Sagoh
The artificial intelligence revolution, while electrifying markets and sparking unprecedented investment in generative models, is poised for a significant evolution. As the initial frenzy for raw AI capabilities begins to mature, a new frontier of opportunity is emerging: the infrastructure and technologies that ensure AI is not only powerful but also reliable, secure, and controllable. This shift, dubbed "AI 2.0," represents a critical pivot for investors seeking to harness the systemic risks and demands inherent in the widespread adoption of intelligent systems.
The Dawn of AI 2.0: From Intelligence Production to Trustworthy Deployment
The initial wave of AI investment, characterized by a relentless pursuit of more capable models and greater computational power, has been nothing short of transformative. In 2025 alone, AI companies commanded a staggering $258.7 billion, representing 61% of all global venture capital, according to OECD data. This "AI 1.0" era focused on the sheer production of intelligence, pushing the boundaries of model performance and efficiency. However, as the landscape of model choices expands – from closed-source behemoths to open-weight and specialized alternatives – the next significant bottleneck is not in creating intelligence, but in its widespread, trustworthy, and scalable adoption.
This is the core of the "AI 2.0" paradigm. It signifies a move beyond simply generating outputs to building the robust infrastructure that governs how AI models, agents, data, and users interact. This encompasses critical elements such as who has access to what, how systems interoperate, how behavior is evaluated, and, crucially, who ultimately retains control. This evolution is driven by a growing recognition that as AI systems become more autonomous, the value of choice, control, and agency across the entire AI stack escalates dramatically. Investors are increasingly advised to look beyond the headline excitement and follow the durable demand signals that are shaping this new phase.
The Imperative for Trust: Demand Signals from Enterprises, Governments, and Users
The demand for AI within enterprises is palpable, yet deep integration remains a significant challenge. A recent study by MIT researchers revealed that by the end of 2025, a mere 11% of S&P 500 companies had AI deeply embedded within their operations. The barriers to adoption are multifaceted, extending beyond mere model performance. Concerns surrounding proprietary data security, vendor lock-in, system reliability, the selection of appropriate models, and the permissible actions of increasingly autonomous AI systems are paramount. These are not abstract ethical debates; they are practical impediments to realizing the full potential of AI in the business world.
Governments, too, are grappling with these complex issues. At the World Economic Forum in Davos, Canadian Prime Minister Mark Carney articulated a compelling vision for "middle powers" seeking greater strategic autonomy and resilience in the face of global technological shifts. His call for AI cooperation, aimed at preventing nations from being forced to choose between dominant "hegemons and hyper-scalers," highlights a growing procurement demand for AI solutions that prioritize openness, interoperability, trusted infrastructure, and genuine optionality. This signifies a clear signal for technologies that empower strategic independence rather than fostering dependency.
The public’s sentiment towards AI is also a critical demand signal. A June 2026 survey by the Pew Research Center indicated that a significant majority of Americans (only 9% expressed more excitement than concern) harbor reservations about the increasing integration of AI into their lives. This wariness underscores the urgent need for AI systems that can build trust and demonstrate tangible benefits without eroding user agency or security.
Collectively, these disparate groups are sending a unified message: trust, agency, interoperability, and control are transitioning from aspirational principles to essential practical requirements. As these requirements begin to shape procurement decisions and product design, they are poised to create entirely new markets.
Mapping the Landscape: The Seven Layers of the Trustworthy AI Stack
To understand where these new markets are emerging, a comprehensive view of the evolving AI ecosystem is necessary. The "Trustworthy AI stack" framework, as outlined by a pioneering venture map, identifies seven overlapping layers that extend far beyond the foundational models themselves.
-
Identity and Authentication: This foundational layer is crucial for establishing the legitimacy of any entity interacting with an AI system. It ensures that both users and AI agents are who they claim to be, a critical step in preventing unauthorized access and malicious activity.
-
Data Agency and Memory: This layer addresses the complex challenges of how data is accessed, shared, remembered, and utilized by AI systems. It encompasses principles of data privacy, consent management, and the ethical retention and retrieval of information, empowering users with greater control over their digital footprint.
-
Orchestration and Interoperability: As AI systems become more complex, this layer focuses on the seamless connection of models, agents, tools, and data sources. It enables disparate AI components to work together harmoniously, facilitating efficient workflows and the creation of more sophisticated AI applications.
-
Models: While the initial investment frenzy focused here, models remain a critical component. However, the emphasis is shifting towards how different model types – closed, open, open-weight, and specialized – can be integrated and differentiated based on their controllability and suitability for specific trustworthy applications.
-
Evaluations: This layer is dedicated to the rigorous testing of AI behavior, performance, and potential risks. It encompasses methods for assessing fairness, robustness, explainability, and the identification of unintended consequences, ensuring that AI systems perform as intended and ethically.
-
Guardrails: Real-time monitoring and constraint mechanisms form the guardrails of the AI stack. These systems act as an intelligent oversight, detecting and mitigating problematic behavior or potential threats as they arise, providing an essential layer of safety.
