Asset managers’ AI spend rises as trust lags
The gist
Asset managers are pouring over $100 million a year into AI, but scaling up from flashy pilots to real business value is hitting a wall of data chaos, workforce skepticism, and regulatory headaches.
What to know
- AI budgets in asset management average $103 million annually, with 35% aimed at cybersecurity and 29% at analytics, but only 19% of firms have scaled autonomous agents widely.
- A single asset manager racked up a $500 million AI vendor bill in just one month, as strict governance, fragmented global data rules, and compliance with regulations like the EU AI Act slow progress.
- Human advisors aren’t out—leveraging AI for client outreach and decision-making helped Travelers boost underwriting profits by 21%, while trust and skill gaps keep humans in the investment loop.
AI’s Physical Infrastructure Boom
AI’s explosive growth is fueling a trillion-dollar race in chips, power, and data center buildouts, but physical bottlenecks like power grid delays and labor shortages are creating lucrative moats for those who can scale the ecosystem.
Investment in AI hardware is roughly evenly split between compute components and the broader physical infrastructure, with networking and power each accounting for about 15-20% of the non-compute hardware value, while the remainder is dedicated to data center buildouts. This balanced allocation underscores that scaling AI capabilities demands not only advanced chips but also substantial investment in the supporting physical ecosystem, which captures significant profit margins and unprecedented growth, as seen in sectors spanning chips, memory, and networking.
Despite a global surge in infrastructure investment—highlighted by large tech companies' expected $1 trillion spending in 2025-2026 and Goldman Sachs’ projection of $7.6 trillion from 2026 to 2031—scaling AI infrastructure faces critical bottlenecks such as power grid connection delays that can take five to seven years, labor shortages with hundreds of thousands of unfilled construction roles, and outdated manufacturing facilities. These physical constraints create a capital-intensive moat around AI infrastructure, with hyperscalers spending over $500 billion annually on capex and pushing toward $600 billion in 2026, signaling that the physical economy is the next frontier for AI expansion.
While early AI capital expenditures focused heavily on chips, data centers, and power, the investment landscape is evolving as bottlenecks shift from raw compute to cooling, power, optics, and memory, with optics and memory emerging as particularly promising segments. This transition reflects a maturation of the AI infrastructure market from broad basket investments to a more selective stockpicker’s market, where the value increasingly concentrates in capital-intensive base layers such as semiconductors, data platforms, and inference engines, rather than the thinning application layer whose margins are compressing as inference costs become variable.
In asset management, AI investment has surged to an average projected spend of $103 million over the next year, with a strategic focus shifting toward cybersecurity (35%), data and analytics (29%), and research and development (27%), reflecting a broader infrastructure buildout beyond software alone. However, scaling AI agents remains challenged by structural issues like data readiness, limited visibility into AI operating costs—with only 4% of firms reporting full transparency—and the complexity of agentic systems, leading to cautious scaling despite heavy investment. Notably, accountability for AI infrastructure decisions is centralized at the highest organizational levels, with nearly 60% of respondents attributing responsibility to the CEO or executive committee.
Governance: AI’s Hidden Cost
Strict data policies, vendor vetting, and fragmented global regulations are driving up compliance costs and slowing AI adoption, with workforce resistance and skills shortages threatening to stall asset managers’ digital ambitions.
Enterprise AI governance remains a formidable challenge as organizations grapple with strict authorization policies to prevent risky or non-enterprise-grade AI tools from compromising data security and compliance. As highlighted in the analysis from June 2026, firms must carefully vet AI vendors to ensure SOC1 and SOC2 adherence, robust data resilience, and business continuity, especially given the complexity of licensing costs that can balloon unexpectedly—as one example cited a staggering $500 million AI bill incurred in just 30 days. This governance burden is compounded by the need to control data sovereignty, with many AI tools storing sensitive information in foreign jurisdictions, forcing enterprises to prioritize solutions that meet stringent compliance certifications to avoid regulatory pitfalls.
Data governance and integration deficiencies significantly hinder AI adoption across regions such as Australia, New Zealand, and EMEA, where fragmented technology estates and unmanaged shadow IT create operational blind spots that erode data quality and confidence. Despite overwhelming prioritization of data integration and governance—94% in Australia and 89% in New Zealand—less than half have formal AI-specific data policies, leaving 40% of leaders in EMEA identifying AI data usage as their top visibility gap. This fragmented landscape not only complicates compliance with evolving regulations like the EU AI Act but also slows the transition from pilot AI projects to scalable, value-generating deployments.
Workforce resistance and talent shortages represent critical bottlenecks in scaling AI within asset management and financial services, where skills gaps, job security fears, and trust issues have caused employee adoption rates to plummet from 37% to 22%. Firms like Coinbase have responded by reducing tech headcount by 14%, reallocating human labor toward uniquely human tasks while emphasizing the strategic imperative for employees to master AI orchestration. Reflecting this urgency, 83% of organizations are investing in reskilling programs and offering salary premiums of up to 15% for AI-capable talent, underscoring that without workforce adaptation, the promise of AI-driven productivity gains remains elusive.
