NEOJAPAN Inc. (3921.T): Deep Dive Research Report

Technology sector, Tokyo Stock Exchange Prime. Report date: 20 October 2026.

Reporting-period anchor. NEOJAPAN's fiscal year ends on 31 January and it reports every quarter. Q2 of the fiscal year ending January 2027 (FY1/27, covering February to July 2026) should therefore have been reported in the second week of September 2026. It was: the half-year results and the briefing came out on 11 September 2026. The next release, Q3 FY1/27 covering August to October 2026, is due around 10 December 2026. This report uses six results events:

1. What the Company Does

NEOJAPAN writes the software Japanese offices use to run their day. Its main product, desknet's NEO ("desk-nets neo"), is groupware: one browser screen holding the shared calendar, the meeting-room booking sheet, the notice board, the leave and expense approval forms, the shared file cabinet, the visitor log and the staff directory. In Japan, a lot of paperwork still moves by hanko (personal seal) stamp and circulated memo. Groupware is the layer that turns that paperwork into screens. NEOJAPAN sells it in two ways. The first is a monthly per-user cloud subscription. The second is an "on-premise" package licence that the customer installs on its own servers. That second option matters a great deal to Japanese local governments and security-sensitive firms that cannot or will not put staff data in a public cloud.

Two products sit next to the groupware. AppSuite is a no-code tool: office staff build small business applications (an equipment ledger, an inquiry tracker, a sales log) by dragging fields onto a form inside desknet's NEO, without writing code. ChatLuck is a business chat tool, the Japanese enterprise equivalent of Slack, built for organisations that want chat with administrator controls and an on-premise option. The group also owns Pro-SPIRE, a contract software-development house, and runs small subsidiaries in Malaysia, Thailand, the Philippines and the United States. Section 2 covers how those pieces fit together.

The founding story. Per the company's annual securities report (yuho), founder and Representative Director and President Akinori Saito (齋藤晶議; registered name 齊藤章浩) has the following career:

• April 1980: joined Nippon Telegraph and Telephone Public Corporation, now NTT.
• August 1990: joined Business Infor Co., Ltd. (株式会社ビジネスインフォア).
• February 1992: founded NEOJAPAN and became its Representative Director and President, the post he still holds.

The filing gives no further detail on his work at either employer. The company was incorporated on 29 February 1992 in Yokohama. It started as a contract developer, building systems for clients that included power utilities and car makers.

The pivot that created today's business came in January 1999. NEOJAPAN began developing and selling a packaged web-based groupware product, iOffice2000, at a time when most Japanese groupware still ran as installed client software. The product line was renamed desknet's: an enterprise edition launched in 2001 and a small-business edition in 2002. It was rebuilt as desknet's NEO in 2012. ChatLuck followed in 2015 and AppSuite in 2017. NEOJAPAN listed on the TSE Mothers market in November 2015, moved to the First Section in January 2018 and to the Prime market in 2022. In 2019 it bought Pro-SPIRE and set up its US and Malaysian subsidiaries. Thailand followed in 2021 and the Philippines in 2024.

The milestone that reshaped the economics came in September 2024. NEOJAPAN raised the list price of desknet's NEO cloud from ¥400 to ¥600 per user per month and AppSuite cloud from ¥320 to ¥500. At the same time it launched a cheaper bundle of the two at ¥800 per user per month. Customers absorbed a 50% increase with almost no added cancellations (Section 4 explains why). That price rise drove the profit step-up in FY1/26. Much of the current investment question is about what drives growth once that one-time lift has worked through.

The value proposition. A Japanese company with 200 staff, or a city hall with 6,000, needs one place where everyone can see who is where, book the shared resources, and route the approval forms that Japanese administrative practice requires. It also needs that place to behave the same way for a decade. NEOJAPAN's pitch rests on three things:

• A familiar screen. Its product looks and works the way a Japanese office already does, for example with a seal-style approval workflow.
• A choice of deployment. The same product runs as cloud or as an on-premise install, including inside Japan's closed LGWAN local-government network.
• Long support. It commits to supporting installed versions for a long time.

What makes it hard to replicate, in headline only (Section 3 has the detail):

• More than 25 years of accumulated Japanese workflow conventions inside one code base.
• A product that runs both as multi-tenant cloud and as a customer-hosted install.
• A government-grade security posture.

None of that is exotic technology. It is the kind of accumulated fit with the customer's way of working that a global horizontal suite does not bother to build.

An example. A designated city (seirei shitei toshi, one of Japan's large cities with prefecture-level powers) in Kanagawa wants one shared groupware for every ward office and bureau. Yokohama did exactly this: it chose desknet's NEO as a common city-wide groupware for up to 60,000 users. Staff log into a portal and see the day's schedule, their bureau's notices and the pending approvals. A ward clerk files a leave request, which goes up a three-stage seal workflow. A facilities team uses an AppSuite app built by a non-programmer to track vehicle bookings. Because the system sits inside the city's own network, resident-related information never leaves it. NEOJAPAN's job is to license and support that install for years, through version upgrades and changes in operating systems and browsers. When the city later wants generative-AI search over its documents, NEOJAPAN provides that as an add-on inside the same screen.

2. Business Segments

NEOJAPAN reports three segments. Software produces almost all of the group's profit. The contract-development subsidiary adds revenue scale and engineers. The overseas segment is a small bet still in its investment phase.

