← Back to 52-Week Low Lens

GMO INTERNET INC

Company Overview

GMO Internet, Inc. is a newly reconfigured Japanese internet infrastructure operator. The legal entity used to be GMO AD Partners, an online advertising business. On January 1, 2025, GMO Internet Group transferred its domain, hosting/cloud, ISP, and related advertising/media operations into this listed shell, which was then renamed GMO Internet, Inc. That matters because many stock screens and English profiles still describe the old ad-only company. They are stale.

Data freshness. The latest clean official annual base is FY2025. More recent data is partial, delayed, unaudited, or estimated. The hard financial base below comes from the FY2025 annual securities report. The important post-year-end events are the March 2026 Turing investment and the April 2026 new-share issuance plus parent sell-down to repair Prime Market free-float non-compliance and fund GPU cloud expansion.

The growth headline looks spectacular because the listed entity changed shape. The real underlying picture is more modest: a sticky infrastructure business growing at low- to mid-single digits, plus a new GPU cloud business that may or may not become a good use of capital. FY2025 quarterly revenue still improved through the year, from roughly ¥19.1bn in Q1 to ¥20.2bn in Q4, while operating profit moved from about ¥1.63bn to ¥2.59bn. That is not a broken operating trend.

Owner earnings are meaningfully lower than the simple P/E implies. That is because this is not a pure software business. Servers, network gear, leased equipment, and systems upkeep are real costs. The April 2026 share issuance also means the current stock price should be judged on a larger share base than FY2025 historical EPS reflects.

How the Company Makes Money

The company now has two real segments. The first, and economically dominant one, is Internet Infrastructure: domains, cloud and rental servers, and internet connection services. The second is Online Advertising & Media, which includes agency work, ad-tech products, and owned media. The new GPU cloud business sits inside infrastructure.

Where do profits actually come from? Mostly from the boring parts: domain registration, hosting, rental servers, and ISP subscriptions. Management’s FY2025 sub-segment presentation showed roughly ¥20.7bn of revenue and ¥5.1bn of operating profit in domain and rental server, plus roughly ¥41.3bn of revenue and ¥7.9bn of operating profit in internet connection, before shared costs and new-business losses. That is the cash engine. Advertising/media is not.

This has historically been a good business because the products are embedded in customers’ daily operations. Moving a domain registrar, mail setup, hosting environment, or broadband provider is possible, but it is a nuisance. That creates switching friction. GMO also benefits from scale, a large installed contract base, brand recognition in Japan, and operational depth. The company says it manages more than 42m domains globally, ranking fourth by managed domains, and FY2025 recurring revenue was 84.9% of sales. Those are real quality markers.

Still, the moat is only moderate. Pricing is visible. Competition is real. Domains and hosting are not patent-protected monopolies. The advertising/media business has clearly weaker competitive protection, and GPU cloud is still an option rather than an established moat. So this is a durable franchise in its core, but not an untouchable compounder.

Why the Stock Fell

The stock has fallen from a post-reorganization peak around ¥3,675 in May 2025 to around ¥600 by the end of April 2026. That is an enormous decline. But the main story is not operating collapse. It is the unwinding of a stock-structure distortion.

After the January 2025 reorganization, the free float was tiny. Management had indicated the tradable share ratio was initially only around 1.44%, and even by December 2025 it was still only 7.7%, far below the Tokyo Stock Exchange Prime Market requirement of 35%. A stock with that little float can trade at absurd prices for reasons that have very little to do with normalized earning power. That is what happened.

Then the opposite happened. The market realized three things. First, the parent had to sell stock or the listing would remain non-compliant. Second, the company itself would likely need fresh capital for GPU cloud. Third, the “AI plus tiny float” story was not the same thing as a wide moat. In March 2026 the company entered the Prime Market improvement period. On April 20, 2026 it set a ¥710 issue price for 30m new shares, alongside a 61.5m-share parent sell-down. By end-April the stock was already trading below that price. The market is clearly skeptical about the returns on the new capital.

So the plain-English answer is this: the stock fell because an earlier scarcity premium disappeared, a very large supply overhang became real, and investors started treating GMO Internet as a capital-hungry infrastructure operator rather than a squeezeable AI proxy.

