menu Shuts Down Sept 30: Not One Number in the Announcement, and a Market That Peaked in 2023

On August 19, 2026, KDDI announced that it will end its food delivery service “menu” on September 30, and that the operating company, menu Inc., will be dissolved and liquidated. The service is still running. The shutdown is a month and a half away.

Filing this under “another one gone” means missing the strangest part. In the same release that announces the wind-down, KDDI writes that on August 31 it will buy up all the remaining shares and make menu a wholly owned subsidiary. Buy it, then close it. And the reason is stated outright.

This publication records the businesses of individuals and small teams in numbers. menu is a capital-backed company and falls outside our coverage criteria. We are covering it anyway, because this exit shows, about as clearly as anything could, the gap between businesses that publish numbers and businesses that do not.

An acquisition whose purpose is liquidation, not operation

KDDI’s release states that it will acquire all of the menu shares held by Rayson Holdings, make the company a wholly owned subsidiary, and then proceed with the process. The stated purposes are handling the various matters that come with ending the service (for users, partner restaurants and delivery couriers) and carrying out the dissolution and liquidation procedures smoothly. Read plainly, the purpose of the acquisition is not to obtain a business but to hold the execution structure for the cleanup on its own.

The sequence goes like this: announcement on August 19, completion of the share purchase on August 31, service shutdown on September 30, then dissolution and liquidation. Rather than consulting a partner holding 49.4% while working through a partner-restaurant base numbering in the tens of thousands and a liquidation, buying them out and consolidating decision-making is faster. M&A is normally done in order to keep a business running. Here it is being done in order to end one.

The language of the shutdown also differs depending on who is speaking. KDDI explains it as “a comprehensive review of the changing competitive environment surrounding the food delivery market and the outlook for the business, from the standpoint of optimally allocating management resources.” menu itself, speaking to ITmedia, said it had “judged that providing the service sustainably going forward would be difficult.” The first is portfolio language. The second is the language of the party involved. The same event, and the difference in position shows up directly as a difference in vocabulary.

The peak was 2023, and the “recovery” after it was bought with discounts

The coarse reading, a market that grew on pandemic demand and shrank after it, is not accurate. The delivery market size published annually by Circana Japan (formerly NPD Japan) moves like this.

The peak was not 2021, the height of the pandemic, but 2023, after movement restrictions were lifted. Then 2024 fell 7.6%. Circana attributes the decline to belt-tightening and the return to commuting, which cut the number of ordering occasions themselves.

The question is what is inside the “+2.0%” of 2025. Circana writes that the background to this recovery is that, in response to the 2024 decline, measures such as offering menu items at the same price as in-store and waiving delivery fees increased, and competition intensified. That is, the market did not return on its own. Operators bought orders back by shaving unit prices and fees. Even where the market size in yen looks flat, the amount left to the operator per order is thinner. In a structure like this, the smaller players run out of breath first.

(There is an inconsistency between the 2024 estimate and the 2025 growth rate, suggesting a later revision. Here we simply lay out each year’s published figure and growth rate as they were released.)

Look at the results of the side that survived

Nor is it a case of “the ones who exited lost and the ones who stayed won.” Demae-can, one of the remaining major domestic players, saw its results deteriorate in the year the market supposedly recovered.

For the first three quarters of the fiscal year ending August 2026 (September 2025 to May 2026), revenue was ¥28.29B (down 6.2% year on year) with an operating loss of ¥6.37B. That is more than double the ¥3.07B operating loss of the same period a year earlier. On the same July 15, the company revised its full-year forecast downward, raising its projected operating loss from ¥4.0B to ¥7.9B. The reason given: order volume and GMV are expected to come in below the initial assumption. Retained earnings stand at ¥-26.97B, and the equity ratio has kept falling, 76.2% (FY August 2024) → 73.7% (FY August 2025) → 67.8%. Users who purchased at least once per quarter fell from 2.9 million to 2.45 million.

From March 1, 2026, the company extended its “in-store price” scheme to all 47 prefectures and began waiving delivery fees. As of June, roughly 21,000 restaurants were covered. What Circana described in the previous section as “offering at the same price as in-store and waiving delivery fees” is, concretely, this. Orders are being won back with discounts, and losses are growing by exactly that much. The same shape shows up in both the market research and a single company’s financials.

Meanwhile, new entrants keep arriving. Rocket Now (operated by the Japanese subsidiary of Coupang of the US) announced on August 7, 2026 that it had expanded its coverage to 420 cities across 46 prefectures. That was 12 days before menu’s shutdown announcement. This is a market where an exit and an expansion land in the same month.

The last numbers menu published were four years old

menu was founded in October 2018, launched in 2019 as a takeout-only app, and expanded into delivery in April 2020. It entered a capital and business alliance with KDDI in June 2021 (becoming an equity-method affiliate), and in April 2023 moved to a joint venture with KDDI at 50.6% and Rayson at 49.4%.

