Amber International Holding Ltd Q2 FY2026 earnings call

Strategic Pivot to AI Agents:

• AMBR is transitioning from a digital wealth management business to a technology company focused on building specialized AI agents.
• CEO Michael Wu emphasized this is a deliberate pivot from strength, backed by positive Q2 financials, rather than continuity.
• The strategy involves operating existing businesses, converting their expertise into AI agents, and then taking those agents to market.

Product Launches and Validation:

• Two agents are currently in market: Embry (personal finance agent) and MIA (marketing operations agent).
• Embry works across users' existing brokerage accounts without requiring asset migration and does not place autonomous orders yet; actions are handled by expert human teams until reliability is proven.
• MIA runs day-to-day campaign operations for over 100 enterprise customers, both embedded in iClick/CMRS services and sold directly as a product.
• AMM (Agentic Marketing/Middle Office?) contributed $3.5 million in its first quarter of recognition, demonstrating that the underlying infrastructure generates economic value immediately.

Operational Philosophy and Moat:

• The company defines an AI Agent as Model + Harness + Purpose. While models and harnesses can be open-source or commoditized, AMBR’s proprietary advantage lies in its deep understanding of 'Purpose'—the specific context, workflow, and customer needs built over years of servicing high-stakes clients.
• They do not compete at the foundation model layer but remain model-neutral, leveraging whatever intelligence best fits the vertical.
• The 'Da Lu' operating model requires internal usage of their own agents to identify flaws and refine workflows before external release.
• Legacy financial infrastructure, regulatory experience, and domain expertise are viewed as critical moats that new entrants cannot easily replicate.

Financial Discipline:

• Operating expenses remained flat at approximately $10 million despite significant revenue growth, demonstrating early operating leverage.
• The company moved from an operating loss of $3.2 million in Q1 to an operating income of $1 million in Q2 without increasing headcount or spending.

Future Roadmap:

• An inaugural investor day is expected before year-end to showcase the broader agent portfolio and transition framework.
• The company is reviewing the structure of legacy financial services; some will remain as infrastructure, while others may be restructured or divested.

Risk and Operational Failures Discussion:

• Management acknowledged that Embry and MIA are early versions and expect significant evolution based on user behavior validation.
• There is no announcement regarding specific operational failures, but the company emphasizes a cautious approach to expanding agent authority, ensuring human oversight remains until reliability is demonstrated.
• The review of legacy businesses introduces uncertainty regarding future structural outcomes, though no specific risks were detailed beyond general strategic transition risks.

Guidance": "Withdrawal of Previous Guidance:

• CFO Josephine Ngai announced the withdrawal of previously issued financial guidance.
• The rationale is that the prior guidance was based on the old business model (digital wealth management) and is no longer an appropriate framework for evaluating the company during its strategic transition to an agentic AI company.
• Management stated they intend to provide updated guidance once they have sufficient operating history and visibility into the financial contributions of the new agentic businesses.
• No new forward-looking numerical guidance (revenue, margins, etc.) was provided for Q3 or full-year 2026.

Qualitative Outlook:

• Management expressed confidence in the long-term opportunity and highlighted the pattern of revenue growth, flat expenses, and expanding margins as the desired trajectory.
• They expect the agentic revenue share to become the center of the company's financial framework over time.

Risks": ["Strategic Transition Risk: The pivot from a proven wealth management business to a new AI agent model carries execution risk. Success depends on the ability to successfully convert legacy expertise into scalable, reliable agents.", "Market Adoption Uncertainty: While early response to Embry and MIA has been encouraging, these are early-stage products. Customer adoption rates and willingness to pay for standalone agents versus bundled services are still being validated.", "Competitive Landscape: Although management argues competitors lack the 'purpose' and 'harness' depth, the AI agent space is rapidly evolving. Large tech firms and model labs could potentially encroach on vertical applications if they choose to prioritize them.", "Regulatory and Compliance Risk: As agents gain more autonomy in high-stakes areas like finance and marketing, regulatory scrutiny may increase. The current model relies on human-in-the-loop controls which must be maintained as automation scales.", "Legacy Business Restructuring: The ongoing review of the legacy financial services portfolio could lead to restructuring costs, divestitures, or changes in revenue streams that might impact short-term financial stability."], "Q&A_Section": "Q: What is proprietary about AMBR's AI agents compared to competitors using open-source models? / A: CEO Michael Wu defined an AI agent as Model + Harness + Purpose. He argued that while models and harnesses are increasingly open-source and commoditized, AMBR's moat lies in its proprietary 'Purpose'—the deep, years-long understanding of specific client workflows, demands, and service delivery methods in wealth management and marketing. Additionally, their harnesses are battle-tested systems built internally, providing a unique value proposition that generic model labs cannot replicate without extensive industry experience.\n\nQ: If investors gave your five largest competitors $50 million tomorrow, what stops them from building Embry and MIA? / A: Wu explained that companies capable of building these agents (large model labs) do not need the money and are focused on winning the 'model war,' not the vertical agent war. Hypothetical competitors lacking this capability would fail because they do not understand the specific 'purpose' or possess the necessary proprietary harnesses. Furthermore, adding an AI bot to existing non-agile services results in the wrong purpose, whereas AMBR’s agents are designed from the ground up around verified customer workflows.\n\nQ: Do you expect to take shares from existing competitors or view the segment as an open and untapped market? / A: Wu described it as both, depending on the horizon. In the near term, MIA takes share from competitors like iClick by offering superior, 24/7 automated service. However, over time, the model opens a much larger untapped market by making high-touch services affordable to smaller customers who previously couldn't afford such levels of attention. Similarly, Embry democratizes private banking-level insights for retail investors, expanding the addressable market beyond just stealing share from incumbent firms."}