What the Market Is Assuming

(a) One-time / cyclical / sentiment-driven factors

• The May 2025 peak was a low-float squeeze, not sustainable intrinsic value.
• The parent’s sell-down was inevitable once Prime Market free-float non-compliance became public.
• The April 2026 primary issue diluted existing holders by about 10.9%.
• The stock is being repriced as a quasi re-IPO with normal float, not as a scarcity asset.

(b) Medium-term business headwinds

• GPU cloud needs heavy capex before investors have proof of durable returns.
• Advertising and media are under pressure as advertisers move work in-house.
• Integration of transferred businesses and overseas units can create execution drag.
• The company is funding growth while also maintaining a very aggressive payout stance.

(c) Potential long-term structural threats

• Parent control may keep capital allocation focused on group objectives rather than minority per-share returns.
• The advertising/media moat is weak and appears to be weakening further.
• Core infrastructure is sticky, but still exposed to price competition and commoditization.
• GPU cloud could become a low-return hardware arms race if scale advantages fail to emerge.

Temporary or Structural?

Reality check versus market narrative. The market is mixing a temporary stock-supply event with a smaller number of real business issues. Those should not be treated the same way.

Structural diagnosis. The essence damage is narrower than the stock chart suggests. The real structural problems are: first, the ad/media franchise is weaker than the market’s top-down “internet ad growth” narrative implies; second, governance and capital allocation deserve a discount because the company paid out essentially all FY2025 earnings while simultaneously issuing new equity for GPU expansion. The core infrastructure business itself still looks intact.

So the right label is mixed: TIME for the stock-price collapse, partial ESSENCE for ad/media economics and capital-allocation quality. That is not the same as saying the whole franchise is broken.

Is the Market Wrong? By How Much?

Time-as-a-moat test. If I had the company’s current market capitalization in cash, I could rebuild parts of this business. I could rebuild ad/media fairly quickly. I could likely assemble a GPU cloud offering if NVIDIA supply were available. I could not quickly recreate a 12m-plus contract base, domain-management scale, brand trust, billing systems, customer support, operating know-how, and distribution relationships in Japan.

That tells me the moat is real but moderate. Time helps GMO. Time does not make it unassailable. This is exactly the kind of business where a fair valuation matters a lot.

Intrinsic value. This is a FY2025-based valuation adjusted for the April 2026 capital raise. I use a post-issue share count of about 304.7m. I do not add the April 2026 cash proceeds as excess cash, because that money came from new shareholders and is earmarked for GPU capex and debt repayment.

The bridge is intentionally simple. Bear assumes large domain transactions fade, advertising stays weak, and GPU cloud earns only modest returns on a bigger capital base. Base assumes the recurring infrastructure core keeps growing modestly and GPU cloud becomes useful but not exceptional. Bull assumes GPU cloud turns into a genuinely profitable new leg without destroying returns on capital.

At roughly ¥183bn market cap, the stock implies an audited FY2025 earnings yield of only about 3.0% and a simplified owner-earnings yield of roughly 2.3%. For a business with parent-control risk, real capital intensity, and an unproven AI-capex program, that is thin. My base value is about ¥57bn below the current market cap, or roughly ¥188 per share lower. The market is only “cheap” if you underwrite something close to the bull case.

My judgment is therefore blunt: the market is right that the old low-float valuation was nonsense, and it is right to demand proof that GPU cloud will earn attractive returns. What the market may be getting slightly wrong is equating stock-supply pressure with a broken core franchise. But that is not enough, at today’s price, to create a margin of safety.

Key Facts, Estimates, and Judgments

Moat & mispricing score: 4/10. The moat is real in recurring infrastructure, weak in advertising, and still unproven in GPU cloud. The market is correct to erase the old scarcity premium, to penalize dilution, and to demand proof that new GPU capital will earn acceptable returns. What it may be underestimating is that the core infrastructure franchise itself did not break; FY2025 beat initial guidance and recurring metrics remained solid. But at roughly ¥183bn market cap, the stock already discounts something close to a successful GPU outcome, so this is not a classic time-over-essence bargain.

CoffeeAnd — 52-week low lens