Kernpunten
- The frenetic pace and scale of AI development is challenging how governments and businesses operate, widening dispersion between genuine beneficiaries, overstated winners, and vulnerable losers across sectors.
- Regulation, security and dependency are increasingly influencing procurement, pricing power and capital flows, creating both durable advantages and hidden risks for companies.
- Flexible long/short absolute return strategies are well placed to navigate mispricing, backing resilient businesses while challenging hype in an uneven AI-driven landscape.
We were fortunate recently to share dinner and conversation with two experienced professionals at the intersection of cyber security, geopolitics and policy, as Janus Henderson hosted a client event exploring artificial intelligence (AI) and its implications for today’s markets.
Approaching the discussion as absolute return investors, what interested us most was not simply where AI innovation is heading, but how it is reshaping competitive dynamics, capital allocation, and risk across sectors. AI has moved beyond a narrow technology lens to become a strategic force, influencing how countries operate, how companies build and defend advantage, and emphasising where markets are struggling to keep up with rapid change.
While investment into AI‑related infrastructure has accelerated sharply (the ‘MAG7’ tech companies are expected to invest over US$680 billion in AI development and infrastructure in 2026), there are still some key questions that sit behind these headlines that are of fundamental importance to how the story of AI might unfold over the coming years. Who controls these systems? Who is trusted to operate them? How will new forms of tech dependence create opportunities and risks on both the long and short side for investors.
The new era of AI geopolitics
One of our strongest takeaways was how deeply geopolitical competition is shaping the AI landscape. In particular, the tension between the US and China has become a defining feature of the global technology stack. It is an active force determining who can build what, where, and at what cost.
A useful way to think about this is to separate where AI is already working well from where it is still struggling. Today, AI performs extremely effectively in the “closed” world: software, data analysis, coding and digital workflows. In contrast, it remains far less reliable in the “physical” world, such as robotics, autonomous vehicles, or real‑world decision‑making.
That distinction matters for investors because it highlights where disruption is likely to arrive fastest. Software‑heavy sectors are already facing competitive pressure from the roll-out of various new AI tools. Meanwhile, physical industries have seen slower, more uneven impacts, driven by regulation or social concerns over safety, rather than by outright AI capabilities.
Policy now sits at the centre of this picture, shaping business models and capital allocation, and influenced by a range of factors, from export controls and subsidies to industrial policy and foreign dependency. Restricting access to advanced chips may slow progress in some regions, but it does not stop deployment or learning altogether. In practice, it changes the route taken and opens the door to innovative workarounds.
There is also a clear sense that different philosophies are at work. In the US, the dominant mindset remains “model first”: build the most advanced intelligence possible, then work out how to monetise it. In China, the instinct is closer to “deployment first”: treat AI as a utility and roll it out at scale, iterating quickly to drive costs down, and penetrate commercial markets.
These approaches offer different strengths and weaknesses, but both systems face binding constraints. In the US, there are infrastructure constraints, with power grids, planning systems and water supplies struggling to keep pace with the growth in demand from data‑centre expansion. In China, access to advanced computing hardware remains an impediment to both evolution and roll-out.
Tech stacks as strategy – why AI is not a single trade
Another important insight was a shift towards thinking in terms of “tech stacks”, rather than individual companies or products. Markets have commonly treated AI as a single trade, where we have seen investors crowd into the most visible opportunities at the same time. But AI is a layered system that stretches from physical infrastructure at the bottom to applications and services at the top.
At the base of the stack sit energy, land, cooling and data centres. Above that come networks, semiconductors, cloud platforms, data, and ultimately distribution into real‑world products and services. Competition, margins and bottlenecks look very different at each layer. The most durable opportunities, or those most susceptible to the impact of AI, might not be the most visible ones.
At the same time, competition within a stack can be brutal. Where technology becomes commoditised or heavily subsidised, margins can be competed away very quickly. Understanding where value sticks – and where it slips – is increasingly important. Learning through feedback and driving down costs can matter just as much as technical brilliance. In some markets, being good enough or cheap enough could outcompete “best in class” technologies that are expensive and/or slow to deploy.
AI is no longer one trade: trust, geopolitics and technology choices are fragmenting markets and widening the gap between winners and losers.
Luke Newman, Portfolio Manager
The importance of trust in AI… and why it is so fragile
Trust can sound abstract, but in AI it is increasingly decisive. In sensitive areas such as finance, identity and data security, trust is already fundamental, directly influencing procurement decisions, regulatory approval and customer adoption.
