Hatchings

JANUARY 2026 – ISSUE 27

The first wave of AI profits has gone overwhelmingly to those selling the picks and shovels. The obvious winners to date are the hardware and infrastructure players – (e.g. Nvidia, TSMC, Broadcom) the hyperscalers (e.g. Microsoft, AWS and Google Cloud) and the data centre developers and operators. AI capital expenditure (capex) is still running at eye watering levels and much of this capex has landed in GPUs, chips and networking within datacentres.
The emerging question is: will software providers get paid for AI capability or does having AI just become the bare minimum expectation in a competitive market? Currently, the answer looks like both. AI is clearly becoming a ‘table stake’, but you can see the beginning of the monetisation in a handful of large software names.

Microsoft

Microsoft 365 Copilot is priced at US$30 per user per month on top of existing 365 licences. While Microsoft doesn’t disclose Copilot revenue directly, there are metrics available. GitHub Copilot, the coding arm of copilot, was last reported to have 1.8 million paying subscribers in 2024.¹ GitHub now generates more than US$1bn in annualised recurring revenue (ARR), with Copilot accounting for more than 40% of GitHub’s growth in ARR.²

Adobe

Adobe discloses more than US$5b of “AI influenced” ARR.³ This revenue is tied to Firefly, Acrobat’s AI Assistant and other generative AI features. Adobe has also disclosed that more than US$250m of ARR⁴ comes from products that are explicitly AI.

Atlassian

Atlassian Intelligence and the newer Rovo AI are bundled into premium and enterprise tiers of software rather than sold as standalone products. Atlassian management has highlighted rapid adoption, with 1.5m⁵ monthly active users of Atlassian AI features in mid-2025, stepping up to nearly 3.5m users late in 2025⁶. The monetisation here is less about AI revenue and more about nudging customers up the software tier stack and defending price increases.

ServiceNow

ServiceNow’s Now Assist generative AI feature is sold as an add-on. ServiceNow management has stated AI-related net new annual contract value is on track to exceed US$500m by the end of 2025.⁷ They have indicated that more than US$1b of AI-related revenue in 2026⁸ will come from a subscription base of US$12–13b.

TechnologyOne (TNE) – AI as pricing power

TNE’s “Plus” launch is the clearest example of explicit AI monetisation. Plus is pitched as an agentic AI layer across the whole ERP. An agent that can query and act across finance, HR, property, assets and budgeting through a natural language interface. It’s integrated directly into existing workflows rather than being a sidecar app.

The commercial model they put around this is quite clear:

• Product embedded AI – Existing customers get AI features inside the 19 individual products at no additional licence fee for the first wave of capabilities. There’s a usage cap where customers receive a limited number of free AI interactions per product per month. Above that, pricing is set on a bundled ARR interaction basis.
• PLUS – the cross-enterprise AI workspace, carries its own annual subscription fee. There is a capped number of PLUS “conversations” per month, with conversation pricing above the cap bundled in ARR bundles.
• Price book uplift – For new products and modules, the company has said the price book will rise by around 10%⁹as AI capabilities are bundled in.

This is a strong example of AI being used as a profit lever. Existing customers get immediate value uplift reducing friction and driving habit formation. Once workflows are re-wired and users are dependent on AI-driven processes, incremental usage beyond the cap converts into high-margin metered revenue.

There is still execution risk here as the product is still in the phase of signing up early adopters. Theoretically, TNE is one of the better examples of AI being used to re-price the entire solution rather than given away as a free feature.

WiseTech (WTC) – AI as a carrot to move the pricing model

At the 2025 investor day WTC framed AI as a productivity tool for all customers, helping accelerate the take up of WTC’s new commercial model known as the Cargowise value pack.

The new value pack model bundles multiple modules and capabilities into simplified packages while removing per seat subscription pricing. Access to these AI tools is tied to these packs in the new commercial model.

While the per customer economics are opaque, company commentary and industry feedback suggest that customers migrating to the new commercial model are expected to spend more in aggregate. AI capabilities are being used as the carrot to accelerate adoption of this model.

95%¹⁰of WTC Cargowise customers have been opted in to the new structure. The largest enterprise customers however, who contribute to the majority of revenue, will transition more gradually as longer term contracts roll off. In practice, the model is similar to the TNE approach, AI tools are embedded and available and therefore we assume will be utilised.

Xero (XRO) – testing and changing habits before monetisation

“Just Ask Xero” (JAX) is XRO’s generative AI agent for small businesses and advisers. First announced in 2023, JAX entered beta testing in 2024 and has been progressively expanded through 2025. It provides a conversational interface inside Xero and can complete common tasks such as generating invoices, editing quotes, paying bills and performing bank reconciliations.

XRO is early in its AI monetisation journey and is taking a workflow first, pricing later approach. The strategy is to embed AI into day-to-day tasks, build reliance and demonstrated value, then introduce charging once customers are reliant. This is a softer approach that TNE and WTC use with AI being effectively free in the early stages, which reflects XRO’s position on the monetisation curve.

