AI Is Not Killing APS. It Is Repricing the Supply Chain Software Stack.

Executive thesis

Generative and agentic AI are forcing investors to reassess where value sits in enterprise software. Supply Chain Planning is now part of that debate.

The evidence available so far does not point to a collapse of APS platforms in favor of AI-native challengers. It points to a repricing of the stack.

Interfaces, reporting, workflow navigation and basic recommendations are becoming easier and cheaper to reproduce. Constraint models, optimization, operational context, orchestration and execution remain much harder to replicate.

That distinction is increasingly relevant for valuation. Scale, installed base, recurring revenue and product breadth still matter, but they no longer guarantee a premium on their own. Investors will want to see whether AI improves growth, lowers implementation effort, raises customer expansion and increases the share of operational decisions that can be automated reliably.

The central question is no longer whether APS survives. It is which parts of APS remain scarce enough to justify a premium.

1. Software is being repriced, but Supply Chain software is still growing

Two market signals are developing at the same time.

Gartner expects spending on Supply Chain Management software with agentic AI capabilities to rise from less than $2 billion in 2025 to $53 billion by 2030. Adoption among enterprises using SCM software is expected to increase from about 5% to 60% over the same period.

At the same time, software valuations have come under pressure as investors consider whether agents could reduce the number of applications, interfaces and software seats companies need.

There is no contradiction. AI can expand the overall market for decision software while reducing the economic value of parts of the existing stack.

The relevant valuation question is where margin and pricing power move next.

2. Kinaxis is the clearest public-market test

There are few publicly listed pure-play Supply Chain Planning vendors, which makes Kinaxis the most useful public proxy.

Its operating performance remains strong. In Q2 2026:

• SaaS revenue increased 20% year over year.
• ARR increased 19% to $465.6 million.
• Total revenue increased 16%.
• Adjusted EBITDA increased 23%.
• Management raised full-year revenue guidance.

Kinaxis has also expanded its AI strategy through Maestro and Maestro Agent Studio, with agents grounded in company data, workflows, intelligence and governance.

Yet valuation has not expanded in the same way. Kinaxis had a market capitalization of approximately C$4.25 billion at the end of 2022. In August 2026 it was about C$4.8 billion. Revenue grew materially faster over that period, which implies a clear compression in the sales multiple.

The market is not rejecting Kinaxis. It is simply not paying extra for the presence of AI alone.

That raises the bar for incumbent APS vendors. The next re-rating is more likely to depend on evidence that AI changes unit economics, customer expansion and deployment intensity than on the number of agents released.

3. Execution-heavy software may be more defensible

Manhattan Associates provides a useful contrast because its portfolio sits closer to execution through warehouse management, order management, transportation, store operations and commerce.

Its market capitalization increased from approximately $7.6 billion at the end of 2022 to roughly $12.6 billion in August 2026, while trailing revenue increased from about $0.77 billion to around $1.1 billion.

An agent may reduce the need to navigate a planning application. It does not remove the need to allocate inventory, release an order, schedule warehouse labor or execute a physical movement.

That gives transaction and execution systems a different defensive profile. AI still needs systems through which actions can be executed.

Descartes Systems Group shows a similar pattern, although with less valuation resilience. Its market capitalization rose from about $5.9 billion at the end of 2022 to around $6.9 billion in August 2026, while trailing revenue increased from roughly $0.48 billion to $0.75 billion. Its sales multiple compressed, but the category has not been priced as if it were becoming obsolete.

4. Private-market valuations remain difficult to interpret

The private market offers useful reference points, but they are stale by definition unless a new transaction occurs.

• o9 Solutions was valued at $2.7 billion in early 2022 and $3.7 billion following an investment in 2023.
• RELEX's last widely disclosed valuation was €5 billion in 2022.
• Anaplan was taken private in 2022 at approximately $10.4 billion.
• OMP has no directly observable public valuation.

These figures should be treated as historical markers, not current prices.

Public markets reprice continuously. Private companies do not. The more useful reading therefore combines the last disclosed valuation with growth, capital allocation and product strategy.

On that basis, the major planning vendors are not retreating. They are broadening their AI capabilities and trying to redefine the boundaries of APS before challengers do it for them.