Despite heavy AI investments exceeding $100 million in asset management, organizational readiness and governance remain the linchpins for successful AI scaling, with accountability concentrated at the CEO or executive committee level in nearly 60% of firms. However, many companies remain in early AI maturity stages—only 12% are leaders—often relying on AI assistants that marginally improve individual efficacy without fundamentally enhancing system efficiency. Clear, outcome-driven governance from top leadership, as one firm demonstrated by committing to a 50% reduction in GNA costs through AI, is essential to bridge the value gap and move beyond fragmented pilot projects toward integrated, autonomous AI agents that deliver measurable business outcomes.
Leadership Drives AI Integration
Scaling AI in asset management hinges on CEO-led experimentation, disciplined change management, and the rise of agentic AI agents that demand new trust, auditability, and human-in-the-loop oversight.
The transition from pilot AI projects to integrated agentic AI in asset management hinges less on technology breakthroughs and more on disciplined management and leadership engagement. Geoffrey Moore’s dual-framework approach, combining protected incubation zones with CEO-led transformation initiatives, enables firms to nurture AI innovation without being stifled by core business demands. Leadership habits such as maintaining rapid experiment velocity, focusing on a single validated hypothesis, and prioritizing high-risk, high-value bets—often with direct CEO involvement—are critical to scaling AI effectively and driving EBIT improvements attributable to generative AI.
Despite significant AI investments topping $100 million, asset managers remain largely in the experimentation phase, with only 19% scaling autonomous AI agents across multiple functions. Employee resistance—rooted in skills gaps, job security fears, and trust issues—poses a formidable barrier, causing adoption rates to drop sharply. In response, firms are intensifying workforce reskilling efforts, with 83% investing or planning to invest in upskilling and over half willing to pay a premium for AI-capable talent, underscoring the organizational shifts necessary to move from pilots to integrated AI deployment.
Agentic AI agents are redefining asset management workflows by autonomously executing complex tasks and coordinating across systems while operating under human guardrails, thus enhancing productivity and decision-making. Unlike opaque large language models, these agents provide traceable, auditable, and sourced outputs with human-in-the-loop oversight, addressing critical trust and governance concerns. Goldman Sachs Asset Management emphasizes that agentic AI is not merely the final step in automation but a discrete technological and organizational shift requiring new leadership strategies to embed AI as autonomous digital co-workers within investment operations.
The evolving AI playbook, bolstered by expanded partnerships and matured production methodologies, is enabling early adopters in asset management to realize measurable ROI from agentic AI deployments. However, scaling AI beyond pilots demands robust governance frameworks and board-level commitment to AI literacy and training, as compliance teams grapple with fragmented data and legacy infrastructure. As Paul O’Brien warns, making AI understanding a board-level priority and instituting mandatory training are essential to prepare organizations for the disruptive potential of integrated AI agents.
Human Advisors’ AI Edge
AI is supercharging human advisors’ productivity and profits, but emotional intelligence and genuine client relationships remain the decisive factor as wealth management becomes a battle of human-AI collaboration.
AI adoption is significantly enhancing operational efficiency and profitability across traditional sectors like financial services and insurance, with Travelers reporting a 21% year-over-year increase in underwriting profits directly linked to AI investments. This technological uplift is not replacing human advisors but rather augmenting their capabilities, enabling them to leverage AI tools for improved risk management and decision-making, thereby driving broader economic benefits beyond the tech industry.
In financial advisory, the competitive edge increasingly belongs to human advisors who adeptly integrate AI into their workflows rather than to AI technology providers themselves. Stan Gregor of Summit Financial highlights a transfer of client relationships not from advisors to AI, but between advisors based on AI adoption, with AI automating administrative burdens and accelerating personalized client outreach through rapid synthesis of market insights—transforming advisors into sophisticated consumers of AI who outpace competitors in earnings growth.
Despite AI’s growing role in data analysis and operational tasks, the human advisor’s personal connection remains indispensable for client retention and nuanced understanding of individual circumstances. As Terry Cook notes, the advisor who builds genuine relationships—such as knowing clients’ families and values—will prevail, especially as AI democratizes technical wealth management tasks like tax mitigation, making emotional intelligence and trust-building the irreplaceable human edge in wealth management.
Investor behavior underscores the complementary dynamic between AI and human advisors: while younger generations like Gen Z and Millennials lead AI adoption to enhance research speed and risk assessment, a majority of investors still rely on human professionals for final investment decisions due to their judgment and accountability. Barry O'Byrne of HSBC encapsulates this balance, emphasizing that clients use AI to explore options but depend on trusted advisors for context and validation, illustrating AI’s role in boosting investor confidence without supplanting the human touch.
