The Invisible Hand(out)

– David Wanis, February 2026

This reporting season we are once again reminded that the stock market is not the economy, and with an increasing percentage of the economy in government hands, the divergence can be stark. With economic statistics showing a continued decline in productivity, undermining GDP per capital and real wage growth, it is tempting to overreact to what disasters must await public company investors when all the pressures of higher interest rates, cost inflation and slower growth hit their financial accounts.

To be clear, these macro factors are all headwinds and higher productivity, higher real GDP growth , lower interest rates and lower inflation, are definitely preferred but what we saw in February puts some faith back into the ability for management teams to rise to the task when needed. And right now they are needed.

February was marked by a crescendo of AI doomer stories – mostly about how software will be universally replaced. That is not what we are seeing, and it made us think about a viable alternative scenario. “Capital Deepening” as the economist nerds like to say is about giving people (aka labour) tools which increase their productivity. In knowledge work that is mostly technology – and mostly software. This is being augmented by AI – a new form of technology which allows new leverage of human labour. What we clearly saw during reporting season was pro-active management teams leveraging employees (labour) against software and AI (tools) to drive real productivity growth for shareholders.

White collar / knowledge work businesses are large software users, but they still probably spend 10x as much on employee wages as software. Using AI to increase labour productivity 5-10% p.a, and not touching their 30 years of business process software investments, rather using AI to leverage existing systems, is a low-risk way of delivering real gains from AI. While there were some sensational headlines from Wisetech (WTC) and Block (XYZ) about 30-40% headcount reductions powered by AI, this is not the norm and arguably for these companies more about post-merger cost rationalisations wrapped up in an acceptable AI narrative.

Below are some of the examples across the small cap market we saw. This is in no way a complete list, and the trend of real and objectively tangible efficiency gains was broad based. We think companies doing nothing about productivity would be facing at least 3-5% operating cost increases, possibly more if their business growth is above GDP growth rates:

Australian Clinical Labs (ACL): Lab operating efficiency as measured by panels per FTE increased 10%. Logistics efficiency (driving samples to the labs) increased 13% with route optimisation software. Final testing of AI and optical character recognition (OCR) to automate pathology request forms at better than human accuracy. Operating costs were down 1.2% YoY.

McMillan Shakespeare (MMS): Increased digital interactions to 83%, improved call centre productivity 16% using AI, increased digital processing of NDIS invoices 57% and increased NDIA integrity checks to 95% (again using AI).  Operating costs were up 1.5% YoY.

Smartgroup (SIQ): Operational and brand consolidation plus process digitisation and AI tools have resulted in an 18.5% increase in customers per FTE over two years. Management expect this to continue and margins to expand. Staff expenses were up 3% YoY.

NIB Holdings (NIB): Half of all chat interactions are automated and 86% of Australian resident claims are processed unassisted via automation. An additional productivity focus on directing insured health spending to the most efficient setting, which increasingly is not the traditional private hospitals. Non-marketing operating expense declined by 0.2% YoY

Superloop (SLC): Embedded AI and workflow automation into four key customer support processes, resulting in expensive inbound voice support calls reducing by 30% despite record organic subscriber growth. Operating expenses as a percentage of revenue were reduced from 15.4% to 13.4%.

Flight Centre (FLT): Deployed AI to triage and automate corporate client queries, driving a 13% increase in productivity per full-time employee and allowing a 6% reduction in corporate headcount over the last year. Operating cost ratios hit record lows, driving a 20% profit jump on just 6% revenue growth YoY in the Corporate segment.

Objective Corporation (OCL): Deployed AI-powered capabilities within its RegWorks software, including AI public forms assistants to streamline community engagement and AI case summaries that deliver immediate actionable insights to government case workers. Administration and other operating expenses increased by 6.7% YoY against a 12% increase in recurring revenue.

Bravura Solutions (BVS): Automated core client integration processes and optimized software development workflows to successfully launch major new client deployments in the EMEA and APAC regions without needing to expand developer headcount. Total operating costs fell approximately 1.5% YoY.

Hansen Technologies (HSN): Established dedicated AI Hubs, successfully embedding eight distinct AI solutions to automate complex billing and compliance workflows for Tier-1 telecommunications operators. Segment operating expenses fell 3.2% YoY against a 7.3% increase in operating revenue.

Temple & Webster (TPW): Integrated AI-driven personalization and automation to reshape the cost base, with 80% of all customer support interactions now handled autonomously by AI tools. Fixed operating costs dropped to a record low of 9.4% of revenue (down from 10.5%) YoY.