2.1 Software (~75% of FY1/26 revenue)

What it does. This segment holds desknet's NEO, AppSuite, ChatLuck and the newer AI services (Section 3). It sells to Japanese companies of every size and to the public sector, through two revenue streams:

• Cloud services, roughly 52% of group revenue in FY1/26. These are per-user monthly subscriptions.
• Products, roughly 21%. These are on-premise package licences plus their annual support contracts.

A small "technical development" line, about 2%, covers custom work and OEM development on top of the products. Starting in FY1/26, the US subsidiary DELCUI Inc. was moved from the overseas segment into Software.

The core capability. The segment knows how to build one Japanese-workflow application that runs either as a multi-tenant cloud or inside a customer's own data centre, including closed government networks, and to keep both versions alive for many years. Rivals have mostly chosen one of the two.

Why it is separate. This is NEOJAPAN's own intellectual property, sold on licence and subscription economics.

Role in the group. Software is the profit engine. Management said it ran at roughly a 40% segment operating margin in FY1/26, against about 30% for the group. Every strategic priority management lists starts here: NEO user growth, AppSuite cross-sell, AI and the on-premise replacement cycle.

Main rivals: Microsoft 365, Google Workspace and Cybozu (Office, Garoon, kintone). Section 5 analyses them.

2.2 System Development Services: Pro-SPIRE (~24% of FY1/26 revenue)

What it does. Pro-SPIRE is based in Ota-ku, Tokyo, and has been fully owned since August 2019. It does contract development and engineer-dispatch work: embedded systems plus business systems such as BI, data warehouses and CRM. Its main customers are life and non-life insurers. In April 2024 it had about 132 staff, around 90% of them engineers.

The core capability. Its domain knowledge of insurance systems and a stable pool of engineers. It has no product of its own.

Why it is separate. It has a different economic model (billable engineer hours), different customers (insurance IT departments, not general affairs managers or municipalities) and a different history (it was acquired).

Role in the group. It brings revenue stability and a source of engineers, but it is not a margin contributor. Its segment profit is thin: in Q1 FY1/27 it was a small fraction of the Software segment's. In H1 FY1/27 its profit was flat because management pulled mid-career hiring forward. Management also names Pro-SPIRE synergies as one of two acceptable M&A themes.

Main rivals: the large population of Japanese contract developers and staffing-style SIers (system integrators) serving financial institutions.

2.3 Overseas (~1% of FY1/26 revenue)

What it does. Local sales of the NEOJAPAN product family, mainly from NEOREKA ASIA in Malaysia, plus NEO THAI ASIA in Thailand and NEOPhilippine Tech in the Philippines.

The core capability. Not yet established. This is a market-entry effort.

Why it is separate. Different currencies, a different go-to-market and early-stage losses.

Role in the group. It is the long-dated growth option, and management describes it plainly as being at the "upfront investment stage" (先行投資段階). The segment is still loss-making.

Main rivals: the same global suites (Microsoft, Google) plus local and regional vendors.

Geographic mix and export history

NEOJAPAN is almost entirely a Japanese business: about 99% of revenue comes from Japan. All of its scale, its public-sector base and its partner channel are domestic. The international story is about seven years old:

• United States, 2019: DELCUI Inc. set up in California, now reported inside Software.
• Malaysia, 2019: NEOREKA ASIA set up. In September 2025 it signed a basic agreement with Invest Johor to build an ASEAN base. In May 2026 it held a regional user conference, "NEO BUSINESS CONNECT".
• Thailand, 2021: NEO THAI ASIA set up.
• Philippines, 2024: NEOPhilippine Tech set up.

The region's logic differs from Japan's. In Japan, NEOJAPAN sells to buyers who already understand groupware. In ASEAN, it sells partly to Japanese-affiliated subsidiaries and partly to local firms that are digitising paper workflows for the first time, competing directly against free or bundled global tools. Overseas revenue is still growing fast from a tiny base, but it does not yet change the group's mix.

3. Products and Operations

desknet's NEO: the core

desknet's NEO bundles more than 25 standard functions into one portal: schedule, facility booking, notices, a workflow engine for approval forms (including seal-style multi-stage routing), timecards, a web mail client, document management, a visitor log, safety-confirmation messages for disasters and an address book. It comes in three deployment forms:

• Cloud edition: NEOJAPAN-hosted and multi-tenant. Standard list price ¥600 per user per month since September 2024. It reached about 577,000 cloud users at the end of July 2026, up 4.9% year on year.
• Package (on-premise) edition: a one-time licence plus annual support, installed on the customer's servers. List examples are about ¥500,000 for 100 users and about ¥4.76m for 1,000 users. Cumulative package sales exceed 5.4 million users, which reflects how deep the installed base is.
• Government-security option: a dedicated cloud environment for desknet's NEO and AppSuite. It was registered on the ISMAP cloud service list on 3 September 2026. ISMAP is the Japanese government's security evaluation scheme for cloud services, and listing is close to a precondition for central-government cloud procurement.

The technical points that matter:

• One code base, two deployments. The on-premise edition can be installed inside the LGWAN, the closed network that Japanese municipalities use for core administrative work. Many cloud-only competitors cannot serve that environment at all.
• Long support windows. Management has made a policy of continuing on-premise support while peers sunset theirs (see the quote below and Section 5).
• AI layered on top. In September 2025 it launched neoAI Chat for desknet's, a generative-AI chat service that works with desknet's NEO and can connect to on-premise deployments. Version 9.7 of desknet's NEO, released in autumn 2026, strengthened the built-in AI assistant. An AI agent inside NEO, designed to automate routine work using calendar and document data, is targeted for spring 2027 (Section 7).

