Digital Media Professionals Inc. Q4 FY2025 earnings call

Company Overview & Core Strengths

• Founded in 2002 as a university spin-out venture, with graphics technology as its core competency; historical milestones include GPU IP adoption in Nintendo game consoles, with 2D/3D integrated graphics LSIs for the amusement market currently forming the core revenue base.
• Core strength is an end-to-end development capability spanning algorithms, software, hardware, from edge to cloud, focused on solving high-priority customer and societal problems leveraging AI/Deep Learning, an area the company has expanded into in recent years.

Fiscal 2025 Financial Summary

• Total revenue hit 3.077 billion yen, up 2% year-over-year, marking the 4th consecutive year of record-high revenue.
• Operating profit fell to 265 million yen (down 63 million yen year-over-year) and ordinary profit fell to 271 million yen (down 58 million yen year-over-year) due to increased R&D spending, including 79 million yen for next-generation edge AI semiconductor development.
• Net profit attributable to parent shareholders fell to 157 million yen (down 174 million yen year-over-year) after a 42 million yen investment securities valuation loss and a 45 million yen deferred tax asset write-down.
• The company maintains a strong balance sheet with total assets of 4.092 billion yen, an equity ratio of 88.2%, and sufficient capital for R&D investment and working capital.

Next-Generation Edge AI Semiconductor Di1 Update

• Development is progressing on schedule; samples are currently available, with mass production shipments scheduled to begin in Japan and Taiwan (the first launch markets) in Q4 of the current fiscal year (January-March 2026 calendar), followed by sequential global expansion. iCatch Technology will sell the chip under the V9 product name in Taiwan.
• Key technological differentiators: The 15mm square (1g weight) all-in-one chip integrates DMP's latest AI processor, 4K HDR ISP, real-time 3D ranging engine, security engine, and a lightweight vector GPU for advanced HMI. It supports INT4/INT8 integer computation plus FP4/FP8/FP16 floating-point computation, and is the world's first edge AI inference chip to implement FP4, making it fully compatible with models trained on NVIDIA's Blackwell architecture.
• It supports 8-channel camera input and 4-pair on-chip hardware stereo vision processing, enabling 360° real-time obstacle detection for drones at low power and weight, a major competitive advantage.
• Key target use cases: Automotive ADAS/driver monitoring (compatible with DMP's existing ZIA SAFE software stack), autonomous mobile robots (compatible with DMP's ZIA MOVE development platform), smart factory picking systems (compatible with DMP's Cambrian system), drones, security cameras, and kiosks.

Robotics & Factory Automation (FA) Business Updates

• DMP has signed agency agreements with 3 major Chinese robotics firms: SEER ROBOTICS (AMR/AGV/AGF controllers, components, and fleet management software), Kinco (low-voltage DC servo motors), and Hinson (AMR/AGV LiDAR sensors). The company aims to become a total AMR solutions provider serving two customer groups: AMR/AGF developers (to which it provides key components) and e-commerce/distribution customers (to which it provides complete units). The Japanese logistics automation market is projected to grow at a 20% CAGR to 123 billion yen by 2030.
• DMP's ZIA MOVE visual SLAM autonomous navigation software has been adopted in GE Creative's food factory AMR lines (YL series heavy pallet lift AMRs and YT series towing AMRs), the first industry implementation of automatic transport between elevators and cold storage facilities, and delivers robust performance in challenging environments like sloped floors, obstacles, and low-temperature high-humidity conditions where LiDAR performs poorly. Commercial order intake began May 14, 2025.
• DMP offers a tiered solution portfolio: custom high-engineering solutions (including Di1, development kits, and professional services for challenging environments like food factories), commodity components from partner brands, and packaged solutions for specific use cases (like the Cambrian picking system), to differentiate from other robotics vendors.

Mid-Term Vision and Growth Strategy

• Targets: 5 billion yen revenue by FY2028 March term, 8 billion yen revenue by FY2030 March term.
• Three-pronged "Trinity Growth Strategy":
  1. Further growth of the core business: Increase added value by expanding adjacent businesses in the amusement industry, and strengthen profitability through product cost reduction.
  2. Expansion into growth areas: Leverage existing AI expertise and track record to launch the Di1 next-generation edge AI chip business, and expand into mobility, smart factory, drone, and smart camera segments.
  3. Capture new business opportunities: Leverage existing FA industry networks from camera system businesses to contribute to further smartification of robotics, factory automation, and distribution automation.

By business segment:

1. IP Core Licensing Business: Revenue of 124 million yen (4% of total revenue), down 27% year-over-year. The decline was driven by the expiration of a large GPU IP maintenance and support project from the prior year, partially offset by new GPU IP licensing revenue, recurring revenue from the safety segment, and running royalties for digital device GPUs.
2. Product Business: Revenue of 2.855 billion yen (92.8% of total revenue), up 4% year-over-year. Growth was driven by strong sales growth of the mass-produced graphics semiconductor RS1 for the amusement market.
3. Professional Services Business: Revenue of 97 million yen (3.2% of total revenue), up 12% year-over-year. Growth came from increased contracted AI development revenue for the broader safety segment, and services for autonomous mobile robots, semiconductor manufacturing equipment, and construction machinery in the robotics segment.

By application segment:

1. Safety Segment: Revenue of 38 million yen, down 46% year-over-year, due to fewer new drive recorder-related projects despite stable recurring revenue and expanded professional services.
2. Robotics Segment: Revenue of 168 million yen, flat year-over-year. The loss of drone camera module revenue was offset by strong sales of the Cambrian vision system (including full in-line adoption at an automaker) and increased professional services revenue.
3. Amusement Segment: Revenue of 2.779 billion yen, up 5% year-over-year, driven by solid mass production shipments of the RS1 graphics semiconductor.
4. Other Segments: Revenue of 90 million yen, down 32% year-over-year, due to the expiration of the prior year's large GPU IP maintenance project.

• For FY2026 March term, DMP forecasts total revenue of 3.25 billion yen, up 6% year-over-year. Growth will be driven by continued solid RS1 shipments for the amusement market, plus limited initial revenue contribution from the new Di1 edge AI semiconductor and FA businesses.
• Management prioritizes aggressive strategic growth investment, so temporary profit decline is expected: the company forecasts operating profit of 20 million yen, ordinary profit of 25 million yen, and net profit attributable to parent shareholders of 20 million yen. Total planned development spending for Di1 is 1.9 million USD in FY2026.
• The company confirms Di1 will begin mass commercial sales in the FY2026 March term, consistent with prior timelines.

No explicit material operational risks or failures were discussed in the available transcript. The primary uncertainty noted is that new businesses (Di1 edge AI semiconductors and robotics FA) are in early stages, with only limited revenue contribution expected in the near term as the company invests to build out the market.

Q: What specific adoption use cases does management expect for Di1 in mobility, smart factories, drones, and smart cameras? What competitive advantages does DMP hold in these segments? / A: For smart factories, Di1 can be integrated into AMR/AGV robots to enable advanced edge AI functionality, and can be combined with DMP's existing Cambrian picking system to enable more sophisticated picking performance that leverages the chip's capabilities. For drones, the biggest industry challenge is short flight time driven by weight and power consumption. Competitor solutions often use large, high-power NVIDIA GPUs that weigh tens of grams and consume 10-30W, cutting flight time short. Di1's 1g weight, low power consumption, and ability to handle performance-heavy stereo vision processing on-chip makes it a far more competitive option, which DMP expects to be a major differentiator with customers. The Di1 is a good fit for smart camera use cases requiring edge AI processing, as well as automotive mobility ADAS applications leveraging DMP's existing ZIA SAFE software stack.