-
Applications: The ultimate manifestation of trustworthy AI lies in its applications. This layer focuses on how these underlying capabilities are integrated into user-facing products and services, where trust, agency, access control, and verifiability become paramount for user adoption and satisfaction.
It is important to note that not every layer will necessarily mature into a distinct, standalone market. Some categories may converge, or established players might integrate certain functions. However, this layered approach provides a crucial lens through which to identify the emerging enabling infrastructure that surrounds and supports the foundational model layer.
Testing the Thesis: Entrepreneurial Activity and Capital Clustering
The true validation of any investment thesis lies in observing whether entrepreneurs are actively building solutions based on its premise. The inaugural "Trustworthy AI Market Map," a first-of-its-kind venture map, aims to do precisely this by identifying a new generation of venture-backed companies dedicated to building the crucial infrastructure that bridges models, agents, data, and users. This map offers an early glimpse into the nascent opportunities emerging from the systemic risks and demands associated with AI deployment.
Co-initiated by Paul Fehlinger, a contributing editor at ImpactAlpha, and Douglas Sloan, the Trustworthy AI Map initiative seeks to empower asset owners and allocators to move beyond merely managing AI risk. The goal is to identify where these risks are actively generating new demand, creating investable markets, and fostering opportunities for systemic capital allocation. The initiative has garnered significant support, with co-convenors including ImpactVC, Andaris, Omidyar Network, ReframeVenture, Mozilla, Norrsken, Seismic Foundation, and Kapor Foundation.
Nineteen venture capital partners, spanning a diverse range of investment philosophies, have contributed nominations, including Benhamou Global Ventures, Bethnal Green Ventures, Eka Ventures, EQT Ventures, ex/ante, Founder Collective, Generation Investment Management, Giant Ventures, Halcyon Ventures, HV Capital, Juniper Ventures, Kapor Foundation, Katapult, Mozilla Ventures, Norrsken VC, Omidyar Network, Revent, SAIF, and Sympatico Ventures. ImpactAlpha serves as the media partner for this significant undertaking.
The initial mapping effort has identified 79 companies across 15 countries, ranging from pre-seed to Series D+ funding stages. Collectively, these companies have raised over $13 billion, with more than 400 investors backing ventures featured on the map. While this funding does not definitively equate to a mature market, it offers invaluable insights into the current clustering of entrepreneurial activity and capital.
Key Observations from the Trustworthy AI Market Map:
Three prominent trends emerge from this comprehensive mapping exercise:
1. Innovation Beyond the Model Layer: A striking observation is that only 17 of the 79 featured companies are directly involved with AI models. The vast majority of activity is concentrated in the adoption infrastructure and the surrounding ecosystem. This includes companies focusing on identity and authentication, data agency and memory, orchestration and interoperability, evaluations, and guardrails. This signifies a clear shift in entrepreneurial focus from foundational model development to the critical layers that enable their practical and trustworthy deployment.
2. Beyond Compliance: Operational Requirements: The landscape does not appear to be driven solely by compliance mandates. Instead, as AI systems become more agentic, questions regarding the identity of an agent, its access permissions, its memory capabilities, its execution authority, and the potential for human intervention are transforming into fundamental operational requirements. This suggests that aspects currently categorized under AI governance are increasingly becoming integral components of active technical deployment infrastructure.
3. A Market in Formation: The data reveals that the market for trustworthy AI infrastructure is still in its nascent stages. Only 19 companies have reached Series B or later funding rounds. While capital is concentrated in a relatively small number of larger companies, many of which are still model-centric, the experimentation and innovation are widely distributed across all seven layers of the stack and across 15 different countries. This dynamic suggests an environment ripe for further disruption and the emergence of new market leaders.
Implications for Investors: Navigating Systemic Opportunities
The current state of the trustworthy AI market presents a compelling opportunity for investors who understand the confluence of systemic risks and emerging demand. This situation bears a resemblance to the evolution of the climate tech investment landscape, where systemic risks, coupled with technological advancements, regulatory frameworks, and growing demand, ultimately transformed fragmented solutions into investable sectors and essential infrastructure.
While AI presents unique challenges and opportunities distinct from climate change, the fundamental lesson of market formation remains relevant. In an increasingly automated world, agency and control are becoming economically valuable commodities. The initial investment cycle of AI has seen an unprecedented influx of capital dedicated to the production of intelligence. The next phase, however, promises to reward companies that can make intelligence, agentic systems, and physical AI demonstrably trustworthy and deployable at scale.
By focusing on the infrastructure layers that enable secure, ethical, and controllable AI, investors can position themselves to capitalize on the durable demand signals emanating from enterprises, governments, and users alike. The Trustworthy AI Market Map serves as a vital compass, guiding investors through this complex and rapidly evolving landscape, enabling them to move from simply mitigating risk to actively investing in the solutions that will define the future of artificial intelligence.
Reynand Wu
Raul Delapena Setiawan