"オンプレミスのサポートをやめません" ("We will not stop supporting on-premise.") - President Akinori Saito, FY1/26 results briefing, 11 March 2026

AppSuite: the cross-sell engine

AppSuite runs inside desknet's NEO and lets non-programmers build database-style applications: forms, lists, calculations, approval routing and dashboards. It is sold as a cloud add-on (or in the ¥800 bundle) and as a package add-on. Because each AppSuite app stores the customer's own operating data and logic inside NEOJAPAN's platform, it is both a revenue line and a source of lock-in. Section 4 covers that lock-in effect.

AppSuite cloud users reached about 132,000 at the end of July 2026, up 38.9% year on year. Q2 FY1/27 added about 13,000 users, the best quarter on record. About 23% of desknet's NEO cloud users now also use AppSuite, against roughly 17% a year earlier. Management targets more than 30%.

ChatLuck: secure business chat

ChatLuck is enterprise chat with administrator controls, audit logs and an on-premise option. It is a smaller product. In H1 FY1/27, large package deals for ChatLuck were the main reason product (package) sales grew.

NEOPORT and LiveX AI: the customer-facing bets

• NEOPORT (2024) is a shared inbox and inquiry-management tool for teams handling customer emails. NEOPORT v2.0 (September 2025) added AI-agent automation of email replies.
• LiveX AI is a US-developed enterprise AI agent that handles customer conversations across chat, voice and email. In October 2025, NEOJAPAN became the first Japanese company to sign as a LiveX AI resale partner, and it began selling the service in Japan. On NEOJAPAN's own support portal, LiveX resolves about 90% of inquiries without a human. In March 2026, NEOJAPAN, ZETA and LIVEX AI formed a three-way alliance to put AI agents into physical retail stores (conversational customer service, product search, review lookup and in-store ad serving). A proof of concept is running with a professional basketball club.

How the products are made and delivered

The value chain is short:

1. Product development. Done in-house in Yokohama (headquarters in the Yokohama Landmark Tower) by the Technical Development Division.
2. Cloud hosting. NEOJAPAN operates the multi-tenant cloud, plus a separate dedicated environment for the ISMAP option.
3. Distribution. A mix of resellers and system integrators (for packages and public-sector tenders) and direct online sign-up (for cloud).
4. Support. Customer success and support teams, increasingly helped by NEOJAPAN's own LiveX deployment.

There are no plants and no inventory. The real constraints are:

• Engineering headcount, which in H1 FY1/27 meant pulling hiring forward at Pro-SPIRE and expanding overseas teams.
• The cost of maintaining many versions for long-lived on-premise installs.
• The release cadence needed to keep pace with Microsoft's and Google's AI features.

The one notable "capacity" event was certification rather than construction: the ISMAP registration in September 2026, which the company had targeted for completion within 2026.

4. Customers

Who buys. NEOJAPAN does not disclose a revenue split by customer type, so this section describes the buying relationship rather than percentages. There are two broad bases.

1. Japanese private-sector organisations of every size. Most cloud users sit at small and mid-sized firms across manufacturing, construction, services, healthcare and distribution. The company also has large licences of more than 1,000 users. In Q1 FY1/26 the count of these large licences fell 1.8% even while the overall cloud base grew, a data point Section 8 returns to.

2. Public bodies. desknet's NEO is used by more than 1,250 government agencies and local governments, including about one-third of Japan's prefectural governments and about 40% of its designated cities. Yokohama (up to 60,000 users) and Kamakura are named public references.

Who decides, and how.

• SMEs. The decision usually sits with the general affairs (somu) or administration manager, often with the president's sign-off, rather than with a dedicated IT team. The criteria are price per user, how easy the screen is for non-technical staff, and whether the workflow forms match existing paper forms. The sales cycle is short: weeks, often starting from a free trial or a reseller recommendation.
• Large enterprises. The information-systems department runs the evaluation, with security, single sign-on and the ability to host internally weighed against "we already pay for Microsoft 365". The cycle runs to months.
• Municipalities and agencies. The information policy division buys through formal procurement or tender, often via a local system integrator. The criteria are LGWAN compatibility, security certification (now ISMAP for cloud), price and long-term support commitments. Cycles follow the April-March budget year and can take a year or more.

Why they choose NEOJAPAN. There are five specific reasons:

1. It runs where the data must stay. It is one of the few mainstream groupware products that still installs on-premise, inside the LGWAN, with a vendor committed to keep supporting it.
2. The screen matches Japanese office work. That includes seal-style approval routing and the many small conventions of Japanese administrative forms.
3. Price per seat. Even after the 2024 rise, the price sits below what a full Microsoft 365 business tier costs for organisations that only need groupware functions.
4. AppSuite. It lets non-IT staff digitise their own small processes without a separate no-code purchase.
5. Satisfaction scores. NEOJAPAN ranked first in information-sharing software and services in Nikkei Computer's 2026-27 customer satisfaction survey (3 September 2026 issue).