As AI becomes more embedded in critical systems, trust is also shaping how companies and governments frame policies around risk, jurisdiction, regulation and dependency, reflecting the scale of potential impact. While this kind of fragmentation is inefficient from a purely economic perspective, with duplicated systems, overlapping supply chains and higher costs, it can still be a rational response where resilience and strategic autonomy matter.
Security sits at the heart of this dynamic. Advanced AI systems can identify weaknesses faster than humans, increasing the likelihood of scares, near misses and scrutiny around system resilience. At the same time, they can accelerate adaptation and defence. Initiatives such as Project Glasswing, where Anthropic has provided government agencies and select organisations early access to its Claude AI models (Mythos), aim to reinforce trust through efforts to strengthen cybersecurity and support public‑sector adoption.
This helps explain why governments are increasingly willing to pay more for trusted technology, even when cheaper alternatives exist. Trust has become a hurdle for access. France’s decision to replace Windows with Linux illustrates how technological dependency is now viewed as a tangible risk. How companies and governments respond, through alignment, policy or regulation, is likely to play an important role in shaping capital flows and the durability of competitive advantages.
The speed of change is exceptional, so how can we measure potential, or gauge whether a business is ripe for AI disruption?
- Evidence that AI is accelerating top-line growth, delivering measurable cost savings, or sustained margin improvement.
- The presence of ‘moats’ that can potentially protect a business from the impact of AI. This can include unique capabilities or long-term contracts, data advantages, credible security credentials, the cost of switching to alternatives, or the presence of existing trust-based relationships with its clients.
- Regulatory risk management measures (ie. procurement standards, legislation or compliance needs) that can insulate the business from the impact of AI driven competition.
Gevolgen voor beleggers
The AI super-cycle has driven markets since 2023, but the trade has often been treated in overly binary terms: crowd into the obvious AI winners and assume today’s leaders will automatically monetise tomorrow’s demand. In reality, the range of outcomes is widening. Some businesses are proving to be durable beneficiaries; others are facing disintermediation faster (or slower) than expected. This is precisely the kind of environment where absolute return strategies, operating a flexible framework of long and short investments, can add value by positioning for both the winners and the losers of AI.
The most durable opportunities, or those most susceptible to the impact of AI, might not be the most visible ones.
Ben Wallace, Portfolio Manager
Key to this is separating AI stories from investment realities. On the long side, we look beyond crowded narratives, toward persistent opportunities in both infrastructure that supports the growth of AI (power, water, data, and critical services) and quieter beneficiaries where AI is driving productivity, cost compression and margin expansion. These companies may not be AI leaders, but they can be net beneficiaries as AI becomes a silent hand of disinflation across services and manufacturing.
On the short side, the focus is areas where valuations already assume flawless execution, where monetisation pathways remain uncertain, or where competitive advantages are being eroded. As AI accelerates the pace of disruption, markets can become inefficient, creating mispricing opportunities for more nimble investors, or those able to look beyond the crowd consensus.
As trust and technology choices reshape the AI landscape, returns are more likely to reward selectivity than simple optimism. This is the kind of environment in which absolute return equity strategies may excel, navigating widening market dispersion by taking both long and short positions, with the flexibility to adjust to a changing environment. AI will not lift all boats equally. As trust, regulation and technology stacks fragment markets, the greatest opportunities for investors may lie in distinguishing the genuine beneficiaries from those whose prospects are being overstated.
Absolute return investing: A type of investment strategy that seeks to generate a positive return over time, regardless of market conditions or the direction of financial markets, typically with a low level of volatility. Please note positive (or absolute) returns are not guaranteed).
Capital allocation: The strategic allocation of distributing money between different investments, or in the case of companies, where the business allocates its financial resources, for example, upgrading equipment, research and development, paying down debt, or returning capital to shareholders.
Disintermediation risk: The risk that intermediary businesses in a supply chain might be removed from the process.
Dispersion: The extent to which a distribution of data points is stretched or squeezed. If the data points cluster around certain values, then dispersion is low, whereas if they are more spread out, then dispersion is high. For example, dispersion in stocks measures the range of returns for a group of stocks. Higher dispersion means that stock returns are spread more widely on either side of the benchmark, creating opportunities for stock pickers to outperform by selecting the winners and avoiding the losers.
Long/short investing: A portfolio that can invest in both long and short positions. The intention is to profit from combining long positions in assets in the expectation that they will rise in value, with short positions in assets expected to fall in value. This type of investment strategy has the potential to generate returns regardless of moves in the wider market, although returns are not guaranteed.
