Beyond winners and losers: How to position a sustainable portfolio for the AI era

Every technology cycle eventually produces its own vocabulary of fear and greed, and this one is no exception. Equity markets in 2026 increasingly trade as a momentum-driven referendum on a single question: Is this company an AI winner or an AI loser? With valuations swinging violently as the narrative reassigns companies from one camp to the other, sometimes on little more than a product announcement or a conference remark.

We understand why. AI is the most consequential general-purpose technology since the internet, and its effects on competitive advantage will be profound. But we believe the binary framing is analytically lazy, and that it creates precisely the kind of mispricing a disciplined, long-horizon process can exploit. A technology that touches everything does not divide the world into two clean categories. It creates a spectrum of exposure, and understanding where each business sits on that spectrum, and why, has become one of the most important tasks in constructing a resilient portfolio.

That is why the Global Sustainable Equity Team categorises companies through four distinct lenses.

The four lenses

Figure 1: Our four-lens framework for categorising the AI value chain

^{Source: Janus Henderson Investors, Global Sustainable Equity Team. For illustrative purposes only.}

AI Enablers provide the foundational infrastructure, data and tools required to build, train and deploy AI systems: advanced semiconductors, cloud platforms and data centres, specialised software, and the data and analytics that feed the models themselves. These businesses sit at the source of the AI value chain, and many map directly onto our Knowledge & Technology theme. Critically for our process, the best of them pass both of our questions convincingly: the diffusion of intelligence through the economy is a powerful social good – in scientific discovery, healthcare, education and productivity – and the enablers’ economics, characterised by deep intellectual property moats and structural demand growth, are among the most attractive in global equities.

EXAMPLES

Physical Bottlenecks are the companies resolving the real-world constraints that limit AI’s growth – power generation and grid infrastructure, cooling and thermal management, advanced networking and interconnects. This is where the AI story and the sustainability story become quite literally the same story. The single greatest obstacle to AI scaling is energy – its availability, its cost and its carbon intensity – which means the companies solving for grid capacity, electrification and energy efficiency are simultaneously enabling the technology and decarbonising it. These holdings can span our Cleaner Energy and Efficiency themes, for example, and they illustrate a point we have long made: the energy transition is not a constraint on technological progress but a precondition for it.

EXAMPLES

AI Beneficiaries integrate AI into their products, services and operations to enhance efficiency, customer experience and innovation. They benefit from AI adoption rather than enabling it – design software that automates and simulates, platforms that personalise at scale, service businesses compressing cost and cycle times. This is, deliberately, where many of our portfolio holdings sit. These are companies with franchises we already wanted to own – durable moats, strong returns on capital, clear thematic alignment – for which AI acts as an accelerant in an existing engine rather than the engine itself. The distinction matters: a beneficiary’s investment case does not depend on the AI narrative holding but is strengthened if it does.

EXAMPLES

AI Resilient businesses are those whose models are relatively insulated from AI disruption – neither heavily reliant on AI, nor likely to be displaced by it in any foreseeable timeframe. Their moats live in the physical and institutional world: infrastructure networks built over decades under regulatory and permitting regimes that will not be granted twice; essential local services whose route density economics support only one or two rational operators; businesses grounded in physical presence and human interaction. AI will make these companies more productive; it cannot make them unnecessary. In a portfolio context, this cohort is the ballast – a source of uncorrelated, idiosyncratic compounding that does not rise and fall with the AI narrative.

EXAMPLES

Two important nuances. First, the lenses are not mutually exclusive: many holdings legitimately span more than one category – an enabler of AI that is also a beneficiary of its own tools, or a resilient infrastructure business that is simultaneously solving a physical bottleneck. We treat this overlap as information, not inconvenience: Businesses that screen well through multiple lenses are often the most robust of all. Second, categorisation is a living judgement, revisited continuously as the technology and each company’s positioning evolve. A lens assignment is the beginning of the debate, not the end of it.

What the framework reveals

Mapping the investible universe through the four lenses aims to create balance by design, not by accident. No single lens dominates, and no single version of the AI future is required.