Switching costs. These are higher than a groupware product's simplicity suggests:

• Leaving means retraining every employee on a daily tool.
• Rebuilding the approval workflows means re-mapping who approves what across the organisation.
• Historic schedules, notices and approval records have to be migrated.
• Customers using AppSuite have to rebuild their custom apps somewhere else.
• For public bodies, a switch can require a fresh procurement and a new security review.

The best evidence is behavioural. Through the September 2024 price increase of 50%, monthly cloud churn stayed around 0.30% (Q2 FY1/26), fell to 0.21% in Q1 FY1/27, and averaged 0.31% across H1 FY1/27, with Q2 at 0.42%. Customers that face a 50% rise and do not leave are telling you what switching costs them.

Concentration. No single customer is disclosed as material, and the base is very fragmented across hundreds of thousands of seats. The concentration that does exist is in a type of buyer: the public sector, plus the on-premise installed base. Pro-SPIRE is more concentrated, with insurance companies as its core clients, but it is a minority of group revenue and an even smaller share of profit.

Contract structures and predictability. Three structures coexist:

• Cloud: monthly per-user subscription that renews automatically.
• Package: a one-time licence followed by annual support and maintenance fees.
• Pro-SPIRE: project- and time-based contracts.

Recurring ("stock") revenue was 81.9% of the total in FY1/26 and 85.6% in Q1 FY1/27. That makes the business highly predictable, apart from two swing factors: one-off package licence sales (sensitive to replacement cycles) and Pro-SPIRE project timing.

5. Competitive Landscape and Moat

Industry structure. Japanese groupware has two layers. On top sit the two global horizontal suites, Microsoft 365 and Google Workspace, which bundle mail, documents, chat and calendar. Microsoft now also bundles AI through Copilot. Underneath sit domestic specialists, which win where Japanese workflow fit, price per seat or deployment flexibility matter more than the breadth of the suite. Cybozu is the largest domestic specialist, with Cybozu Office for SMEs, Garoon for large organisations and kintone for no-code. NEOJAPAN is the second domestic name. A tail of chat and point tools also competes for parts of the workflow.

User surveys show the shape. In BOXIL's survey of 1,704 users:

Other surveys that count suite subscriptions rather than groupware users put Microsoft and Google far ahead. There is no audited share table. The fair summary is that NEOJAPAN is a clear number two among domestic specialist groupware vendors, with a disproportionately strong position in the public sector.

Why NEOJAPAN wins or loses, against each:

• Against Microsoft. NEOJAPAN wins where the buyer needs seal-style approval workflows out of the box, an on-premise or LGWAN deployment, or a lower price per seat for staff who do not need a full productivity suite. It loses in large enterprises that have standardised on Microsoft 365 and see groupware as a feature they already own. The fall in large licences noted in Section 4 is consistent with that pattern.

• Against Google. NEOJAPAN rarely meets Google in the public sector. It loses cloud-native SMEs that live in Gmail and Docs.

• Against Cybozu. This is the most direct contest, and the one where the field is shifting toward NEOJAPAN. Cybozu has chosen to move customers to the cloud:

  • It phased out sales of the package edition of Cybozu Office from 2021 to 2026, and support ends by 30 September 2027.
  • It will begin ending sales of package Garoon in October 2027, with the final licence date 31 October 2031 and support ending 31 January 2033.

  Every on-premise Cybozu customer that cannot or will not move to a public cloud has to choose a new vendor. NEOJAPAN has said, in effect, that it will be the last major vendor still supporting on-premise. In Q2 FY1/26 management described its strategy as "establishing our position as the No.1 company" in on-premise groupware as competitors withdraw. In no-code, kintone is the larger and better-known product. AppSuite's advantage is that it lives inside NEO with no separate purchase.

• Against kubell. ChatLuck competes on administrator control and on-premise deployment, not on network size. It is a niche win, not a share battle.

Barriers to entry. These are moderate for a generic groupware product and high for NEOJAPAN's specific niche:

• A new entrant would need years of Japanese workflow detail.
• It would need a dual cloud and on-premise code base.
• It would need ISMAP and LGWAN credentials, which take time and audit expense.
• It would need a reseller network that public bodies trust.

The bigger threat is substitution by a suite the customer already pays for, not a new specialist.

Structural shifts. Three are under way:

1. The domestic specialists are splitting on deployment. Cybozu is going cloud-only. NEOJAPAN is staying dual-mode. That creates a one-time migration pool through 2027 to 2033.
2. AI is moving the value from features to agents. Microsoft's Copilot sets the expectation. NEOJAPAN's answer is neoAI Chat, the AI agent planned for spring 2027 and LiveX.
3. Government security standards are tightening. ISMAP has become a gate for public cloud procurement, which favours vendors that invest in certification. NEOJAPAN cleared that gate in September 2026.

Moat: Narrow - switching costs, regulatory or qualification barriers

Customers accepted a 50% price increase with monthly churn staying around 0.2% to 0.4%, which is direct evidence of switching costs (Section 4). The LGWAN and on-premise capability plus ISMAP registration create a qualification barrier in the public sector that global suites and cloud-only rivals do not fully meet. The moat is narrow, not wide, because it has never been tested against a determined, bundled AI push by Microsoft into NEOJAPAN's core municipal and SME base. The drift of large licences shows it is already thin at the enterprise end. The single thing most likely to erode it is Microsoft making groupware functions plus Copilot "free enough" inside a suite that public bodies are already allowed to buy.

