stock.

AI infrastructure is one of the hottest ways to play the AI supertrend. These hard assets power AI models and are essential to the industry, so they may generate strong returns regardless of which model or company wins the AI race. So, investing in companies that operate AI data centers can be an attractive, “agnostic” way to invest in AI without having to pick which model company will win.

Here are four AI data-center companies that recently announced they’re planning to IPO or exploring the possibility and a few key details on each.

4 Upcoming Data-Center IPOs: Overview

Company Estimated capital raise Estimated IPO valuation Details
Vantage $10 billion $100 billion 35 data centers across five continents
DayOne $5 billion $20 billion More than 500 megawatts of data-center capacity operational or under construction in Asia
CyrusOne $5 billion Uncertain More than 60 data centers in the U.S., Europe, and Asia
Switch $10 billion $80 billion, including debt Data centers in at least four U.S. states
Source: Reuters, Bloomberg, SiliconAngle

Vantage is the key name here. If it goes public, it would likely be the largest data-center IPO ever. But none of the other potential IPOs are small fries either, with estimated valuations that range from $20 billion to $80 billion (CyrusOne’s is somewhere in the middle). Those valuations put these companies firmly in the mid-cap to large-cap categories.

Whether these firms are worth those valuations is another question, and we’ll need to see each company’s IPO prospectus before deciding. The devil is in the details, as usual.

Investors looking to get a slice of IPOs often turn to other publicly traded securities that already own a piece of them. That’s what investors could do by buying EchoStar (SATS), which owned a stake in SpaceX (SPCX), before its official debut in June .

In the case of these data-center companies, Switch is majority-owned by DigitalBridge (DBRG), according to Bloomberg. DigitalBridge is also a large investor in Vantage, the big fish here.

So, buy DigitalBridge and get exposure to these IPOs? It’s not quite that simple.

That’s because SoftBank (SFTBY) is on the verge of closing its acquisition of DigitalBridge, which was announced in December 2025. It’s an all-cash offer of $16 per share, and with DigitalBridge trading just below that price, there’s not much upside before or after the sale.

So, the best way to get exposure is to invest in SoftBank. But even then, it isn’t quite that simple. That’s because you also get exposure to the variety of other investments in SoftBank’s portfolio.

That said, you may be interested in some of SoftBank’s other investments, such as OpenAI. The Japanese company has invested $34.6 billion in the AI lab in its SoftBank Vision Fund 2 since September 2024. Overall, it has committed to investing $60 billion in OpenAI, of which $55 billion has already been placed. The investment would give SoftBank a 13% share in OpenAI, per CNBC.

The downside of investing in SoftBank to get access to these upcoming data-center company IPOs is that you’re diluting your investment in those data centers. That is, you won’t get the concentrated exposure to the data-center companies that you may want.

Risks to Watch for in AI Data-Center IPOs

While investing in AI infrastructure such as data centers may offer an agnostic play on who wins the AI race, like all investments, they’re not risk-free. Investors should watch for the following risks when considering whether to buy a new data-center company IPO.

Valuation

The owners of data centers are taking them public because they think they can realize attractive values while investors are still willing to pay up.

A look at the relative valuation of other data-center firms, as well as their historic valuations – say, looking at their enterprise value relative to earnings before interest, taxes, depreciation, and amortization (“EBITDA”) – can give you an idea of whether a stock’s valuation is high.

Indebtedness

As major real estate investors, these data-center firms will have substantial debt on their books. That’s just the nature of the beast. But a close look at the level of indebtedness (often measured as a multiple of EBITDA) can show whether the company is overly leveraged.

If you think AI is in a bubble or eventually will be, you’ll want to assess whether the company can sustain its debts if things go awry. Are the company’s debt maturities spread out over time, or does it have a bunch of maturities hitting in the next few years? Does it have high- or low-cost debt?

Customer Concentration

How much are these data-center companies exposed to any individual customer? It may be much higher than in many industries, given the massive investments by hyperscalers such as Microsoft (MSFT), Amazon (AMZN), Meta Platforms (META), and Alphabet (GOOGL). AI labs OpenAI and Anthropic have also been securing a lot of data-center capacity over the past year.

If a user – even a third-party user like an AI lab piggybacking on a hyperscaler – can’t pay its bills, the data center has a problem, especially if customer concentration and debt are too high.

Geographic Concentration

Having a larger geographic footprint may offer a data-center company more access to clients and opportunities to grow. It also reduces the risk of any one geography negatively impacting the firm. But geographic diversification also offers another defensive benefit, as AI data centers have experienced significant pushback in the U.S. and abroad for a host of reasons.

Even U.S.-focused companies may diversify their exposure across more AI-friendly areas.

AI IPOs Are Flooding the Market

Investors with a longer time horizon should be paying attention to the flood of AI-related IPOs hitting the market this year. This level of issuance often coincides with a market top, as insiders – those who know a company best – look to capitalize on investors’ “irrational exuberance.”

The fact that four data-center owners decided in virtually the same week to go public or at least explore the idea feels like “smart money” looking to time the market while it’s hot, rather than a coincidence.

This flood of AI data-center IPOs follows on the heels of the largest semiconductor IPO ever, when Cerebras Systems (CBRS) went public in May, and when the Elon Musk-helmed SpaceX priced the company at a stunning $1.77 trillion valuation in June.

The market is also keyed up for the IPO of AI lab Anthropic, which is trying to wrest the title of largest-ever IPO from SpaceX with a reported $2 trillion offering in October. It’s trying to go public quickly while the market remains on fire. ( Here’s how you can buy Anthropic before its IPO .)

This kind of huge IPO with lots of baked-in optimism signals the end of a bubble market.

But none of this is to say that short-term traders can’t make a profit, though they’ll need to be quite nimble to do so. SpaceX soared for a few days after its IPO, peaking at more than $225 per share, but has since plummeted on concerns about billions of shares coming to market .

Other AI trades, such as the memory chipmakers Micron Technology (MU) and SK Hynix (SKHY), have become more volatile following once-in-a-generation bull runs over the past year.

Again, not that traders can’t make money here – volatility is greater for this crew – but it’s becoming riskier. Meanwhile, long-term investors may see AI as a great opportunity even when all the hype dies down.

So, look closely at the next round of AI data-center companies to see what you’re getting in return for what is highly likely to be a steep IPO price. A truly great investment will still be great, even when the noise subsides.

Regards,

James Royal, PhD

Editor’s Note : Elon Musk just built the world’s most powerful AI data center. Called Colossus, it promises to make Grok the world’s #1 AI model in 2026 and beyond. But a new kind of AI model is about to explode on the scene, triggering a $100 trillion “reset” of the AI markets. Billion-dollar money manager Louis Navellier (our colleague over at InvestorPlace) calls it: “Artificial Superintelligence, but better.” And it could leapfrog ALL current AI models starting this year. For details, including the name and ticker of the company behind this breakthrough, go here .

The post AI Data-Center IPOs Are Set to Flood the Market. This Japanese Tech Stock Gets You Access Now appeared first on MarketWise .