AMBA

• AGI and Physical AI Strategy: Management emphasizes leadership in Application-Generated Intelligence (AGI) and is expanding into a full-stack physical AI platform provider. The company is shifting focus toward operational efficiency at the edge, utilizing open-weight and distilled models on-premise to reduce latency, protect data, and lower bandwidth costs.
• New Product Launches: Introduction of the X7 standalone AI accelerator, currently sampling, designed to serve as an AI coprocessor for host processors like ARM or x86. This complements existing AI SoCs (CB75, CB72, N1 series) and targets the edge infrastructure market.
• Market Expansion (SAM): The five-year Serviceable Addressable Market (SAM) forecast has been revised upward from $8.5 billion (FY27) to $22.9 billion (FY32), reflecting a ~20% CAGR. IoT markets now represent ~70% of the terminal year SAM, driven by edge infrastructure and robotics.
• Go-to-Market Partnerships: Two major indirect channel partnerships announced: Capgemini (focus on large enterprise deployment and system integration) and Macnica (focus on distribution to small/mid-sized customers and ISV ecosystem). These aim to drive significant incremental revenue over seven years.
• Semi-Custom Strategy: First semi-custom project underway (2nm CV4 SoC), with first production revenue expected in FY2028. This strategy allows customization while retaining IP to sell to non-competing customers.
• Design Wins: Notable wins include Tier 1 automotive suppliers (Audi, VW) for in-cabin monitoring, Canon/Suprema/IDIS for AI monitoring, Moultrie for trail cameras, and a quadruped robot using CB72 for robotics validation.

• Q3 FY2027 Revenue: Expected range of $115 million to $124 million, with a midpoint of $119.5 million, anticipating favorable seasonality and growth led by AGI demand in IoT.
• Q3 FY2027 Gross Margin: Non-GAAP gross margin expected between 59% and 60%.
• Q3 FY2027 Operating Expenses: Non-GAAP operating expenses expected between $56.5 million and $59.5 million.
• Full Year Outlook: Management maintains confidence in the long-term growth trajectory but notes uncertainty regarding Q4 due to memory supply constraints. The full-year revenue growth target of 10-15% remains under review based on Q4 visibility.

Total revenue for Q2 FY2027 was $108.1 million, representing a 7.7% sequential increase and a 13.2% year-over-year growth, slightly exceeding the guidance midpoint. Automotive revenue set a new record, driven by strong commercial vehicle adoption of AI, and slightly outpaced IoT business growth. Enterprise-driven businesses within IoT outperformed consumer-led segments. Specific product segment financials beyond total revenue and general market split were not itemized in absolute terms.

• Memory Supply Chain Constraints: Surging memory prices and scarcity are impacting the entire industry. While Umbrella does not buy/resell memory (thus no direct gross margin impact), high memory costs may reduce customer volume/orders. Q4 performance depends on securing sufficient memory for customers.
• Execution Risk in New Channels: The transition to indirect sales channels (Capgemini/Macnica) involves a ramp-up period; meaningful revenue is expected in 2-3 years, creating execution risk during the initial phase.
• Competitive Landscape: The edge AI accelerator market is crowded with competitors including NVIDIA, Qualcomm, and ~50 startups. Success depends on delivering superior power efficiency and mature software/hardware integration.

Q: Analyst asked about the X7 accelerator's role relative to existing portfolios and its competitive landscape.

A: CEO clarified X7 is a standalone AI accelerator that can bundle with any host CPU (ARM/x86) or own SoCs, allowing performance upgrades without board redesign. It competes in a crowded space against GPUs and specialty chips, but Umbrella differentiates via superior power efficiency (4-5W envelope) and low memory footprint (4MB), addressing critical cost and efficiency needs in edge infrastructure.

Q: Analyst inquired about the rationale behind the two distinct indirect channel partnerships (Capgemini vs. Macnica) and their revenue modeling.

A: Management explained these are complementary strategies. Macnica focuses on a 'volume play' aggregating small/mid-sized fragmented customers previously unserved by direct sales. Capgemini focuses on a 'value play' targeting large enterprises for complex, scalable deployments. Both utilize the Cooper software platform, with modeling projecting half-billion-dollar potential each over seven years, though timelines differ.

Q: Analyst questioned the impact of memory shortages on gross margins and future demand pull-forwards.

A: CFO and CEO stated memory costs do not impact Umbrella's gross margin directly as they do not hold inventory of memory. However, high memory costs could suppress customer device volumes, indirectly affecting chip orders. Management sees little revenue impact in Q2/Q3 but remains cautious on Q4 visibility pending customer feedback on securing memory supplies.

Q: Analyst asked about the robotics market traction and whether it drives the SAM increase.

A: CEO confirmed robotics is a significant driver of the SAM expansion, citing design wins worth ~$100M and additional pipeline additions. He noted the robotics market mirrors the autonomous driving sector from seven years ago: early stage focused on modular perception solutions, with a roadmap toward integrated domain controllers. Umbrella offers both, supporting data collection and auto-labeling services to aid customer training.

Q: Analyst asked if the long-term gross margin target would change due to the new partnership structures.

A: CEO affirmed the long-term non-GAAP gross margin target remains 59%-62%. Although the indirect channel is in early stages, management sees no immediate need to adjust the target. They will monitor the ramp-up closely and inform investors if structural changes to margins become necessary as these partnerships scale.