DreamArts Corporation Q4 FY2025 earnings call

• 2025 Fiscal Year Overall Result
  • DreamArts achieved both revenue and profit growth in FY2025, with material cost control driving cost ratio improvement. Growth investments in talent recruitment and promotion activities were implemented, and still delivered a 26% YoY increase in operating profit.
• 3-Year Mid-Term Management Plan (2026-2028)
  • Core basic strategy is "Democratization of Digital", aimed at promoting citizen development by leveraging in-house frontline personnel instead of relying on external vendors, to solve the bottleneck of limited IT human resources in Japanese enterprises that need to both maintain existing systems and advance new digital transformation projects. The plan sets 5 critical success factors (CSF):
  1. MCSA (Mission Critical System Aid): Covers peripheral core front-end system areas around enterprise ERP. Multiple projects have already been implemented, and new projects are progressing smoothly.
  2. Global Connect: Aims to support Japanese manufacturing enterprises' overseas operations centered on SmartDB. SmartDB has already been used in 28 countries and regions by clients including Asics and NX Group, and will be actively promoted as part of the mid-term plan.
  3. DAPA (DreamArts Practical AI): Integrates practical AI into all DreamArts products, with the concept of embedding AI into business decision processes. DAPA is currently in pilot use, with a formal product launch scheduled for April 2026. DAPA builds a refined AI prompt database that automatically dispatches optimized prompts to users, allowing large-scale enterprise users to leverage AI without needing to create their own prompts. Vector layer for DAPA is scheduled for release in September 2026, and ontology layer (meaning/ semantic schema ledger) development will start in 2026 for launch in 2027.
  4. PLG (Product-led Growth): Builds growth mechanisms into products to expand user base and increase average customer monthly usage fee, and will enter full-scale development in the second half of 2027.
  5. EC2 (External Capability & Capacity): Integrates external capabilities to drive growth, centered on the SmartDB qualification certification system. There are currently 3,831 certified SmartDB personnel: 79% are client-side frontline personnel, 21% are partners, and 38% are female, which breaks the stereotype that women are less capable in IT and advances citizen development. New AI-related qualification badges will be added after Spring 2026, and the company targets 10,000 certified personnel in 2026. Partner expansion is also progressing, with more large partners expected to join in 2026 and 2027.
• Shareholder Return and Corporate Value Improvement Initiatives
  • Dividend policy shifted to progressive dividend, targeting maintaining or increasing dividends with a 30% payout ratio as benchmark (up from the previous 20-30% target). FY2025 dividend is 20 yen per share.
  • Implemented a 3-for-1 stock split to improve liquidity and expand the investor base.
  • Introduced an employee stock compensation plan to increase employee sense of belonging and align incentives with corporate performance and share price.
• Current CSF Progress
  • MCSA is progressing very actively; EC2 user communities have become more active, with active knowledge sharing among users. Global Connect targets full non-stop operation by the end of 2026, to support 24/7 global access. DAPA pilot projects are ongoing, with vector layer on track for September 2026 release. EC2 qualification growth is on track, with multiple client companies already integrating SmartDB qualification into their personnel evaluation systems.

Full fiscal year 2025 (ended December 2025) consolidated results: overall revenue hit the initial plan, with growth driven by the Cloud Business. On-premises Business and Professional Services Business both outperformed plan, lifting total company revenue. The Horizontal SaaS segment grew 17.2% year-over-year, achieved 62.5% gross profit margin (up 5.4 percentage points YoY), cash flow grew 10.1% YoY, stock revenue ratio reached 87.5% (close to 90%), of which cloud revenue accounts for approximately 80% of total stock revenue. Horizontal SaaS average monthly ARPA reached 1.639 million yen, and net revenue retention (NRR) hit 109.8%. Operating profit grew 26% YoY, as revenue growth and cost reduction absorbed increased selling, general and administrative expenses from growth investments.

• FY2026 (ending December 2026) full year consolidated revenue guidance is 6.25 billion yen. Growth is planned to center on the cloud business: Horizontal SaaS expects steady upsell growth, Vertical SaaS has ongoing national chain introduction projects and other large projects in progress. On-premises business factors in expected churn during the cloud migration promotion process. Professional services plans for a slight revenue increase from implementation and development support projects, consistent with FY2025 performance.
  • Management expects an increase in revenue but a decrease in profit YoY for FY2026, which aligns with the mid-term management plan. As the first year of the 3-year mid-term plan, DreamArts will actively invest in talent recruitment, advertising promotion, and product development to achieve mid-term targets, and expects increased material costs related to AI development.
  • Mid-term plan targets for FY2028: 380 client companies, over 10% market share among approximately 3,700 large enterprises with 1,000+ employees, 8.7 billion yen in total revenue. Operating profit is expected to recover starting in FY2027 and achieve strong growth in FY2028, after the temporary YoY decline in FY2026.
  • Management expects the new digital transformation trend driven by democratization of digital in Japan, which the company has prepared for, may arrive earlier than the original 2029 expectation.

• Market concerns that low-cost, powerful AI tools such as Claude Code and AI agents may pose a disruptive threat to DreamArts' business model. Management notes that specific risks cannot be predicted in advance, as threats may materialize suddenly.
  • If Claude Code and AI agents do not function well or fail to gain widespread adoption, it will limit the incremental revenue opportunity DreamArts expects from integration with these tools.
  • DreamArts' current size is too small relative to the size and volume of opportunities in the large enterprise market, creating a capacity gap that may limit growth if the company cannot scale personnel fast enough.

Q: How do you envision the 2026 business environment, and what traction do you see for your AI functionality?

A: I believe the 2026 business environment is very positive. Amid the recent upward trend of the Nikkei 225, we see widespread momentum for enterprises to actively implement business transformation and adopt new structures and initiatives, which we can see from the daily sales deal reports we receive. The period of widespread AI proof-of-concept (the "AI PoC festival") has concluded, and the market has now entered the phase of advancing practical AI application. Many enterprises have realized that just enabling individual employees to use consumer-facing AI tools like ChatGPT does not directly deliver large improvements to overall organizational capability, so there is growing interest in our DAPA approach. DAPA does not simply add a chat interface to SmartDB or Shop-Ran; instead, it builds an AI prompt database structure that we describe as "automatic transmission for AI utilization", to solve the problem that manually creating prompts leaves large performance gaps between skilled and unskilled users, and creates a barrier to large-scale enterprise AI adoption. This practical approach is seeing rising market demand and expectations.

Q: Over the next 3 to 5 years, what are the worst-case to best-case scenarios for Claude Code adoption by Japanese enterprises?

A: The worst-case scenario is that Claude Code and AI agents do not gain much adoption or do not function well. However, even in this scenario, and in all scenarios, Claude Code and AI agents are ultimately positive for DreamArts. SmartDB holds a critical foundational position in enterprise cloud computing as the owner of the user and permission ledger. For large enterprises, governance is extremely important: it requires strict control and audit logging of which user belongs to which organization, what access and update permissions they have based on organizational hierarchy, and when users viewed or updated data, all of which is managed by SmartDB's ledger. SmartDB can also handle the complexity of multiple concurrent roles that is common in Japanese organizations, which cannot be managed by AI tools like Claude Code alone. Any new application built with Claude Code or any AI agent that accesses enterprise data and participates in decision-making will need to reference and integrate with SmartDB's user and permission ledger, because Claude Code cannot automatically build and manage this ledger on its own. The best-case scenario is that Claude Code becomes widespread, many new applications are built with it, and widespread integration with SmartDB becomes the norm. In this scenario, we can add new billing options for integration with Claude Code and AI agents, which is expected to drive revenue growth.

Q: The market appears to be concerned about the threat from AI technologies from companies like Anthropic, and risks can materialize suddenly. What preparations or business model changes have you made in case low-cost powerful AI tools become widespread?

A: Since we do not know the exact nature of a potential sudden risk in advance, we need to respond when and if it materializes. To prepare, we are strengthening our focus on the critical ledgers that large enterprise computing depends on. As part of DAPA, we plan to add a new ledger: an ontology ledger, which is a map of meaning that formalizes and records what data and information means across an organization to eliminate ambiguity. We plan to start development of this ontology layer (semantic schema) in 2026 and target launch in 2027. Combined with our existing user permission ledger, this creates two complementary core ledgers. We are building three integrated layers for AI: the vector layer (scheduled for September 2026 release) that enables intuitive data access for LLMs, the ontology layer that adds strict semantic control to resolve ambiguity in vector-based access, and the existing process and database layer already built into SmartDB. Together, these three layers cover the complex organizational structures and cross-functional autonomous decision-making processes that are characteristic of Japanese enterprises out of the box, and enable no-code citizen development. This positioning is the most effective preparation we can make against any potential competitive threat.

Q: AI agents could allow you to speed up development with fewer engineers, but you are still actively hiring. Do you have any plans to slow hiring to focus on improving productivity per employee?

A: I fully agree with the core idea that we should improve productivity per employee and slow hiring when it makes sense. However, DreamArts is currently far too small relative to the size of the large enterprise market we serve and the volume of opportunities we are seeing, so we need to scale to an appropriate size. In addition, AI agents are still in the early stage comparable to early autonomous driving: demo videos show strong performance, but there are many practical issues to resolve for real-world deployment, including questions around responsibility, insurance, and regulatory compliance for real-world use. While AI agents will improve individual productivity in many use cases, for work tied to organizational decision-making, you still need formal management of who submitted an item, who approved it, who sent it back with comments, which AI agents cannot do on their own because they lack the underlying user and permission ledger. AI agents are not a threat to us; if they become widely adopted, they will need to integrate with SmartDB, and we can charge AI agents for access, so more AI agents means more billing targets for us. Because of the market size opportunity we face and our current limited capacity, we still need to increase headcount, so we expect employee numbers to continue growing for the foreseeable future.

Q: What role does SmartDB play as a System of Record? Is data stored within SmartDB, or is it just a front-end for data stored in other systems, which could make it easier to replace?

A: Data is stored and retained within SmartDB; as the name suggests, it is a database, which is a core priority for us. It is a System of Record, and it also includes a powerful no-code process engine that is fully integrated with the user and permission ledger we already discussed. For example, a 20,000-person personnel application system built by frontline users via no-code citizen development runs on the SmartDB process engine, and SmartDB stores all transaction records, so it functions as an MCP server. When AI agents want to access enterprise data, they send a request to SmartDB, which verifies the AI agent's identity and checks the permissions tied to its owner before providing access, so we sit in a foundational controlling position. Additionally, data stored in SmartDB is structured and vectorized in an AI-ready format, comparable to pre-milled rice ready for cooking, unlike data stored in general data lakes which is not prepared for effective AI utilization.

Q: 2026 is the first year of the mid-term management plan. Which of your CSF strategies are you seeing good traction for, and what progress have you made?

A: MCSA is progressing very actively. For EC2, user communities have become very active; we recently held a community event with 3 existing users of SmartDB for ERP front workflows and 1 upcoming user, and there was very engaged discussion from afternoon into the evening, with new insights generated from knowledge sharing among users. We aim to provide Best Practice as a Service, not just SaaS, which is core to delivering enterprise value through software. Global Connect is an initiative that started over 10 years ago; enabling fully non-stop operation with no maintenance outages is required for broader adoption, and we plan to achieve this within 2026, which will greatly accelerate promotion of this initiative. DAPA has multiple pilot projects ongoing, with architecture and design work progressing well, and the vector layer is on track for September 2026 release, so progress is on schedule. PLG started with foundational work and will go full force in the second half of 2027, but we expect it to deliver meaningful impact. For EC2, we are targeting 10,000 certified SmartDB personnel in 2026. We already have some clients approaching 200 certified personnel, 90% of which are frontline employees not from IT departments, which means a large enterprise of 20,000 total employees can have 200 people capable of building enterprise system applications in-house, which is unprecedented. Multiple clients have already added SmartDB certification to their personnel evaluation frameworks, some provide financial support for certification, and more are using it for promotion decisions, which originated from client requests for us to build this system. EC2 is also progressing very well.