AI's agile “appliers”

A study from MIT in 2025 found 95% of generative Artifical Intelligence (AI) projects in businesses were failing to deliver a meaningful return on investment at the enterprise level. A plethora of pilot projects failed to scale and deliver tangible results. But huge advances are being made and we are seeing real world benefits from agile businesses applying AI. At Equitable Investors we have pragmatic AI-leveraging investments and we utilise AI tools ourselves, while recognising their shortcomings. We are cognisant of the likelihood that a lot of capital may be torched, especially among deployers of AI infrastructure, before the winners emerge and the long-term benefits of applying AI are enjoyed economy-wide. But the sunk costs of “hyperscalers” represent an opportunity for agile businesses.

The leap

Large Language Models (LLMs), the mode of “AI” we usually refer to, are designed to create text. The term "Large" refers to the model's architecture, which contains trillions of parameters; and "Language" indicates that the model is trained on vast amounts of text written in natural languages. LLMs operate by processing "tokens" - basic units of input such as a single word, character, or number. Computing power has increased exponentially and the technology is undergoing a significant "agentic leap" as. Agentic LLMs are not purely reactive. They can reason, plan, and autonomously execute multi-step workflows to achieve specific goals.

Leveraging “hyperscaler” spend

While hyperscalers (Google, Microsoft, Meta etc) are still spending hundreds of billions of dollars in the Installation Phase (building the compute rails) and investors are concerned the returns are not flowing rapidly enough, the Deployment Phase is well underway. Deployment is being spurred on by Agentic LLMs, and is set to radically change some business models. In this process, we see the value migrating from the “suppliers” of compute to the “Appliers”. 

Economically, the cost structure of LLMs is bifurcating: the cost to train frontier models is rising into the billions, while the cost of inference (running the model to generate responses) is plummeting - inference costs per token (the standard unit of measurement) dropped by more than 99.5% over the past five years.

Capital-light businesses - including software developers but really any company seeking to apply technology to create operating leverage or competitive advantage - have a never-seen-before opportunity to enhance if not reinvent their businesses by applying established and emerging AI technologies at low cost.

 “What we see right now in the AI space is it's really kind of a barbelled market demand where on one end, you have the AI labs who are spending gobs and gobs of compute right now, along with what I would consider a couple of runaway applications. And then at the other side of the barbell, you've got a lot of enterprises who are getting value out of AI in doing productivity and cost avoidance types of workloads. These are things like customer service or business process automation or some of the fraud pieces. And then in that middle of the barbell are all the enterprise production workloads. And I would say that the enterprises are in various stages at this point of evaluating how to move those, working on moving those and then putting them into production. But I think that middle part of the barbell very well may end up being the largest and the most durable.”
- Amazon CEO Andrew Jassey (Q4 2025 earnings call)

Software and Software-as-a-Service (SaaS) businesses can win or lose here. They need to be “Appliers” (a term we have borrowed from Polen Capital). A successful software business does not build its “moat” just from coding. Proprietary content, knowledge and sovereign or proprietary data, hard-earned reputations, marketing smarts, regulatory barriers, existing customer relationships, positive business cultures - they are all ingredients. AI can be applied to enhance these qualities and drive operating leverage. But flimsy businesses that have no proprietary “value-add” will be exposed.

• ServiceNow has reported it now resolves roughly 80% of support tickets autonomously, saving 400,000 labor hours annually

• Amazon’s AI shopping assistant (“Rufus”) has been used by >300m customers

To extract value, it may well be the case that SaaS “Appliers" will have to shift from charging on a “per user” (or “per seat”) basis to monetizing the outcomes - essentially taking a share of the productivity gains enjoyed by their customers. It may also be the case that “Appliers” benefit while the broader economy does not if efficiencies lower demand for labour and consumption slows in response.

With quarterly cash flow reports behind us and the half-year reporting season weell underway, there has been plenty of opportunity to hear from companies lately about their adoption, plans and fears for AI. Here’s what some of the companies we have been listening to have been saying:

Respondents to a KPMG survey of tech executives reported that the average ROI on their internal corporate investment in AI stands at 200% (2x) but KPMG found that higher investment does not guarantee better returns and smaller organisations have been getting more leverage out of their investment in AI. “In early stages of maturity, smaller, focused investments generate higher rates of return,” KPMG noted. High performers show high ROI (4.5x), even with lower investment levels relative to revenue.

 

Indications from our portfolio and the broader market are clear: the AI 'Deployment Phase' is already yielding tangible results for disciplined “Appliers”. Whether it is HSN achieving a 29.2% EBITDA margin through AI productivity or ServiceNow autonomously resolving 80% of support tickets, the operational leverage is no longer theoretical. While 95% of generic enterprise AI projects may currently struggle to scale, Equitable Investors remains focused on smaller, agile organisations, where KPMG identifies the highest ROI on AI investment. The winners are successfully breaking the link between revenue growth and labor costs - a magic formula for growing small companies.