Total revenue for Q2 2026 was $13.9 million, a 39% increase from Q1's $10 million. The company reorganized its reporting into two primary segments:

1. Digital Assets Platform Revenue: $6.6 million (47.5% of total revenue). This includes wealth management solutions ($5.3 million), execution services, and payment solutions. Wealth management specifically grew to $5.3 million from $4.3 million in Q1, driven by stronger demand for diversified and newly launched products.

2. Agentic Revenue: $7.4 million (53.2% of total revenue). This segment reflects the initial contribution from AMM ($3.5 million in its first quarter of recognition) and marketing/enterprise solutions ($3.8 million). Combined with the digital assets platform, agentic-related revenue reached $10.1 million, slightly exceeding the previous outlook of $9-10 million.

Gross margin expanded significantly to 79.5% from 67.7% in Q1, primarily due to the mix shift toward higher-margin agentic activities and core wealth management.

Withdrawal of Previous Guidance:

• CFO Josephine Ngai announced the withdrawal of previously issued financial guidance.
• The rationale is that the prior guidance was based on the old business model (digital wealth management) and is no longer an appropriate framework for evaluating the company during its strategic transition to an agentic AI company.
• Management stated they intend to provide updated guidance once they have sufficient operating history and visibility into the financial contributions of the new agentic businesses.
• No new forward-looking numerical guidance (revenue, margins, etc.) was provided for Q3 or full-year 2026.

Qualitative Outlook:

• Management expressed confidence in the long-term opportunity and highlighted the pattern of revenue growth, flat expenses, and expanding margins as the desired trajectory.
• They expect the agentic revenue share to become the center of the company's financial framework over time.

• Strategic Transition Risk: The pivot from a proven wealth management business to a new AI agent model carries execution risk. Success depends on the ability to successfully convert legacy expertise into scalable, reliable agents.
• Market Adoption Uncertainty: While early response to Embry and MIA has been encouraging, these are early-stage products. Customer adoption rates and willingness to pay for standalone agents versus bundled services are still being validated.
• Competitive Landscape: Although management argues competitors lack the 'purpose' and 'harness' depth, the AI agent space is rapidly evolving. Large tech firms and model labs could potentially encroach on vertical applications if they choose to prioritize them.
• Regulatory and Compliance Risk: As agents gain more autonomy in high-stakes areas like finance and marketing, regulatory scrutiny may increase. The current model relies on human-in-the-loop controls which must be maintained as automation scales.
• Legacy Business Restructuring: The ongoing review of the legacy financial services portfolio could lead to restructuring costs, divestitures, or changes in revenue streams that might impact short-term financial stability.

Q: What is proprietary about AMBR's AI agents compared to competitors using open-source models? / A: CEO Michael Wu defined an AI agent as Model + Harness + Purpose. He argued that while models and harnesses are increasingly open-source and commoditized, AMBR's moat lies in its proprietary 'Purpose'—the deep, years-long understanding of specific client workflows, demands, and service delivery methods in wealth management and marketing. Additionally, their harnesses are battle-tested systems built internally, providing a unique value proposition that generic model labs cannot replicate without extensive industry experience.

Q: If investors gave your five largest competitors $50 million tomorrow, what stops them from building Embry and MIA? / A: Wu explained that companies capable of building these agents (large model labs) do not need the money and are focused on winning the 'model war,' not the vertical agent war. Hypothetical competitors lacking this capability would fail because they do not understand the specific 'purpose' or possess the necessary proprietary harnesses. Furthermore, adding an AI bot to existing non-agile services results in the wrong purpose, whereas AMBR’s agents are designed from the ground up around verified customer workflows.

Q: Do you expect to take shares from existing competitors or view the segment as an open and untapped market? / A: Wu described it as both, depending on the horizon. In the near term, MIA takes share from competitors like iClick by offering superior, 24/7 automated service. However, over time, the model opens a much larger untapped market by making high-touch services affordable to smaller customers who previously couldn't afford such levels of attention. Similarly, Embry democratizes private banking-level insights for retail investors, expanding the addressable market beyond just stealing share from incumbent firms.