Frontiers and frontlines: notes from the road

As with any investment, your capital is at risk.

On the surface, a frontier AI lab, a skilled nursing facility, a gas turbine and a garbage truck have little in common. But after several weeks on the road across America, a pattern began to emerge. In each case, the most interesting investment question was not what the company does today, but what its constraints reveal about tomorrow: compute, labour, power, logistics, culture, ambition.

Back in Edinburgh, the US Growth team sat down to share the most salient insights from the road – a collaborative process that helps connect the dots. It is through this process that we seek to develop what we call ‘uncommon understanding’ – insight that is not yet reflected in share prices. 

The frontier: AI reaches warp speed

AI has reached a tipping point. As portfolio manager Gary Robinson laid out in AI agents: why 2025 rewrote the AI timeline, our assumptions for AI have been revised upward.

Across dozens of meetings with leading large language model (LLM) labs, AI-native software companies and established software incumbents, what we heard was consistent and striking: companies are generating real, meaningful returns on their AI spending. For some, Research and Development (R&D) is improving, customer service is transformed, or efficiency gains are changing how organisations are structured.

High-profile layoffs at hyperscalers and other tech firms overshadow what we heard on the road. Productivity gains are unlocking new capacity, not simply eliminating jobs. One company had reduced an engineering team from 10 to eight. The striking point was not the reduction. It was what happened next: the two ‘freed’ engineers were redeployed to projects that had sat for months in the someday pile.

That suggests the industry is entering a transition and that the investments are rational. It suggests the bullish case for compute demand is, if anything, too conservative. One CEO described his company as consuming “basis points” of the AI tokens he expects it to consume long term. Taken across the economy, and based on conversations, one portfolio manager describes it as “every flop of compute the world can produce will find a buyer for the foreseeable future”.

It’s showing up in the numbers. Hyperscalers’ capital expenditure (capex), hardware pricing, backlogs, labour costs – they’re all rising. These are real-world constraints. Consistent feedback from companies across the AI infrastructure supply chain suggests we may be in a genuine capital-expenditure super cycle.

That is not a quarter-by-quarter observation. It is a multi-year one.

Software’s fitness test

On the wrong side of the AI disruption appears to be software – a painful transition period is underway. The transition from seat-based to consumption-based pricing is akin to what happened when on-premises moved to the cloud: companies that embraced the change were punished in the short run, but those that did not make the transition were ultimately the ones that destroyed value over the long run.

The current transition is applying a similar fitness test.

The team met with a wide array of software-as-a-service (SaaS) businesses. From industry pioneer Salesforce to the lesser-known Roper Technologies, an eclectic mix of small software businesses. Roper recently acquired Subsplash, a cloud-based software and fintech company serving more than 20,000 churches and faith-based organisations. Niche.

Most interesting to us, however, was the CEO’s comment that it has yet to see any signs of AI threats in the numbers. And, in his opinion, the transition is likely to create just as much opportunity as risk. Better products, new features, faster shipping and deeper integration into customer workflows. He conceded the range of outcomes is wider, but that leaves scope for upside, too.

With obvious parallels to the cloud transition, in our view, the founders and leadership teams with the vision, adaptability and cultural fitness will be those most capable of succeeding in the AI era. Owner mindsets with high agency. One founder described themselves as ‘running into the fire’. They need to make sure they run fast enough and are resilient enough to tolerate any short-term pain in pursuit of the long-term prize. That has become a prerequisite.

Shopify is a prime example of an organisation that has adapted to the early development of AI agents. Our view is that agentic commerce – transactions executed by autonomous AI agents on behalf of consumers – will expand the share of total commerce that flows through digital channels. Shopify already operates trusted rails for ecommerce, enabling merchants around the world. It is now actively positioning that infrastructure for the agentic age: channel-agnostic, indifferent to whether the buyer is a human or a bot. That is what running fast looks like.

The frontlines: off the beaten track

Beyond AI and software, the most interesting conversations were about constrained supply. NextEra, one of the largest electric power utilities in the United States, told us that the much-touted ‘gas renaissance’ powering AI datacentres is, in their view, largely not taking place.

Turbine lead times stretch to the end of the decade. The skilled welders and pipeline engineers who would install them have migrated to liquefied natural gas (LNG) export terminals on the Gulf Coast. Actual gas-fired generation on the US grid fell in 2025. Instead, companies are turning to solar, battery and the electrical grid to power the datacentres of the future.

In addition, it’s very important for us to underwrite the competitive positions of the businesses that we invest in. That underwriting cannot be done from a desk. It is done by visiting management, touring facilities and seeing how the business operates in the wild, not as it appears in an investor presentation.

One stop took us far from the language of model weights, tokens and datacentres: Pacifica Nursing & Rehab Center, an Ensign Group facility in the San Francisco Bay Area. This is the unglamorous but essential part of investing: getting close to the day-to-day reality behind the numbers.

Ensign is run by thoughtful, empathetic people doing vital, often unseen work, caring for the sick and the elderly. What mattered was not a polished presentation about culture, but the repetition of small operational behaviours: how managers spoke about staff, how care routines were handled, how local autonomy showed up in decisions. That is the sort of evidence that rarely appears in a model but can matter enormously over the long term.

SpaceX: the final frontier

One of our most recent trips was to SpaceX, our largest private company investment. SpaceX headlines focus on whether, when and at what price the company might list. That is not the question that matters most to us.

We first backed SpaceX at a valuation of around $30bn. But the significance of that early investment is not simply the gain to date. It is that, as long-term investors across private and public markets, we are not required to treat a future listing as the end of the journey. Our advantage is patience.

That is why these trips matter. Meeting companies across America – from AI labs and software businesses to utilities, nursing facilities and SpaceX – helps us connect signals that can look unrelated in isolation. Together, they form a picture of where constraints are emerging, where ambition is concentrated, and where exceptional companies may compound through structural change.

Our job is not to predict precisely when a bottleneck shifts, which quarter the agentic commerce thesis shows up in earnings, or the exact number SpaceX prices its IPO at.

Our job is to develop uncommon understanding in a small number of exceptional companies positioned to compound growth through whatever happens, then back them long enough for that compounding to do its work.

Risk factors 

The views expressed should not be considered as advice or a recommendation to buy, sell or hold a particular investment. They reflect opinion and should not be taken as statements of fact nor should any reliance be placed on them when making investment decisions.

This communication was produced and approved in May 2026 and has not been updated subsequently. It represents views held at the time of writing and may not reflect current thinking.

Potential for Profit and Loss 

All investment strategies have the potential for profit and loss, your or your clients’ capital may be at risk. Past performance is not a guide to future returns.

This communication contains information on investments which does not constitute independent research. Accordingly, it is not subject to the protections afforded to independent research, but is classified as advertising under Art 68 of the Financial Services Act (‘FinSA’) and Baillie Gifford and its staff may have dealt in the investments concerned.

All information is sourced from Baillie Gifford & Co and is current unless otherwise stated. 

The images used in this communication are for illustrative purposes only.

196532 10063089