6. Industry

What drives demand. Japanese demand for groupware and collaboration software rests on five forces:

1. A shrinking workforce. Japan's labour shortage pushes employers to digitise paper and seal workflows instead of adding administrative staff.
2. Work-style reform and remote work, which the pandemic made permanent.
3. Replacement cycles in the installed base. The October 2025 end of Windows 10 support was a recent one. Vendors withdrawing on-premise products, as described in Section 5, create forced migrations.
4. Public-sector digitisation. Policy pressure on local governments, central-government cloud adoption under ISMAP, and continuing LGWAN-based operations at municipal level.
5. Generative AI, which is turning groupware from a record-keeping tool into an assistant layer. That raises potential spend per seat but also invites the suites to absorb the category.

Size and growth. Deloitte Tohmatsu MIC's 2025 report on collaboration and mobile-management software (a broad market that includes groupware, business chat, web conferencing, workflow, document management and online storage) puts the market at ¥327.3bn in FY2024. It forecasts 15.7% annual growth to ¥678.6bn in FY2029. Groupware is a mature subset growing more slowly than the total. The fast-growing parts are AI, no-code and chat, which is exactly where NEOJAPAN is adding products.

Where NEOJAPAN sits in the supply chain. It is an independent software vendor at the application layer:

• It sits above infrastructure: its own servers or data-centre space for cloud, and the customer's servers for packages.
• It sits below the system integrators and resellers who deliver to end customers, especially public bodies.

It does not depend on a hyperscaler marketplace for distribution. It does partner on AI models and agents: LiveX is resold, and neoAI uses third-party large language models.

Import substitution. The Japanese market is a long-running contest between imported global suites and domestic products. There is no protectionist policy. What exists is the practical effect of data-sovereignty and security rules: ISMAP is a government-wide security standard and LGWAN separation applies to municipalities, and both favour vendors that can deploy in specific environments. Domestic specialists keep share where the work is Japanese-specific. Global suites take it where the buyer treats groupware as one feature among many.

Regulation.

• ISMAP listing is the gateway to central-government cloud procurement.
• LGWAN separation rules shape what municipalities can run where.
• Japan's data-protection regime governs where personal data may be held.

These rules raise the cost of serving the public sector and are a net positive for vendors that have paid that cost.

Cyclicality. Low. Groupware seats follow employment, not GDP, and subscriptions renew automatically. In a downturn, new seat additions and package purchases slow, and Pro-SPIRE-style contract development is the most cyclical part of the wider IT services market, because insurers can defer projects. Public-sector demand follows the April-March budget year and is countercyclical, if anything.

Tailwinds and headwinds, industry-level.

• Tailwinds:
  • Structural pressure to digitise administrative work.
  • Forced migrations as on-premise products are withdrawn.
  • Government security standards that reward certified domestic vendors.
  • AI adding a new paid layer to the seat.
• Headwinds:
  • Suite bundling, especially Microsoft's AI-inclusive tiers.
  • Commoditisation of basic calendar and chat functions.
  • Engineering wage inflation in a tight labour market.
  • The risk that AI agents make the "groupware screen" itself less central.

7. Growth Triggers

Ranked by how much each could change the business:

• Capturing on-premise customers that Cybozu and others are abandoning. Management frames on-premise as a share-gain opportunity while competitors exit (the competitor timelines are in Section 5). It has repeatedly committed to keeping on-premise support (Q2 FY1/26 briefing, 12 Sep 2025; FY1/26 briefing, 11 Mar 2026). The large ChatLuck and package deals in H1 FY1/27 were cited as early evidence (Q2 FY1/27 briefing, 11 Sep 2026). Repeated across three briefings.

• An AI agent built into desknet's NEO, targeted for spring 2027. It is designed to carry out routine work using the user's calendar and document data, and the next-generation version is to connect with outside systems through APIs (Q2 FY1/27 briefing, 11 Sep 2026). It follows the September 2025 launch of neoAI Chat for desknet's and NEOPORT v2.0 (Q2 FY1/26 briefing, 12 Sep 2025), and management's stated AI positioning:

  "誰もが安心して快適にAIを使える" ("AI that anyone can use safely and comfortably") - FY1/26 results briefing, 11 March 2026

  The pitch is governance, security and audit trails for organisations that cannot use consumer AI freely.

• Raising AppSuite adoption among NEO cloud users above 30%. It was about 17% in mid-2025 and about 23% in mid-2026. The ¥800 bundle is the main tool, and management expects AppSuite to keep growing at 30%+ (Q2 FY1/26 briefing, 12 Sep 2025; FY1/26 briefing, 11 Mar 2026; Q2 FY1/27 briefing, 11 Sep 2026). Repeated in every briefing for over a year.

• Public-sector cloud through ISMAP. The government-security option for desknet's NEO and AppSuite cloud was registered on the ISMAP cloud service list on 3 September 2026. That opens central-government and security-sensitive cloud procurement that was previously closed (company announcement, 3 Sep 2026; referenced in Q2 FY1/27 materials).

• The three-year plan to FY1/2029. It targets roughly 7% annual revenue growth and 8% annual operating profit growth, built from NEO expansion, AppSuite cross-sell, overseas and LiveX AI (FY1/26 briefing, 11 Mar 2026; reaffirmed in the Q2 FY1/27 briefing, 11 Sep 2026).

