It’s becoming clearer with each passing day that AI ranks among one of the most seismic events of the past 300 years, up there with the Industrial Revolution and the development of the internet. There’s no doubt that it has had an equally significant impact on industries, sectors, systems and procedures across the world.

Certainly, much has been made of the positives in the effects and transformation it could bring to the financial and investment sector in particular, as well as the companies connected or adjacent to this. However, so huge is the scale and so rapid the advances, that it can be near impossible to seek a defining perspective that one can trust, as well as the confidence and clarity that can feed into future planning and decision-making.

We’ve already explored how AI is reshaping multiple aspects of private markets and, in our CIO commentary each Tuesday, we can see how it’s having an effect on public ones, too. Huge gains have been seen in those companies developing AI and LLMs such as Anthropic, OpenAI and xAI, as well as those producing the microchips and associated technology needed to propel the sector such as Nvidia, Intel as well as countless other SMEs and startups.

Recent data suggests that AI’s economic impact is real but it is also disparate and uneven. A 2024 McKinsey Global Institute report estimates that generative AI could contribute between $2.6 trillion and $4.4 trillion annually to global GDP, largely through productivity enhancements in knowledge work. Similarly, Deloitte highlights that early adopters are seeing efficiency gains of 10-20% in specific workflows, particularly in customer service, coding, and operations. These developments, while significant, are still largely concentrated in those firms with the infrastructure, data architecture and management discipline needed to deploy AI effectively.

An increase of AI adoption does not automatically translate into signs of increased productivity. There is no doubt that it is a result of its utilisation, but it is still not running parallel with the increased upsides as some would have expected.

Arbra has previously explored how the depreciation curve of AI assets is steep, as value accrues not from ownership of models, but from their integration into revenue-generating processes. This turns AI from a capital expenditure story to an operational discipline.

At the same time, capital markets have been quick to price such bullish scenarios. Bloomberg Intelligence reports that AI-related equities have driven a disproportionate share of market returns in 2024–2025, with a small cluster of mega-cap firms accounting for the majority of gains. This concentration should ask us to pause and question: are these gains built on lasting value, or are they being driven by market enthusiasm and expectations?

A useful way to break this down is through productivity metrics rather than valuation multiples. For instance, BCG analysis shows that companies successfully embedding AI into core processes outperform peers in earnings before interest and taxes (EBIT) margins by up to 6 percentage points across a three-year period. However, the same research notes that more than 70% of AI initiatives fail to scale beyond pilot phases, which is an important reminder that execution risk also remains high in the sector.

Today, AI’s impact on public and private markets is diverging in both pace and visibility.

Certainly, in the former, AI has become a dominant thematic driver, influencing sector rotations, capital flows, as well as index performance. The Financial Times has reported that AI-linked stocks have contributed significantly to the resilience of US equities despite macroeconomic headwinds, and could be masking broader market fragility. This dynamic creates both opportunity and distortion as valuations increasingly reflect future optionality rather than current earnings power.

Private markets, by contrast, offer a more nuanced picture. Here, AI is less about headline valuations and more about operational leverage. Venture and growth investors are increasingly focused on companies that use AI to compress cost structures, accelerate time-to-market, or unlock new revenue streams in underpenetrated geographies. Data from PitchBook suggests that AI investment in private markets remains significant, even as capital becomes more selective and hardened on proven use cases.

Importantly, private markets also provide exposure to the infrastructure layer that reinforces the AI ecosystem, such as data centres, specialised chips, energy supply, and connectivity. These “picks and shovels” assets often exhibit more predictable cash flows and clearer pathways to monetisation than application-layer startups. For sophisticated allocators, this creates an opportunity to capture AI-driven growth without relying on the more hype-driven valuations.

The central tension in AI investing today is between narrative velocity and operational reality. The speed at which technological capabilities are evolving has also outpaced the ability of many organisations to absorb, analyse and act on them effectively. It’s in this area where nebulous information allows for risk and opportunity to reside.

For investors, the implication is clear: it’s no longer enough to assess the technology itself. What matters just as much is whether a company has factors such as the right people, clean and usable data, and the ability and facility to turn AI tools into tangible business outcomes. In practice, that means looking closely at how AI is utilised in its day-to-day operations and execution, as opposed to buzzwords, pitch decks and demo videos.

Assurances by those companies using the most advanced models won’t lead to better investment choices alone, but efficacy will be proven by those who can translate commitments into sustained and compounding productivity gains.

This is particularly relevant to those with exposure to frontier and alternative markets, where AI adoption has the ability to leapfrog legacy systems, but could also face constraints in infrastructure and talent. At Arbra, our core competencies are derived from combining robust due discipline and deep financial expertise with macro insight and knowledge of the conditions “on the ground”.

It’s in this style of diligent capital allocation, coupled with a deep awareness of the local context as well as operational feasibility, where we see AI becoming the real differentiator.