In its growth years, menu was a company that published numbers. As of April 2021 it covered all 47 prefectures with roughly 53,000 partner restaurants. The June 2021 alliance release with KDDI put the figure at 60,000. The official site carries “over 91,000 restaurants nationwide *as of August 2022.” In October 2024 it announced that it had taken the top spot in new downloads among delivery apps for the first half of 2024.

The losses can also be traced as far as the statutory financial notices in the Official Gazette go. A net loss of ¥2.86B for the fiscal year ending August 2021. ¥2.63B for the period ending January 2022, with retained earnings at ¥-5.49B. A net loss of ¥3.19B and net assets of ¥500M for the period ending January 2023. It is clear that the company was burning ¥2–3B a year while adding partner restaurants. Financial notices after that cannot be traced, and KDDI’s earnings report for the fiscal year ending March 2026 contains no mention of menu and no related impairment.

And this release contains no user count, no restaurant count, no courier count, no revenue and no loss, not a single number. The most recent scale figure left on the official site is the 91,000 restaurants as of August 2022, which is to say four years ago. How many restaurants will be affected at shutdown, and how many couriers will lose work, has not been disclosed at this point. On handling partner restaurants and couriers, the release says only that it will proceed step by step “while giving full consideration to the impact”, no specifics on compensation or migration paths.

The language of exiting has barely changed in four years

menu is not the first departure from food delivery in Japan.

Rakuten Gurunavi Delivery in 2022, Chompy in 2023, menu in 2026. Four years apart, and the sentences are nearly identical: “intensifying competition” and “allocation of management resources.” When seven departures over five years can all be explained by those two phrases, it is more natural to read it as a structural problem in the market than as a matter of individual skill.

Only Wolt’s ending is somewhat different in character. It came as part of parent company DoorDash’s international portfolio review, exiting Japan, Qatar, Singapore and Uzbekistan simultaneously, announced on February 25, 2026 and shut down on March 4, just eight days from announcement to closure. By that standard, menu’s month and a half sits on the long side.

What becomes visible next to 358 published cases

As of August 20, 2026, this publication carries 358 cases. Among the 216 that disclose monthly revenue, the median is ¥1.5M. Narrowed to cases in Japan, ¥250K. The ¥2–3B menu was burning in a year exceeds a thousand years of a business at that median. It looks like a different world, one too far away to compare.

Even so, there is exactly one thing that can be compared cleanly: whether you publish numbers when you close.

The ten closed or withdrawn cases on this site all ended with the operator disclosing the figures. The record of running two rental spaces for 29 months at a loss of about ¥2.2M was written by the operator down to the breakdown, ¥730K in revenue, 95 bookings. The same goes for how a fan-work platform with 1 million monthly unique users and 10,000 active members came to close, and for why a logo generation service that ran five years at $3,000 a month was wound down. Nobody has any obligation to write about failure, and yet they wrote.

Meanwhile a business into which hundreds of billions of yen in capital must have been poured over five years leaves not one number behind when it ends. The volume of disclosure is determined not by the size of the business but by whether there is a motive to disclose. A subsidiary of a listed company has no motive to write out the performance of a business it is closing. An individual has a motive to leave a record of their own four years. This asymmetry is what creates a situation where an enormous industry called food delivery ends without anyone knowing how much was spent or how many restaurants were lost.

How far this generalizes

What can be drawn from this case is not the simple lesson that “you must not enter a market where orders can only be sustained by discounting.” It can only be said in a more limited form.

One point travels beyond food delivery: a business that needs subsidies (coupons, free delivery, in-store price parity) to sustain growth loses demand the moment the subsidies stop. Circana’s market data and Demae-can’s results back this up from separate angles. Among our own cases, there are a number of operators who chose to stay small and profitable by raising prices and narrowing their customer base. What they are avoiding is the same trap.

Plenty stays specific to this case. This is a platform market holding three sides at once (restaurants, couriers, users), and it does not apply to every industry. Nor can menu’s recent financials be confirmed, so its current financial position, including whether it is insolvent, is unknown. Reading it as “it failed because it had capital behind it” is also too quick. In the same market, Uber Eats has reported three consecutive profitable years since 2023. This should be treated as a question of unit economics rather than the presence or absence of capital.

And the numbers in this article have their own limits. Market size is an estimate by a research firm, and menu’s scale stops at a self-reported figure from four years ago. The true shape of a business that has ended never survives unless the people involved write it down. That blank is exactly why this publication collects individuals’ records of failure.

The service runs until September 30. How the handling of partner restaurants and couriers is disclosed from here is the part worth watching, as it could change the assessment of this exit.