Spatial computing is an emerging technology within the investment management industry. Some investment managers are bullish and are actively exploring it by building their applications and experiences. For example, Fidelity Investments released an immersive metaverse experience, “The Fidelity Stack” ¹. “The Fidelity Stack, built in Decentraland, features a multilevel design complete with a lobby, dance floor, and rooftop sky garden for users to explore on foot–or even through teleport. In the Invest Quest at The Fidelity Stack, users are challenged to traverse the building, learning the basics of ETF investing while gathering “orbs” along the way.”¹

Spatial computing applications exist across learning, client interactions, and wealth planning. Hyperpersonalized GenAI client experiences can be amplified through spatial computing. Imagine engaging with your financial advisor in a personalized space in augmented—put on your virtual reality (VR) headset at home, work through wealth planning scenarios, and see your surroundings adjust as you interact with different levers. GenAI agents can generate images, voice, and text to create a tailored client experience. 

Despite the potential, challenges include technical limitations (processing and battery life), data retention and privacy, ergonomic form factor, limited user access, and the need for skilled engineering talent.

While large language models (LLMs) excel at quick insights on many topics, cases requiring specialized knowledge or unique terminology remain challenging to solve efficiently and cost-effectively. SLMs have advantages over LLMs by being efficient (lower resource needs, faster processing, lower latency) and having a more compact focus. “One such SLM, developed by Kohei Watanabe, a member of the Lazard Quantitative Equity team, is Latent Semantic Scaling (LSS). LSS is a semi-supervised document scaling technique that locates documents in any language on user-defined dimensions… [It] estimates the semantic proximity between words in the data, while the seed words define the dimensions of interest… In addition to being transparent—we can completely unpack and fully understand LSS outputs—this level of precision comes at a fraction of the cost if we were to attempt the same exercise using an LLM.”² The team leveraged LSS to gain insights from annual reports in Mandarin, identifying companies least impacted by the volatile Chinese housing market, without needing translation. 

Other opportunities for SLMs in investment management include chatbots for proprietary analysis and financial documents, compliance tools trained on specific rules, and advisor assistants trained to understand financial data and concepts. As agentic AI transforms how we do work, SLMs will be instrumental in enabling adoption at scale, with each agent being powered by an SLM specific to its unique tasks. 

Adopting SLMs comes with challenges. Companies need to design a multiagent architecture, where specialized SLMs are responsible for specific functions, similar to a microservices approach in software development. Ensuring accuracy and reliability through robust monitoring and data privacy guardrails will be crucial.

Many technology teams at investment management firms have been focused on migrating from mainframe to cloud and are now revisiting their data/cloud strategy as AI becomes a competitive imperative. Investment managers should ensure that their data centers can support scaling AI infrastructure and energy needs. AI workloads demand low-latency, high-data-rate transfers and increased computational demands, requiring a shift to high-power, AI-ready infrastructure. Firms are investing in core infrastructure and partnering with leading providers to incorporate this trend into their investment theses. BlackRock, Microsoft, Global Infrastructure Partners, and MGX announced the Global AI Infrastructure Investment Partnership in September 2024, “with a goal of building the backbone of future AI infrastructure… [to] build data centers and … the supporting grid energy infrastructure to power them” ³. In the same month, “Blackstone, along with the Canada Pension Plan Investment Board (CPP Investments), agreed to buy AirTrunk … the fastest-growing data center platform in the Asian-Pacific region.”³

Other hardware advancements include laptops and desktops with AI chips and copilots built in, and futuristic screens for simulations and what-if scenario analysis.

Robust hosting strategies will likely become critical to balance on-premises solutions, hyperscalers, and emerging providers, which will help navigate regulatory, operational, and cost constraints.

At many AI-pioneering investment managers, the AI agenda is driven by IT and data, analytics, and AI teams with business sponsorship. Early successes have elevated these teams to partners required to achieve AI-led transformation initiatives.

Investment managers are increasingly adopting modular and vendor-driven solutions that streamline operations and provide flexibility. Vendors like Aladdin and Charles River offer scalable platforms consolidating data from public and private markets. This can free internal business and technology/data teams to focus on non-commoditized aspects and become data-driven for agility, personalization, and operational efficiency. “As Vanguard CEO Tim Buckley and CIO Greg Davis explain, AI condenses earnings and research reports, enables faster and more accurate decisions, and simplifies advice. And that is only the beginning.”⁴. To fuel these prospects, specialized tech talent is a must.

As AI initiatives scale, IT teams will revisit their operating model through five pillars: infrastructure, engineering, FinOps, talent, and innovation. There will be a shift from human-in-charge to human-in-the-loop, transforming IT delivery. New roles like prompt engineers will expand, reshaping the talent mix. AI-driven automation could reduce business teams’ reliance on IT. However, areas such as model risk management, monitoring, and LLM ops will become more prevalent, especially as AI regulations become clearer.

These shifts are essential for firms aiming to stay competitive and responsive to expanding the role of AI in meeting evolving client and stakeholder needs. 

As investment management firms embrace innovations and gear up for the future, they must also adapt to new and evolving cybersecurity risks. While quantum computing is still emerging, it offers strong potential for evolving securities trading, optimizing fund asset management, and predictive analytics. However, as quantum computing adoption grows, so will concerns around data privacy and security. While building quantum-secure protocols may not currently be top of mind, investment managers are prioritizing the maturity of their cyber programs. JP Morgan’s Global Head of Cybersecurity Awareness Program emphasizes that “Building a united and secure oversight framework—across cybersecurity, risk management and business resiliency—is a top priority for our firm” ^5. A well-defined roadmap to reach a post-quantum world is critical to ensure investment management technologies can continue to operate within privacy and security constraints. 

2025 sees the rise of new GenAI features within core systems such as trading, customer relationship management (CRM), human resources (HR), and finance. Within the front office, AI/GenAI capabilities are being released in Aladdin Copilot8 and Bloomberg GPT7. Asset servicers like State Street offer front-to-end solutions with AI embedded throughout the platform, offering AI-fueled “front to back and AI in between” ⁶ solutions. There is also excitement around leveraging AI features from CRM and finance/HR systems to transform how work gets done. Investment managers will likely need to update workflows to embed the full power of GenAI within these systems to achieve impact on their business.

Investment managers are integrating GenAI into their core systems, either by transforming existing systems or building new ones.  They are using GenAI to develop the AI capabilities and/or migrate in-house systems and tools to a more modern tech stack. Many legacy systems lack documentation and experienced engineers, but AI-assisted solutions can help overcome these challenges at scale. Investment managers must address the complexity of rolling out these features and ensure proper education on the technology and its limitations. Implementing controls, such as human-in-the-loop processes, ongoing monitoring, and prompt maintenance, is essential to safeguard GenAI deployments, especially for mission-critical core systems.