Reflections on 2025: Building the Foundations for Scalable, Data-Driven AI, Looking Ahead to 2026
- kevinm26
- 2 days ago
- 5 min read
Updated: 16 minutes ago
This piece is commentary and analysis, not investment advice.
The past year has been a busy yet highly productive one at Iona Star as we continue to build a portfolio of companies aligned with our long-term technology vision. We have met with many exceptional founders developing businesses with compelling, well-defined, and differentiated value propositions.
Across every prospective investment, our focus remains consistent: the defendability of the moat, be it technological, data-driven, or structural, and the clarity with which that moat translates into sustainable commercial advantage.
AI Became a Business Imperative in 2025
In 2025, AI moved decisively from experimentation to execution. Technology-enabled businesses of all sizes and across all industries are now actively exploring and adopting AI to drive workforce efficiency, enhance customer experiences, and improve forecasting, decision-making, and operational resilience.
A recent McKinsey report surveying more than 2,000 businesses across 100 countries found that nearly two-thirds are already running limited AI pilots, with a similar proportion actively exploring AI agents.
Building on large language models (LLMs), AI agents are software systems that can sense their environment, reason and plan, and act autonomously to achieve defined objectives. They are increasingly deployed to automate clerical tasks, power marketing and customer engagement, accelerate sales cycles, support compliance and QA functions, and more.
Organisations that successfully integrate AI agents into human workflows stand to unlock significant efficiency gains, driving both revenue growth and cost reduction.
Understanding the True Cost of AI
While AI promises lower operating costs over time, deploying it introduces a new and often poorly understood cost base. Businesses are increasingly required to model and manage their AI cost base, encompassing the full spectrum of costs involved in developing, deploying, and extracting value from AI systems.
Some of these costs are well publicised: specialist talent, computational power, data centre construction and cooling, and rising energy demands. Less discussed, but equally material, are data-related costs, including acquisition, licensing, preparation, enrichment, and long-term storage.
For a technology that is fundamentally dependent on data inputs, these costs can be substantial and, if poorly managed, value-destructive.
Data as the Critical Enabler
At Iona Star, our investment thesis is centred on enabling organisations to leverage AI productively, responsibly, and cost-effectively. One of our key focus areas and passions is efficient, high-quality data provisioning, ensuring that datasets are accurate, structured, enriched, and contextually usable at scale.
Our team brings decades of experience from financial services and data-intensive industries, where sophisticated data architectures have long underpinned competitive advantage. We are now applying this experience across the broader AI ecosystem, which is even more dependent on access to diverse, high-quality datasets.
We work closely with our portfolio companies, partners, and their customers to help them extract value from AI applications operating on multimodal data, including event-driven data, historical time-series, and unstructured sources, such as text, images, audio, and human-generated conversations.
By guiding organisations through data landscapes, procurement strategies, licensing models, and delivery architectures, we help accelerate the transition from deterministic to probabilistic decision-making, a necessary shift for navigating uncertainty in markets, technology roadmaps, customer behaviour, and geopolitical and economic conditions.
From Raw Data to Actionable Intelligence
A prerequisite for accurate, real-time prediction is the ability to process and analyse vast quantities of raw data efficiently.
By curating, enriching, and organising data at scale, businesses can assign probabilities to outcomes, make informed predictions, and trigger autonomous actions through AI agents. To describe this process, the term "data hydration" comes to mind: the transformation of raw data into datasets suitable for high-performance AI inference.
Just as hydration supplies living systems with vital water, data hydration supplies AI systems with vital intelligence.
We are also increasingly focused on what we refer to as the big data hyperverse, the expanding universe of structured, unstructured, and synthetic datasets that, when combined with deep domain expertise in data and analytics, materially improve model accuracy, robustness, and real-world utility.
Looking ahead to 2026, real-world model data will come even more sharply into focus. We expect to announce a number of developments and investments aligned with this expanding data hyperverse, including access to new sources of high-value, domain-specific data; advanced approaches to data generation and enrichment; and technologies that improve how complex, multimodal datasets are structured, governed, and consumed by AI systems. As models and agents become more autonomous and context-aware, differentiated access to diverse, high-quality real-world data will be a decisive driver of performance, reliability, and economic efficiency.
Modernising Asset Ownership and Liquidity
Another area of interest is the transformation of financial markets by leveraging the opportunity that tokenisation promises. To date, institutional adoption has been slower than expected due to infrastructure limitations, transaction costs, and a lack of compelling large-scale use cases. Early efforts, such as bank-led tokenised bond issuances on private blockchains, have delivered limited incremental value but have not driven widespread market participation.
We believe this dynamic could change from 2026 onwards. The market is increasingly shifting toward the tokenisation of large, illiquid asset classes that are poorly served by traditional exchanges but offer predictable, asset-backed cash flows. Sectors such as data centres, renewable energy infrastructure, cargo shipping, electric vehicle charging networks, and real estate present significant opportunities to unlock new liquidity and institutional yield.
Portfolio Introductions
During 2025, Iona Star invested in a number of innovative companies building critical infrastructure across AI, data, and enterprise technology. Below are brief introductions to a selection of our portfolio and the roles these companies play in advancing trusted, scalable, and enterprise-ready AI.
ABT is building a next-generation tokenisation platform focused on real-world assets, enabling institutions to unlock liquidity in traditionally illiquid asset classes. As regulatory clarity improves and institutional adoption accelerates, ABT is well-positioned to support secure, scalable tokenised asset issuance and management across global markets.
More to come on AkashX soon.
Craxel consolidates fragmented data into a unified, AI-ready knowledge asset. Its unprecedented speed and efficiency at scale uniquely delivers the real-time contextualised information needed for decision making—whether by humans or machines.
CUBIG is a global leader in synthetic data generation and management, and AI data security, helping enterprises unlock the value of sensitive data through privacy-preserving, high-fidelity synthetic alternatives.
HyperLayer is modernising financial services infrastructure through programmable, AI-driven platforms that sit atop legacy banking systems, accelerating product innovation and time-to-value for institutions.
INQDATA is redefining market data access through its Market Data as a Service platform, offering financial institutions a scalable alternative to entrenched legacy systems and enabling rapid, multi-bank deployment.
OriginalVoices is pioneering AI-powered, consumer-grounded Digital Twins that deliver real-time behavioural insights at scale, transforming how enterprises access authentic human intelligence.
Working closely with AI-Services leader, Brainpool, Tunedd applies AI to due diligence and risk assessment, enabling faster, higher-quality decision-making for investors and professional services firms.
Looking Ahead to 2026
As we enter 2026, we remain focused on expanding our portfolio, deepening our investment thesis, and supporting founders building the infrastructure required for the next generation of AI-enabled businesses.
Alongside the companies highlighted above, we have invested in other businesses that will be announced later in the year, further strengthening our exposure across data, AI, and financial infrastructure. We look forward to sharing more detail on these investments in due course.
We will continue to publish insights as our thinking evolves and as our portfolio grows, alongside announcing further investments that translate our conviction into long-term value as we scale the fund.
Expect to hear more from us soon.