• M&A. Management says surplus cash is earmarked for acquisitions of businesses that offer desknet's NEO cross-sell opportunities or synergies with Pro-SPIRE (Q2 FY1/27 briefing, 11 Sep 2026). No target has been named.

• LiveX AI and retail AI agents. This covers the Japanese resale of LiveX AI (launched October 2025), the ZETA/LiveX alliance for in-store retail AI (March 2026), and proof-of-concept expansion, including with a professional basketball club (Q1 FY1/27 materials, 11 Jun 2026; Q2 FY1/27 briefing, 11 Sep 2026).

• ASEAN expansion toward 10,000 cloud users. The vehicles are Malaysia (Invest Johor basic agreement, Q2 FY1/26 briefing, 12 Sep 2025), Thailand and the Philippines. Headcount and marketing are being built up during the investment phase (Q2 FY1/27 briefing, 11 Sep 2026).

8. Key Risks

8.1 Bundled suites with AI absorb the category. Mechanism. Microsoft 365 already includes calendar, chat, documents and a low-code platform, and Copilot raises the value of staying inside one suite. If Microsoft's ISMAP-listed government cloud and AI tiers become the default for municipalities and mid-sized firms, the standalone groupware seat loses its reason to exist outside on-premise installs. The process would be gradual: fewer new seats first, then large licences migrating at renewal, then pressure on price. Early evidence. The decline in >1,000-user licences while the total cloud base grew (Section 4). Calibration. High probability of a slow drag, moderate probability of a material one within five years. This is the risk that matters most.

8.2 The price lever has been pulled, and growth now needs volume. Mechanism. The September 2024 price rise drove FY1/26. Guidance for FY1/27 implies mid-single-digit growth, and Q1 FY1/27 growth was the slowest in years because the price effect had worked through. From here, growth needs NEO seat additions (about 5% a year), AppSuite attach and new products. Another general price increase would test the switching-cost evidence again. Not raising price leaves growth near the base rate. Calibration. High probability of a moderate drag on growth. It is not a threat to the franchise.

8.3 On-premise replacement demand is lumpy, then falls away. Mechanism. The strategy of being the last on-premise vendor turns competitors' withdrawals into a pool of orders that peaks around specific deadlines (Cybozu Office support ending in 2027, the Garoon sales and support wind-down from 2027 to 2033) and around operating-system cycles. Package licence sales are one-time. Q1 FY1/27 product sales fell before large deals lifted Q2. If the pool turns out smaller than expected (because customers choose the cloud, or Microsoft), or if NEOJAPAN wins it early and then faces a gap, product revenue will swing, and the on-premise code base becomes a maintenance cost carried for a shrinking group of customers. Management has explicitly tied its positioning to this window:

"No.1企業としての地位を確立する戦略" ("a strategy to establish our position as the No.1 company") - Q2 FY1/26 briefing, 12 September 2025, on on-premise groupware as competitors withdraw

Calibration. Moderate probability. The effect is quarterly volatility more than lasting damage.

8.4 AI execution: being late or second-rate. Mechanism. The spring 2027 AI agent (Section 7) has to be good enough that customers see a reason to stay rather than move to Copilot. NEOJAPAN builds on third-party models and a resold US agent (LiveX). If model costs rise, if the partner's priorities change, or if the in-house agent ships late or underwhelms, the "safe AI" pitch becomes "slower AI". Calibration. Moderate probability. The impact rises over time because the AI layer is where suites will compete.

8.5 Capital allocation: overseas spending and M&A. Mechanism. Two uses of cash are open-ended:

• Overseas. After about seven years the overseas segment is still loss-making, and management is increasing headcount and marketing.
• M&A. Management has signalled acquisitions using the cash pile but has not defined price discipline. Previous acquisitions (Pro-SPIRE) produced a low-margin services arm.

An expensive deal outside the core, or overseas losses that never close, would dilute a high-return software franchise. Calibration. Moderate probability. The impact depends on the size of any deal.

8.6 Founder concentration and succession. Mechanism. The founder-President, now 64, controls about 46% of the shares and is the product and strategy authority (Section 10). There is no public succession plan. A sudden departure, or an unclear handover, would hit decision-making in a 300-person company where the founder still personally presents results. The arrival of an activist holder (Section 11) adds the possibility of a public disagreement over capital policy. Calibration. Low probability in the near term. High impact if it happens.

8.7 Public-sector procurement concentration. Mechanism. A meaningful share of the brand and of the on-premise base sits in prefectures, designated cities and agencies (Section 4). A central policy shift (mandated common platforms, shared government groupware, or procurement that favours suites) could change renewal decisions across many customers at once, instead of one at a time. Calibration. Low to moderate probability. The impact would be concentrated.

9. Walk the Talk

Results events used: June 2025 (Q1 FY1/26), 12 Sep 2025 (Q2 FY1/26), 11 Dec 2025 (Q3 FY1/26), 11 Mar 2026 (FY1/26), 11 Jun 2026 (Q1 FY1/27) and 11 Sep 2026 (Q2 FY1/27). The most recent is 39 days old. Full Q&A transcripts exist only for the half-year and full-year briefings. For the Q1 and Q3 periods this section relies on the results summaries, decks and press coverage.

Commitments worth a closer look.