Long position: A security that is bought with the intention of holding over a long period in the expectation that it will rise in value.
Magnificent Seven (MAG 7): The term ‘Magnificent Seven’ refers to the seven major technology stocks – Apple, Microsoft, Nvidia, Amazon, Tesla, Alphabet, and Meta – that have dominated markets in recent years.
Profit margin: The amount by which the sales of a product or service exceeds business and production costs.
Project Glasswing: A new initiative bringing together a group of tech giants, including Apple, Anthropic, Google, Microsoft and NVIDIA, in an effort to secure the world’s most critical software, helping to reshape cyber security for governments and companies in a future where AI models will become increasingly capable of finding and exploiting software vulnerabilities. Claude Mythos Preview is a system provided to Project Glasswing partners, and organisations that build and maintain critical software infrastructure, so they can seek to identify, secure against, or mitigate risks from new iterations of AI systems before they are released to market.
Short investing: Fund managers use this technique to borrow then sell what they believe are overvalued assets, with the intention of buying them back for less when the price falls. The position profits if the security falls in value.
Top-line growth: An increase in a company’s gross revenues or sales over a specific period, which is used as an indicator of business growth.
Dit zijn de standpunten van de auteur op het moment van publicatie en kunnen verschillen van de standpunten van andere personen/teams bij Janus Henderson Investors. Verwijzingen naar individuele effecten vormen geen aanbeveling om effecten, beleggingsstrategieën of marktsectoren te kopen, verkopen of aan te houden en mogen niet als winstgevend worden beschouwd. Janus Henderson Investors, zijn gelieerde adviseur of zijn medewerkers kunnen een positie hebben in de genoemde effecten.
Resultaten uit het verleden geven geen indicatie over toekomstige rendementen. Alle performancegegevens omvatten inkomsten- en kapitaalwinsten of verliezen maar geen doorlopende kosten en andere fondsuitgaven.
De informatie in dit artikel mag niet worden beschouwd als een beleggingsadvies.
Er is geen garantie dat tendensen uit het verleden zich zullen doorzetten of dat prognoses worden gehaald.
Reclame.
Belangrijke informatie
Lees de volgende belangrijke informatie over fondsen die vermeld worden in dit artikel.
- Aandelen/deelnemingsrechten kunnen snel in waarde dalen en gaan doorgaans gepaard met hogere risico's dan obligaties of geldmarktinstrumenten. Als gevolg daarvan kan de waarde van uw belegging dalen.
- Als een Fonds een hoge blootstelling heeft aan een bepaald land of een bepaalde geografische regio, loopt het een hoger risico dan een Fonds dat meer gediversifieerd is.
- Het Fonds kan gebruikmaken van derivaten om zijn beleggingsdoelstelling te verwezenlijken. Dit kan leiden tot hefboomwerking, wat de resultaten van een belegging kan uitvergroten en waardoor de winsten of verliezen van het Fonds groter kunnen zijn dan de kosten van het derivaat. Het gebruik van derivaten gaat ook gepaard met andere risico's, waaronder met name het risico dat een tegenpartij bij derivaten niet in staat is om haar contractuele verplichtingen na te komen.
- Als het Fonds activa houdt in andere valuta's dan de basisvaluta van het Fonds of als u belegt in een aandelenklasse/klasse van deelnemingsrechten in een andere valuta dan die van het Fonds (tenzij afgedekt of 'hedged'), kan de waarde van uw belegging worden beïnvloed door veranderingen in de wisselkoersen.
- Wanneer het Fonds, of een afgedekte aandelenklasse/klasse van deelnemingsrechten, tracht de wisselkoersschommelingen van een valuta ten opzichte van de basisvaluta te beperken, kan de afdekkingsstrategie zelf een positieve of negatieve impact hebben op de waarde van het Fonds vanwege verschillen in de kortetermijnrentevoeten van de valuta's.
- Effecten in het Fonds kunnen moeilijk te waarderen of te verkopen zijn op het gewenste moment of tegen de gewenste prijs, vooral in extreme marktomstandigheden waarin de prijzen van activa kunnen dalen, wat het risico op beleggingsverliezen verhoogt.
- Het Fonds kan geld verliezen als een tegenpartij met wie het Fonds handelt niet bereid of in staat is om aan zijn verplichtingen te voldoen, of als gevolg van een fout in of vertraging van operationele processen of verzuim van een derde partij.