5. AI-native challengers are entering through focused wedges

Private capital is clearly funding newer Supply Chain software companies, but most are entering through narrow decision domains rather than rebuilding the entire planning stack.

• supplier and value-chain intelligence;
• supplier risk;
• network design;
• forecasting;
• inventory;
• logistics finance;
• exception management;
• decision intelligence.

This is a credible route into the market. A new entrant does not need to replicate an entire APS platform if it can become materially better at a valuable decision.

The challenge comes later. As these companies move toward constrained enterprise planning, they must handle feasibility, complex constraints, industry models, optimization, scenario consistency and execution integration.

That is where the incumbent advantage may still be strongest.

6. The moat is moving below the interface

AI reduces the cost of reproducing several forms of software value. Conversational interfaces, reporting, simple workflows, basic forecasts and recommendation layers are becoming easier to build.

They may remain useful, but they are becoming less scarce.

By contrast, representing a complex industrial network across factories, shared capacities, alternative routings, material constraints, allocation rules and multiple planning horizons remains difficult.

The key asset may therefore sit below the application layer.

For APS vendors, that asset can include:

• constraint models;
• optimization IP;
• enterprise knowledge models;
• industry-specific planning logic;
• orchestration across decisions;
• integration with execution;
• implementation knowledge accumulated over time.

If those assets remain necessary and differentiated, AI can strengthen the incumbent position by making them easier to access and use.

If agents can bypass them, reproduce them or offer sufficiently good substitutes, the valuation premium can erode quickly.

7. Incumbency can still be an advantage

For established APS vendors, legacy depth can work in both directions.

OMP is a good example. Much of its value sits in production logic, constraints, optimization models, scenarios and industrial planning expertise rather than in the interface itself.

If that depth can be exposed through a simpler and more agentic operating model, incumbency becomes an asset. The vendor gains AI-native interaction without rebuilding APS-grade decision logic from scratch.

If product complexity prevents faster deployment and simpler workflows, the same legacy becomes a drag on competitiveness.

The dividing line is therefore less about age and more about whether an incumbent can separate its decision depth from the historical complexity of its product experience.

8. Investors will eventually focus on economics, not features

Agent announcements are becoming common across enterprise software. They are unlikely to remain a meaningful valuation signal on their own.

A more useful scorecard would include:

AI-driven growth
Does AI create incremental ARR or simply become part of the existing offer?

Customer expansion
Does AI improve retention and platform penetration?

Implementation productivity
Can deployments be completed materially faster and with less consulting effort?

Delivery intensity
How much configuration and services work is required for each dollar of recurring revenue?

Software-company productivity
Does AI improve R&D, support and implementation economics?

Decision automation
What share of operational decisions can customers execute autonomously within defined rules, constraints and authority thresholds?

The last metric could become especially important. A large catalogue of agents means little if customers still make almost every operational decision manually.

Economic value appears when AI changes how many decisions can be made faster, more consistently and with less operating effort.

The Dataleo View

The market is not pricing the death of APS. It is reducing the premium attached to parts of enterprise software that are becoming easier to reproduce.

Kinaxis illustrates the pressure. Strong growth and a broader AI proposition have not led to multiple expansion. AI capability, by itself, is not enough.

Execution-oriented platforms appear better protected where they control transactions that still have to happen regardless of the interface used above them.

Private capital is funding AI-native challengers, but mainly through focused decision wedges rather than wholesale APS replacement.

The valuation debate therefore comes down to a more practical distinction: commoditizable application value versus scarce decision value.

For incumbents, the opportunity is to expose decades of planning logic through a much simpler operating model. For challengers, the challenge is to prove that better interaction can scale into reliable decision-making under enterprise constraints.

The next valuation premium in Supply Chain software is likely to accrue to the vendors that combine AI-native simplicity with decision depth that remains difficult to reproduce.

Methodology note

This analysis combines public financial information, public-market capitalization data, disclosed private funding rounds and vendor strategy available through August 2026. Private-company valuations reflect the last disclosed transaction and should not be interpreted as current market values.

Public-market valuation changes cannot be attributed to AI alone. Growth, interest rates, acquisitions, governance, guidance and broader software-market sentiment all affect multiples.

The objective is not to establish causality between individual AI announcements and share-price movements. It is to assess how AI is changing the market's view of where durable value sits in the Supply Chain software stack.