The price rise. This was the call with the most at stake: raise the core cloud product's price by half and the add-on's by more than half, while offering a bundle to steer customers toward AppSuite. Management said churn would stay contained. It did, quarter after quarter. The bundle also worked as designed: AppSuite attach rose by about six percentage points in a year. A management team that understands how much pricing power it has, and uses it once, carefully, is rarer than one that announces growth initiatives.

The FY1/26 forecast. The original plan was set with the price effect only partly visible. Management did not raise it until the third quarter, and then still beat the raised number. That pattern (a conservative start, a late raise, a beat) repeats in FY1/27: Q1 and H1 both came in ahead of the internal plan, and the full-year forecast stayed unchanged at the half-year. The company's own framing in March 2026, as reported, was that results "exceeded the plan after the upward revision".

The softer spots. Overseas has no measurable milestone beyond a user target, and the segment has stayed in an "investment phase" since the subsidiaries were set up in 2019. Pro-SPIRE's profit has been flat-to-thin, and management's explanation each time (hiring pulled forward) is reasonable but has not yet been followed by margin expansion. Neither was a broken promise. Both are areas where management has avoided making a promise that could be checked.

Assessment. On the core software business, this management guides conservatively and delivers. The price rise, the dividend path, ISMAP and the AI launches all happened as described or earlier. The pattern is a cautious initial plan followed by quiet beats, not optimism followed by excuses. The one place to discount the narrative is outside the core (overseas and M&A), where the language is aspirational and no measurable target has yet been set.

10. Ownership, Governance and Shareholder Friendliness

Part 1: Ownership and control. NEOJAPAN is a founder-controlled company. Per the FY1/26 annual securities report, President Akinori Saito beneficially owns 6,441,900 shares (45.73%). That figure explicitly includes the 5,200,000 shares (about 37.1%) held by Proceedus Inc., an asset-management company over whose voting the President has authority, plus about 1,241,900 shares (8.85%) he holds directly.

Other large holders on the 31 January 2026 register:

Since then, Kaname Capital, L.P. has disclosed a stake of just over 5%, with a stated purpose that includes making important proposals (the transaction is in Section 11). By category, domestic corporations (mostly Proceedus) hold about 38%, individuals about 42%, foreign holders about 13% and financial institutions about 7%. There is a single class of shares and no cross-holding or pyramid structure was found.

What this means for a minority shareholder:

• One person effectively holds just under half the votes. He can block special resolutions, which need two-thirds, on his own, and in practice decides ordinary ones.
• The roughly 25% held by the next three individuals and the employee association means the true free float is well below 50%.
• No minority holder or activist can force a change in strategy or capital policy without the founder's agreement.

Part 2: Governance.

• Board. Eight directors, three of them outside directors designated as independent (a tax accountant, a person with banking and corporate-systems experience, and an attorney). The audit and supervisory board model has three statutory auditors, two of them outside and independent. The President, who is also the controlling shareholder, is the top executive. There is no separate independent chair.
• Related party. The only related-party item disclosed is that an outside director's tax firm buys NEOJAPAN cloud services worth under ¥100,000 a year. The company judges this immaterial to independence (FY1/26 securities report).
• Other flags. No auditor change, modified audit opinion or insider share pledge appears in the filings reviewed. The extracts reviewed did not show whether executive pay is tied to measurable performance targets.

Beyond the control concentration described above, governance is unremarkable.

Part 3: Capital returns.

Dividends. Annual DPS was ¥23 for FY1/24, ¥40 for FY1/25 and ¥52 for FY1/26, and ¥54 is forecast for FY1/27. That is three consecutive increases. The step-up in FY1/25 came when the company introduced an interim dividend, raised its target payout ratio and adopted a progressive (no-cut) dividend policy in March 2025. The payout target now stands at about 40%. The FY1/26 dividend was raised twice during the year as the forecast was raised.

Buybacks and dilution. Older programmes, outside MoatMap's window:

• March 2024. NEOJAPAN bought back 850,000 shares (about 5.7% of shares then outstanding) for about ¥951m through a single off-market ToSTNeT-3 transaction (Tokyo Stock Exchange off-hours trading system). Major shareholder Katsuya Otsubo's holding fell by the same amount (filings of 14 and 18 March 2024).
• 31 July 2024. All 850,000 shares were cancelled (board resolution announced 30 May 2024).

That was a single block purchase, not an open-ended programme, so the shares authorised and the shares bought were the same. No other repurchase programme was found in FY1/24 through FY1/27 to date.

Recent window, per MoatMap. No buybacks recorded between 22 July and 20 October 2026. Treasury shares stood at 69,700 (about 0.5%) on 31 July 2026.

Net change. Issued shares fell from about 14.9 million to about 14.1 million over the three years, a reduction of roughly 5.7%, all from the 2024 cancellation. There has been no meaningful option dilution.

Verdict: Returns Capital. The company has raised its dividend every year under a stated progressive policy and retired about 5.7% of its shares, though a growing cash balance that management has earmarked for M&A, not distribution, keeps it short of returning everything it earns.