JAX remains in beta testing and is positioned as a way to deepen engagement, simplify workflows and automate routine processes rather than as a separate paid tier. Management commentary has emphasised getting customers to use AI in their everyday workflows while XRO tests usage patterns and customers’ willingness to pay. For now, AI is primarily a tool to defend the core franchise and create future pricing power, rather than a driver of near-term P&L.

Pro Medicus (PME) – the AI works, but monetisation is slow

PME is a good example of the other side of the AI monetisation equation. The technology is clearly valuable, but the economic plumbing (reimbursement and willingness to change) lags.

PME’s Visage Breast Density Imaging platform is an AI classifier, developed with Yale University on a dataset of 33,000 mammograms (164,000 images)¹¹. It is designed to provide more consistent density scores. The algorithm matches the consensus of 5 radiologists, which is better than 1 radiologist on their own, so is pitched as an extra set of eyes on a scan.

The product is an add on module to their flagship product Visage 7. It is commercially available but not contributing in a meaningful way. Given a lack of reimbursement among AI products in the field, radiology groups often bundle AI costs into existing service fees (capturing value indirectly through productivity) or ask patients to opt in and pay out of pocket, which raises ethical and equity questions. There is inherent friction for a radiologist telling patients there is an optional non-rebatable fee for an extra layer of AI eyes.

In this context, AI risks becoming a ‘table stake’ for PME, necessary to stay competitive with other AI-enabled vendors, but difficult to monetise directly until reimbursement frameworks catch up. More realistically, AI will be monetised indirectly through PME’s broader enterprise contracts, helping to improve win rates, increase contract sizes and deepen customer stickiness. Futher, PME has an alternate path to AI profitability which involves becoming the aggregator and orchestrator of high-quality medical imaging algorithms. PME do not need to build every proprietary model themselves if they can be the trusted platform that enables and integrates the models.

A few conditions seem necessary for AI to be more than just a ‘table stake’ where companies can achieve returns on their AI investment.

1. Unique, high-quality data at scale – if a model can be trained on data that very few have, the AI outputs become much harder to commoditise and therefore potentially easier to monetise. In contrast, if your primary value proposition is stitching together information that already exists elsewhere, AI agents can often replicate that with access to APIs and other public data, this information will be harder to monetise.

• TechnologyOne’s advantage is mission-critical, regulated public-sector ERP data with a single-instance SaaS architecture, which makes cross-module AI feasible.

• Xero and WiseTech both sit on highly granular transactional data

• PME’s deep relationships with tier-1 academic hospitals and research capabilities use anonymised patient data.

2. Workflow integration and habit formation that reduces friction and increases the odds that users become more dependent on the relevant tech solution.

3. Clear economic outcomes for the user – AI that demonstrably saves time, improves output or reduces error is easier to monetise than AI that delivers marginal convenience.

Is AI just the new ticket to play? 

The emerging evidence from case studies suggests AI is a mix of defence and differentiation initially used to further entrench existing customers, justify price increases and lift customers to higher priced tiers. 

Over time, as workflows are redesigned and users become dependent on AI-mediated processes, it becomes a structural lever for pricing power with usage-based fees on top of subscriptions, and the ability to reset price books.

The companies best positioned are those with large, proprietary, high-quality data sets and deep workflow ownership. 

¹  Microsoft FY24 Q3 earnings remarks
https://www.microsoft.com/en-us/investor/events/fy-2024/earnings-fy-2024-q3

² Microsoft FY24 Q4 earnings remarks
https://www.microsoft.com/en-us/investor/events/fy-2024/earnings-fy-2024-q4

³  Adobe Q3 FY25 earnings release
https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/11905202/aiy4w5teshy5t.pdf

⁴  Adobe Q3 FY25 earnings release
https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/11905202/aiy4w5teshy5t.pdf

⁵  Atlassian TEAM Q3 FY25 shareholder letter
https://s206.q4cdn.com/270053503/files/doc_financials/2025/q3/TEAM-Q3-2025-Shareholder-Letter.pdf

⁶  Atlassian TEAM Q1 FY26 shareholder letter
https://s206.q4cdn.com/270053503/files/doc_financials/2026/q1/TEAM-Q1-2026-Shareholder-Letter.pdf

⁷  Q3-2025 Earnings Call transcript
https://www.alphaspread.com/security/nyse/now/investor-relations/earnings-call/q3-2025

⁸  Second Quarter 2025 Financial Results
https://www.sec.gov/Archives/edgar/data/1373715/000137371525000274/erq2fy25.htm

⁹  FY25 results call transcript coverage
https://finance.yahoo.com/quote/TNE.AX/earnings/TNE.AX-H2-2025-earnings_call-348394.html/

¹⁰ WiseTech 2025 Investor Day CEO/Executive Chair speech
https://www.wisetechglobal.com/media/vhffu5my/wtc-2025-investor-day-ceo-and-executive-chair-speeches.pdf

¹¹  Visage blog
https://blog.visageimaging.com/blog/visage-breast-density-ai-classifier-from-conception-to-routine-clinical-use

Disclaimer: