Digital Media Professionals Inc. Q3 FY2026 earnings call

• Company Overview & Core Differentiation
  • DMP was founded in 2002, is the only Japanese company with proprietary GPU technology, and focuses on low-power embedded GPUs and edge AI inference for automotive, industrial, and consumer devices, in contrast to NVIDIA's focus on data center training and high-end PC graphics.
  • Core competitive advantage: End-to-end integrated development across algorithms, software, and hardware, with domain-specific optimization matching customer application requirements for power, performance, and cost, a capability rare among Japanese semiconductor firms. This enables faster convergence from customer requirement to mass production compared to multi-vendor development models.
  • The firm operates 2 R&D centers (Tokyo, Japan and Ho Chi Minh City, Vietnam) with approximately 65 total employees.
• Product Segment Highlights
  • RS1 Amusement SoC: Industry-first real-time 3D graphics for amusement machines, optimized for the industry's unique high-performance video processing requirements, delivering cost reduction for customers while growing DMP's market share.
  • Cambrian Vision System: Uses physical AI and digital twin simulation pre-training to enable robust operation even with variable ambient light, and is the only system globally capable of automated picking of micro parts (<1mm), glossy objects, and transparent objects, addressing long-standing unmet needs in automotive and logistics manufacturing.
  • A3000V2 AI IP: Is the first edge AI processor globally to implement NVIDIA's new FP4 technology, delivering improved performance and reduced memory usage for generative AI models, and supports common AI frameworks including TensorFlow and PyTorch.

The provided transcript does not include detailed absolute financial figures or revenue contribution percentages for individual segments. It confirms that the firm's business is structured across four core product/segment areas:

1. Fabless Semiconductor (Amusement Segment): RS1, a real-time 3D graphics SoC for pachinko/pachislot amusement machines. The segment is a stable mature revenue generator for DMP, with growing market share across top industry players including Sammy, Universal Entertainment, and their joint venture Zeal, serving the annual 1.5 million unit Japanese amusement machine market.
2. Robotics Segment: Cambrian Vision System, an AI-powered robot picking system sold exclusively in Japan by DMP as the general agent for UK-based Cambrian Innovation. It is currently gaining traction with major Japanese automotive manufacturers and consumer goods players such as Kao.
3. Semiconductor IP Licensing: Three core IP products are available for licensing: A3000V2 AI inference processor, legacy GPU IP, and stereo vision camera IP. Cumulative shipments of customer devices using DMP IP total approximately 200 million units, with deployments in products including Nintendo 3DS, most Japanese digital camera models, and high-end Hisense/TVS REGZA televisions.
4. New Segment (In Development): Next-generation DMP edge AI inference chip, currently under development with mass production scheduled to start in early 2026.

• Mass production of DMP's next-generation edge AI inference chip is scheduled to begin in early 2026 (Q4 FY2026), with full-year revenue contribution starting from FY2027.
  • DMP expects the new edge AI chip to become the firm's long-term growth platform following the RS1 amusement chip, serving a projected global market size of over 1 trillion yen by 2032.
  • Management expects no profitability from the new chip segment in its first full year (FY2027), with a clear path to profitability expected by the second full year (FY2028), with revenue growing year-over-year through 2027 and 2028.
  • The RS1 amusement chip segment will be maintained as a stable, long-term revenue and profit base to support new growth initiatives.
  • DMP will expand its robotics solutions business beyond the Cambrian Vision System, adding synergistic complementary robotic products to deliver full solutions to Japanese customers.

• The edge AI semiconductor market evolves extremely rapidly, with new AI architectures and technologies released every 6-12 months, requiring continuous fast catch-up to maintain competitiveness.
  • Large global players including Apple, Qualcomm, Intel, and NVIDIA are increasingly entering the edge AI space, increasing competitive pressure on smaller focused players.
  • Semiconductor development requires high upfront investment, with historically low success rates for new chip projects, though DMP's 20 years of development experience reduces this risk relative to new entrants.
  • DMP's size means it has limited engineering headcount, requiring close partnership with Taiwan's iCatch Technology to scale development capacity and speed.
  • The firm must prioritize R&D investment for new growth initiatives in the near term, pushing out potential shareholder dividend distributions to a later date.

Q: Can you explain why edge AI rather than cloud AI is the right market for DMP, and why DMP's product is needed instead of large NVIDIA chips? / A: Around 40% of global data is image/video data from cameras and mobile devices, and 98% of this data is currently discarded because it is too expensive to transmit to the cloud for processing. Edge AI can process this data locally on-device. Large data center GPUs from NVIDIA consume too much power and generate too much heat for small embedded devices like surveillance cameras, which often run on battery and operate in high ambient temperatures. Low-power, high-performance small chips like DMP's are required for these use cases.

Q: What is DMP's competitive position relative to global competitors in the edge AI market? / A: While overseas vendors from the US, Israel, and other regions have entered the market, DMP holds a strong advantage in the Japanese market because it works closely with customers through all stages of AI model development and tuning. DMP is also differentiated by its partnership with iCatch Technology, which provides world-leading camera peripheral technology, enabling high-precision AI based on high-quality image input. DMP's ability to develop new chips quickly to match fast-changing AI trends is also a core competitive advantage, as the edge AI market evolves very rapidly.

Q: With R&D spending of 530 million yen on the new edge AI chip, how does DMP expect to recoup this investment, and what is the revenue timeline? / A: The new chip will start volume production in early 2026, so it will only contribute one quarter of revenue in FY2026, with full-year contribution starting in FY2027. DMP plans to scale revenue rapidly through expanding application segments and global expansion via its partnership with iCatch. Profitability is not expected in the first full year, but a clear path to profit should emerge in the second year. Customer acquisition will run in parallel with product development rather than waiting for full product launch to speed up revenue growth.

Q: How does DMP plan to differentiate itself from large global players like Apple and NVIDIA in this market? / A: Large global players mostly focus on mass-produced general-purpose chips, while DMP's core strength is domain-specific optimization for particular customer applications (called domain-specific LSI). This trend toward domain-specific chips is accelerating globally even at big technology firms like Google and Meta, so DMP's focused business model aligns with broader industry movements, creating a clear differentiated niche for the firm.