The classification — splitting developers of record into four capital lineages

The developer field in the building plan data contains 50 different name strings. Some are single names; one lists seven companies jointly. Laying them out as-is reveals no structure, so we split them into four types on a single question: what is that entity’s core business? Where a name was borderline, we gave priority to the capital lineage further upstream. JR Tokai Hotels and Hankyu Hanshin Hotels, for example, are hotel operating companies, but because the land originates from the parent group’s holdings and rail-corridor assets, we placed them under rail and infrastructure.

Because the classification involves editorial judgment, the borderline cases are worth stating. Hulic Hotel Management is the operating arm of a real estate group, but since the developer of record is the operating company itself we placed it under operator self-development. The seven-company consortium in Naka Ward, Yokohama, led by Mitsui Fudosan also includes Keikyu Corporation and Tokyu, but we classified it by its lead company as developers and corporates. Kumoidori 5-chome Redevelopment Co. in Chuo Ward, Kobe, is not an “association,” but because it is the vehicle producing reserved floor area out of a redevelopment scheme it is included under redevelopment associations.

41.5% of rooms are developer-led, 34.2% are hotel operators building for themselves

By type, developers and corporates lead with 25 projects and 6,306 rooms (41.5% of rooms). Hotel operator self-development follows at 26 projects and 5,197 rooms (34.2%), then rail, power and telecom infrastructure at 11 projects and 2,476 rooms (16.3%), and redevelopment associations at 5 projects and 997 rooms (6.6%). Two projects totalling 213 rooms have a blank developer field; rather than infer a name we left them in a separate bucket.

Rank the same data by project count and the order flips. Operator self-development is the largest at 26 projects (37.7%), edging past the 25 of developers and corporates. Because average rooms per project run 252.2 for developers against 199.9 for operators, the room-count ranking reverses. In terms of how often a party actually appears in the market as a buyer of land, hotel operators themselves are the counterparty you meet most frequently.

Sorted by scheduled completion year: 22 projects and 5,304 rooms in 2025, 20 projects and 4,639 rooms in 2026, 17 projects and 3,618 rooms in 2027, 8 projects and 1,428 rooms in 2028, and 2 projects and 200 rooms in 2029 or later. The declining count in later years does not mean the pipeline is drying up. Building confirmation applications are normally filed one to two years before opening, so the further out the year, the more plans are simply not yet in the dataset. Figures from 2028 onward should be read as the floor of the currently confirmed pipeline.

All 26 operator projects are single-use — the four types are not fighting over the same sites

This is the metric that separates most cleanly in the whole dataset. All 26 hotel operator self-development projects, without exception, are single-use hotel buildings. By contrast, all 5 redevelopment-association projects are mixed-use, and 9 of the 11 infrastructure projects are mixed-use, leaving just 2 single-use (18%). Developers and corporates split almost evenly, with 13 single-use projects (52%).

The difference goes beyond product type. A single-use hotel building goes up on a mid-size site that works at roughly 3,000–8,000 m² of gross floor area. A mixed-use building stacking offices, residential and retail is an order of magnitude larger, and median floor count for the infrastructure type stands out at 28. In other words, the sites hotel operators are hunting and the city blocks infrastructure players and redevelopment associations handle are simply different in size. “Competition for land” does not mean all four types blending evenly: it appears as a two-tier structure in which operators and developers converge on mid-size sites that work as single-use, while the large blocks are locked up by infrastructure and redevelopment players.

22.1 months for operators, 67.0 for redevelopment associations — who carries the time

For the 58 projects with both a scheduled start date and a scheduled completion date, we measured the months in between. Medians: 22.1 months for hotel operator self-development (N=18), 34.5 months for developers and corporates (N=24), 50.6 months for rail and infrastructure (N=11), and 67.0 months for redevelopment associations (N=5). Shortest to longest is a threefold spread.

It would be a mistake, however, to read this gap as a measure of sponsor competence. Ignore type and cut the data by single-use versus mixed-use alone, and you get 23.7 months single-use (N=32) against 48.5 months mixed-use (N=26) — a spread of almost the same width. Within developers and corporates the same split appears: 27.1 months single-use, 46.0 months mixed-use. What sets the duration is principally the use mix of the building, and the four types are strongly correlated with that use mix, which is why the difference surfaces along type lines. An earlier construction-period analysis found the same direction — 1.96 years single-use against 4.00 years mixed-use — and these figures are consistent with it.

The practical implication runs in the opposite direction from the sequence you might expect. On redevelopment-association projects, five and a half years pass from the urban planning decision through rights conversion, groundbreaking and completion. The capital tied up over that period, and the inflation risk carried alongside it, ultimately get priced into the acquisition cost of the reserved floor area — that is, into the rent or acquisition price the hotel operator pays. The length of time is an internal cost to the sponsor and, simultaneously, a cost of goods for the tenant moving in.

Median gross floor per room: 18.8 m² versus 75.5 m² — the same “hotel plan,” different design logic

For the 37 single-use projects with a known gross floor area, we divided gross floor area by room count to get floor area per room. The median is 18.8 m² per room for hotel operator self-development (N=23) and 75.5 m² per room for developers and corporates (N=12). Within the same planning dataset, a fourfold spread.

This is not a question of better or worse but of different products. Operator self-development skews to lodging-focused formats, designed to extract as many rooms as possible from a constrained site. APA Group’s nine projects in particular run 15.5–20.3 m² per room (median 16.4 m²), an efficient use of floor that secures more rooms from the same gross floor area. That floor efficiency is the source of the land-acquisition power discussed below.