11. Insider Activities

MoatMap's Japanese large-holding data records one insider event in the last 12 months. On 11 May 2026, President Akinori Saito (registered name 齊藤章浩, a ≥5% holder) filed change report No. 4 under the 5% rule (EDINET large shareholding change report, 2026-05-11; reporting obligation date 2026-05-01). MoatMap lists it as a buy of 332,449 shares, equal to 2.37% of shares outstanding. The filing does not support reading this as a fresh open-market purchase:

• It reports a joint holding with Proceedus of 6,441,900 shares: 1,241,900 held directly and 5,200,000 through Proceedus.
• Those are the same share counts the company's register showed on 31 January 2026.
• The rise in the reported ratio (Proceedus from about 34.9% to 36.9%, Saito from 8.51% to 8.82% against the previous report) matches what happens when the share count shrinks: the July 2024 cancellation of 850,000 shares (Section 10) raised every holder's percentage without any purchase.

No price or value is disclosed, and no trades within the window were visible in the sources reviewed. This filing is treated as a recalculation of an existing stake, not as conviction buying.

The other ≥5% event in the window was an outside holder buying. Kaname Capital, L.P., a Japan-focused fund described in coverage as an activist, crossed 5% on 9 April 2026, holding 714,200 shares (5.07%). Its report gave "pure investment and, depending on circumstances, making important proposals" as the purpose (EDINET large shareholding report, filed 2026-04-30 and 2026-05-01; reported value about ¥1.2bn). No director or officer sold shares in the window.

Read: neutral on management, with a mildly positive outside signal. The founder's stake is unchanged and very large. The new money in the window came from an outside fund that may push on capital policy.

12. Scenarios

Bull case. The on-premise strategy pays off at scale. Package Cybozu Office customers whose support ends in 2027, and the first wave of package Garoon customers, choose desknet's NEO because it lets them stay on their own servers. Many then later move to NEOJAPAN's own cloud instead of a rival's. ISMAP registration opens central-government and security-sensitive cloud deals that were previously out of reach. The spring 2027 AI agent lands as the "safe AI" option for municipalities and regulated firms that cannot adopt Copilot freely, turning AI into a paid add-on per seat. AppSuite moves past a 30% attach rate. In two to three years NEOJAPAN is the default domestic groupware for anyone who needs on-premise or a Japanese workflow, with spending per user rising through add-ons rather than list-price increases. If a disciplined acquisition is made, it adds a product that NEO's installed base can absorb.

Base case. Management delivers roughly what it has guided. NEO cloud seats grow in the mid single digits, AppSuite keeps compounding, and on-premise replacement deals show up as lumpy but positive quarters through 2027. The AI agent ships around spring 2027 and holds customers more than it brings in new ones. Pro-SPIRE stays a steady, low-margin contributor, and overseas keeps growing from a tiny base without breaking even. The dividend rises a little each year under the progressive policy. The track record in Section 9 (conservative plans, quiet beats) suggests reported results will come in slightly ahead of plans without any break in the trend. The open variable is how the cash pile is used, and whether Kaname Capital publicly presses for an answer.

Bear case. 8.1 and 8.4 compound: Microsoft's AI-inclusive tiers become the default for mid-sized firms and municipalities, and NEOJAPAN's agent arrives late or feels second-rate. Large licences keep drifting away, and new seat growth slows to near zero. 8.2 then bites, because without a further price lever or AppSuite attach gains, growth stalls in the low single digits. 8.3 adds volatility: the on-premise migration pool proves smaller than hoped, many abandoned Cybozu customers go straight to the cloud, and the package business shrinks after the 2027 deadlines while the on-premise code base still has to be maintained. 8.5 makes it worse if management, under pressure to deploy cash, buys a low-quality services or overseas business. 8.6 hangs over all of it: with one person controlling just under half the votes, the strategic response depends entirely on the founder.

Sources:

• logmi Finance: NEOJAPAN Q2 FY1/27 briefing (preliminary)
• logmi Finance: NEOJAPAN Q2 FY1/27 briefing
• logmi Finance: NEOJAPAN FY1/26 briefing
• logmi Finance: NEOJAPAN Q2 FY1/26 briefing
• logmi Finance: NEOJAPAN company page
• Zaikei: H1 FY1/27 results
• Zaikei: Q1 FY1/27 results
• Zaikei: Q3 FY1/26 results and guidance revision
• Zaikei: FY1/27 outlook (Oct 2026)
• media-ir: Q1 FY1/27 results
• The Globe and Mail / TipRanks: Q1 profit
• Wikipedia (JA): ネオジャパン
• NEOJAPAN: shareholder information
• donburi.accountant: FY1/26 securities report officer section
• kitaishihon: NEOJAPAN board
• NEOJAPAN: ISMAP registration
• NEOJAPAN: desknet's NEO cloud price revision
• desknet's NEO official site
• ASCII.jp: Yokohama City adopts desknet's NEO
• Cybozu: package Cybozu Office end of sales and support
• Cybozu: package Garoon end of sales and support
• Deloitte Tohmatsu MIC: collaboration software market 2025
• BOXIL: groupware share survey
• TechnoEdge / PR TIMES: LiveX AI resale partnership
• PR TIMES: ZETA, NEOJAPAN, LIVEX AI alliance
• Nihon M&A Center: Pro-SPIRE acquisition
• Yahoo! Finance Japan: dividend history
• Nikkei: ToSTNeT-3 buyback result notice, 14 Mar 2024
• IRBANK: NEOJAPAN large holding reports
• M&A Online: large holding reports for 3921
• Matsui Securities: Kaname Capital large holding report
• Kabutan: Saito change report No.4
• Yahoo! Finance Japan: Cybozu
• Companies Market Cap: Microsoft
• Macrotrends: Alphabet market cap
• Buffett Code: kubell