Accent Group (AX1): Leveraged digital platforms and automated supply-chain and inventory management systems to successfully open 27 new stores and grow wholesale sales by 9.4% without increasing support team and distribution headcount. Cost of Doing Business (CODB) as a percentage of revenue fell 40 basis points to 44.3% YoY.

Tyro Payments (TYR): Automated internal merchant workflows and acquired Thriday, an AI-powered financial management and automated accounting product, to successfully manage a 38% increase in banking users without linearly scaling back-office headcount. Operating expenses reduced 2.9% YoY and have only grown 0.3% p.a over the past three years.

PEXA Group (PXA): Leveraged a custom AI-first approach to automate key internal development workflows, cutting estimated software development timelines by 80% and reducing manual task time by 25% for its internal data and population forecasting tools. Capital expenditure and underlying operating costs were reduced by 12% YoY.

Megaport (MP1): Automated multi cloud application orchestration with 65% of all connections now happening entirely autonomously on its Software Defined Network (SDN). Direct network operating costs were flat as a percentage of revenue YoY.

Iress (IRE): Executed an “Agentic AI innovation” framework to augment enterprise delivery, automate compliance activities, and streamline the software development lifecycle for its core wealth management platforms. Total operating costs decreased 3.8% YoY.

Of course, once again this was “the most volatile reposting season ever” – which really depends on what you own. Nonetheless, here are some stats: The Small Ordinaries saw downward 12-month forward EPS revisions of -1.4%, with small industrials faring worse at 2.4% downgrades. Midcaps fared better (+0.9%) and February was revenge of the large caps with the ASX Top 50 seeing upward revisions of 3%.

Large cap heavy weights such as the banks, BHP, Woodside, Origin, QBE and Woolworths all saw meaningful upgrades.

Within the small cap sectors some of the larger downgrades included TPG (communication), G8 Education and Bapcor (Consumer Discretionary), Graincorp and Inghams (Consumer Staples), Viva Energy (Energy), HMC Capital (Financials), Healius and Neuren (Healthcare – which was the worst performing sector on EPS revisions), Electo-Optical Systems and Austal (Industrials), Siteminder (Information Technology), Iluka (Materials) and Lifestyle Communities (Real Estate).

The bright spots where we saw upgrades included Superloop and Aussie Broadband in Communications who are seeing TPG as the place to gain share now, Tabcorp, Accent Group and Universal Stores in an otherwise tough Consumer Discretionary result season, Ridley and A2 Milk in Consumer Staples, Amplitude and Bannerman in Energy, Regal Partners and Tyro in Financials,  Integrated Diagnostics in a very tough healthcare set of results, Monadelphous, NRW Holdings and Reece were stand outs in industrials. Megaport saw large positive revisions and saw share price collapse with the SaaS names, Bravura and IRESS also saw upgrades. A broad range of metals and mining names saw upgrades and Pexa seems to have got a better handle on costs than market expected in real estate.

We think there are currently three things going on with software stocks, and the risk of AI disruption expanding to other sectors means this may be applicable elsewhere in the future:

1. Valuation: multiples got far too extreme and investors became far too complacent of these multiples. Narratives became more important than fundamentals and the stocks were fragile to bad news. A simple example here is Technology One (TNE) – a stock well known to many small cap investors but long since promoted out of our universe to the ASX top 50. TNE forward EPS growth has for the past 15 years consistently been estimated by the market at around 15% per annum. This level of growth was priced at 14x P/E in 2010 to 2013 and more recently expanded to a multiple of 86x by May 2025. This for a company which is truly high quality – none of the problems we discuss shortly – however making the investment case work at 86x for 15% p.a growth was a stretch.

Source: Bloomberg, Longwave Capital Partners, 28 Feb 2026

2. Quality: a lot of low-quality behaviour by technology and software companies was overlooked while the narrative held up. Unproven financial economics (negative free cash flow), poor operating efficiency, egregious stock-based compensation (SBC) being added back and dilution ignored, questionable capitalisation policies and a focus on adjusted EBITDA pre-SBC as the preferred “earnings” measure. Narrative and recent share price appreciation became an accepted short cut for quality.

3. AI Disruption: this is the pin which is popping the balloon, and the risk is real. It probably won’t play out the way the market thinks, and it could take longer, but the market focuses on the end state (terminal multiple) not the timeframe over which the disruption occurs. This makes “buying the dip” when things are still getting worse, and valuations are still excessive, fraught.