This is a deliberate response to what we see as the defining risk of the current market – narrative concentration. When a single theme drives an outsized share of index returns, portfolios that simply ride the theme are making an implicit, highly correlated bet – and portfolios that avoid it entirely are making the opposite one. Our approach is to be exposed to AI the way we want to be exposed: Through companies that clear our sustainability and wealth-creation hurdles individually, diversified across the value chain, with a substantial foundation of holdings whose compounding does not depend on how the AI story unfolds. If the buildout accelerates, the enablers, bottleneck-solvers and beneficiaries participate fully. If it disappoints, the resilient cohort and the beneficiaries’ standalone franchises provide the stability. In neither scenario are we hostage to a single narrative.

Moats in the AI era: Broader than they appear

Running the investible universe through the framework can also sharpen our thinking on what constitutes a durable competitive advantage in an AI world. The instinctive answer – physical assets are safe, digital businesses are vulnerable – turns out to be far too simple. Durable moats in the AI era come in several distinct forms:

Figure 2: Four sources of durable competitive advantage in the AI era

^{Source: Janus Henderson Investors, Global Sustainable Equity Team. For illustrative purposes only.}

• Physical and regulated networks: Infrastructure assembled over decades under permitting regimes and local market structures – often natural monopolies or duopolies – that no amount of algorithmic capability can replicate. AI makes these assets more productive; it cannot rebuild them.
• Network effects: Platforms whose value compounds with participation – marketplaces, ecosystems and two-sided networks where each additional user deepens the moat. AI can improve the experience inside the network; it struggles to bootstrap a rival network from zero.
• Proprietary data and standards: Businesses whose datasets, ratings, scores and benchmarks are embedded in the workflows and regulations of entire industries. Far from being disrupted by AI, these companies own the very raw material AI needs – their data becomes more valuable as more intelligence is applied to it.
• Trust, brand and human relationships: Franchises grounded in physical presence, service and accumulated confidence, where the product is ultimately a human experience that intelligence can augment but not substitute.

The common thread is duration, which we deem to be an essential component of a holistic definition of sustainability. In each case, the moat’s source is something slow to build and slow to erode, which is exactly what allows us to underwrite these businesses over the multi-year horizons our process demands, and to hold a collection of genuinely non-correlated, idiosyncratic compounders rather than one crowded bet expressed many times over.

Sustainability is the through-line

It would be easy to read all of this as a technology piece with a sustainability wrapper. We would argue the causality runs the other way. Our sustainable development themes are what led us to define opportunities in the first place, and AI has amplified the relevance of nearly every one of them.

The Knowledge & Technology theme captures the enablers democratising access to intelligence and information. Cleaner Energy and Efficiency capture the bottleneck-solvers ensuring the AI buildout is powered, cooled and connected sustainably – arguably the largest electrification and efficiency investment programme in history. Our Health, Water, Efficiency and Quality of Life themes are populated with beneficiaries and resilient compounders whose services the world needs regardless of the technology cycle, and which AI is making more productive.

And through it all, the two questions remain the constant. Is the world a better place because of this company? AI has raised the stakes on this question – a technology this powerful creates its own environmental footprint and social risks, and we aim to hold AI-exposed companies to account on energy sourcing, resource intensity, responsible deployment and workforce impact through active engagement. Is this company going to grow wealth? AI has raised the stakes here too – accelerating the compounding of the well-positioned and shortening the runway of the complacent. A process that insists on both answers, is in our view better suited to this environment than either uncritical AI enthusiasm or reflexive AI avoidance.

Conclusion: Resilience is a portfolio property

The market wants AI to be a simple story of winners and losers. Our framework insists on a more granular truth: That a technology touching everything demands a holistic approach to the whole value chain – enabling it, solving its physical constraints, benefiting from it, and compounding independently of it – with every holding, in every lens, still required to make the world better and grow wealth at the same time.

That is not hedging. It is about participating fully in progress while refusing to bet the outcome on any single version of the future. In a black-and-white momentum market, we think that balance – and the discipline behind it – is itself a source of durable advantage.