At the top of the developer-and-corporate range, meanwhile, sit projects in Hakone (187.3 m² per room) and Kyoto (152.2 m² per room). These are resort and upper-tier products with large baths, restaurants and banquet space, committing an order of magnitude more floor per room. Among operators too, Kyoritsu Maintenance, which runs onsen-and-dining stay formats, spans 35.9–124.9 m² per room, and Imperial Hotel’s Gion, Kyoto project sits at 196.4 m² per room — the most floor-intensive of any operator self-development here.

Sites spread across 23 prefectures, Tokyo leading with 18 projects and 3,497 rooms — metro dependence is flat across types

The project sites span 23 prefectures. Tokyo leads with 18 projects and 3,497 rooms, followed by Hokkaido with 9 projects and 2,345 rooms, then Kyoto, Chiba and Kanagawa with 5 each. Hokkaido mixes large resort projects in Kutchan (500 rooms) and around Niseko with projects in central Sapporo, averaging 260.6 rooms per project — above Tokyo’s 194.3.

Concentration in the three major metropolitan areas (Tokyo, Kanagawa, Chiba, Saitama, Osaka, Kyoto, Hyogo and Aichi) runs 68% for developers and corporates, 65% for operator self-development, 73% for rail and infrastructure and 60% for redevelopment associations — almost no difference by type. A tidy split of “operators in the regions, developers in the centre” does not appear in this data. The difference lies not in geography but in the use mix and site size covered in the previous section.

¥6,300 required ADR at zero land cost, ¥17,000 at central Tokyo land prices — a 2.7x asymmetry

Of the four types, only rail, power and telecom infrastructure operates under a structurally different condition: because it already owns the land, no external land cost has to be injected into the project. JR East’s projects in Shinagawa, Takanawa and Shibaura, TEPCO Power Grid and NTT Urban Development in Uchisaiwaicho, and Tokyu Group in Dogenzaka all rise on land the group already holds. To size what that difference is worth, we model a 200-room, single-use hotel.

■ Model assumptions

■ Results by land condition

The result is unambiguous. With zero land cost, an ADR of about ¥6,300 is enough to clear the 5.0% target. Buy the site at the average commercial land price of the 23 wards’ “other wards” (the 18 wards outside the five central ones) and you need about ¥8,500; at the all-23-wards average, about ¥11,600; at the five-central-wards average, about ¥17,000. Lightest to heaviest condition, a 2.7x spread.

The estimated transacted ADR for Tokyo business hotels averages ¥14,273 over the most recent 12 months (August 2025 – July 2026, N=894–924 properties). Apply that, and a zero-land-cost project still clears its target yield with ADR running 56% below the prevailing level, whereas a project bought at the all-23-wards commercial average has 19% of headroom, and at the five-central-wards average the prevailing ADR does not reach the required level at all. That is precisely why the combination of acquiring land at the average central-five commercial price and building a single-use, lodging-focused hotel does not appear even once in this dataset. Every plan standing in the five central wards shows up instead as a large mixed-use block.

■ Sensitivity analysis (ADR required for a 5.0% target NOI yield)

Push occupancy down to 75%, or squeeze the GOP ratio to 35%, and the zero-land-cost case still works below ¥7,200. Run it the other way and the five-central-wards case still needs ¥15,881 at 88% occupancy and ¥15,150 at a 45% GOP ratio — the distance to the prevailing ADR does not close (each sensitivity moves one variable at a time). The presence or absence of a land cost is not a gap that better occupancy or GOP can bridge.

In the 2026 Official Land Price Survey (as of 1 January 2026), commercial land in Tokyo rose 12.2% year on year, and 13.8% across the 23 wards. Even in the “other wards” outside the central five, the rise was 14.0% — the increase is not confined to the centre. The required ADR for a buyer of land ratchets up every year, while for a party using land it already owns it rises only by construction cost inflation. Read alongside our earlier work backing out the supportable ground rent under a leasehold structure, it becomes clear that the form of land tenure is itself a primary variable in project feasibility.

12 groups hold multiple projects — 42 projects, 8,361 rooms; the other 27 are one-offs

After consolidating groups, 39 distinct developers remain. Of these, 12 groups have two or more projects in planning, accounting for 42 projects and 8,361 rooms (55.0% of rooms). The remaining 27 projects belong to single-project sponsors — regional operating companies and one-off investment plays, widely dispersed.

The largest single supplier is APA Group (under the APA Home and APA Mansion names) with 11 projects and 2,714 rooms, or 17.9% of rooms. Roughly one in six of all 69 projects belongs to that group. As the previous section showed, it has standardised a floor-efficient design and can stack more rooms than rivals onto a site that works as mid-size. The ability to bid a higher total — room rate times room count — on the same piece of land is what underpins that volume. From the perspective of someone looking for land, it is a counterparty you will meet with high probability on mid-size urban sites.

Route Inn Japan follows with 4 projects and 726 rooms, Kyoritsu Maintenance with 4 and 642 rooms, and Toyoko Inn with 2 and 406 rooms (one of them jointly named with a real estate company). The leaders in operator self-development are all national chains — sponsors with standard designs and in-house construction capability. On the developer side, Mitsubishi Estate Group leads with 5 projects and 1,375 rooms, and among single projects the 1,000-room plan in Mihama Ward, Chiba (the mixed-use development in front of Kaihin-Makuhari Station) stands out.

What it means if you are looking for land — three readings

Frequently asked questions