Two mid-sized distributors — call them A and B, a composite rather than real firms — bought the same AI tools in the same quarter. Same market, same suppliers, similar teams. Eighteen months later, B quotes in minutes and ships more orders per person than it ever has. A has a licence bill, and an order desk that works exactly as it did before.
Nothing about the technology explains the difference. Something about the businesses does — and it is the same thing that sits behind one of the most important, and most hopeful, findings in the productivity evidence.
The frontier isn't slowing. The rest are.
Treasury's March 2026 submission on productivity puts it plainly: for most firms, gains come "primarily through diffusion rather than frontier innovation" — the broad adoption and effective use of technologies that already exist [1]. Only 1 to 2% of Australian firms innovate in ways that are new to the world; the Productivity Commission's inquiry into the rest is titled, pointedly, Innovation for the 98% [3].
And the 98% have been absorbing proven ideas more slowly. Linking Australian firm data with international data, Treasury researchers found that between 2002 and 2016 the productivity of global frontier firms grew 60%, Australian frontier firms 25%, and other Australian firms 15% — and that Australian firms now catch up to the global frontier more slowly than they used to, most strikingly in services [2].
Year by year, those differences look small. Compounded, they are not. Pick a pace and a horizon and watch what the difference does.
The Pace Gap
A Small Difference in Pace Becomes a Wide Gap
What happened — and what it compounds to
What happened
Productivity growth, 2002–2016
- Global frontier firms+60%
- Australian frontier firms+25%
- Other Australian firms+15%
Based on Australian Government Treasury data (Working Paper 2022-01). Average labour productivity of each group, indexed to 2002; firms move between groups over time.
What it compounds to
27%
Gap opened
At 1.0% a year for 10 years, output per worker rises 10%. A firm moving at the frontier’s pace rises 40% — opening a gap of 27% from a standing start.
Illustrative. Paces are the compound annual rates implied by the 2002–2016 growth above; “stretch” is a what-if. Both paths start level, so the gap shown is the ground that opens from today.
| Your pace | 5 years | 10 years | 15 years |
|---|---|---|---|
| Typical (1.0% a year) | 13% | 27% | 42% |
| Local leader (1.6% a year) | 9% | 19% | 30% |
| Stretch (2.5% a year) | 5% | 9% | 14% |
| Frontier (3.4% a year) | 0% | 0% | 0% |
The model carries the encouraging point. The gap is built from pace, not from a fixed disadvantage — and pace is something a business can change. Moving from the typical pace to a stretch pace means losing about a third as much ground over a decade.
AI is widening the gap — and opening a door
Generative AI could have been the great leveller: the same tools, available to everyone at once. Treasury describes its upside as "potentially very significant, but uncertain" [1], and so far the adoption data say the uncertainty is being resolved unevenly. Across OECD countries the share of firms using AI more than doubled in two years, to 20.2% in 2025 — but 52.0% of large firms use it, against 17.4% of small firms [4]. In Australia, 43% of small and medium businesses reported some level of AI adoption over the December 2025 to February 2026 quarter [5].
Adoption is not the same as benefit, and that is where the gap is really set. Studying earlier waves of general-purpose technology, economists found that technologies like AI "enable and require significant complementary investments" — new processes, business models and skills — and that the payoff shows up only once those are made [6]. The return arrives when the work is redesigned, not when the licence is signed. That is the whole difference between distributor A and distributor B.
The door is open, though. The fastest growth in AI use across the OECD came in industries that had been behind — up 59.1% in a year in construction and 62.5% in accommodation and food services [4]. And among Australian businesses already using AI, broad use across the business is rising while one-off experiments fade: those who have committed are "doubling down" [5]. Late starters are moving quickly, and the ones who commit keep going.
Frontier distance: four reasons firms fall behind
If invention isn't the constraint, what is? The evidence points to four things that decide how quickly a business absorbs what already works. Together they set its frontier distance:
- Capability. The Productivity Commission calls management capability "a critical determinant of adoption of best practice"; a 0.1 lift in its management capability score (on a 0 to 1 scale) is associated with 6.2% higher labour productivity [3].
- Complements. The process, product and skills redesign a new technology needs before it pays [6].
- Connection. The Commission sees signs that the "diffusion machine" has weakened, with slower growth in labour mobility, business entries and exits, and investment in "capital with brains" such as equipment, R&D and software [3].
- Challenge. Declining business dynamism and competitive pressure appear to account for half to three-quarters of the slowdown in Australian firms' catch-up [2].
Frontier Distance
Four Reasons Firms Fall Behind — and the Way Back
Tap a reason to see the signals and the first move
Reason 1 of 4
Capability · Managing for improvement
The management habits that let a business notice where it stands and act on it: measuring the work that matters, setting clear targets, and turning what one team learns into how everyone works. It is the capacity to absorb a better idea — and every other move depends on it.
Signs it is holding you back
- Performance is reported, but rarely changes a decision
- A better method stays with the team that found it
- Nobody can say how far you are from the best in your field
The first move
Choose three operating measures that genuinely matter, review them weekly with named owners, and make “what did we change because of this?” a standing question.
In the example
Distributor B began publishing order cycle time and pick accuracy by site every Monday, and asked each site lead one question: what will you copy from the best site this week?
None of the four is about buying technology. Each is about how a business learns. Distributor B did not have better tools than A. It managed for improvement, rebuilt the work around the tool, borrowed know-how from outside, and held itself to the best it could see.
Where do you sit?
Frontier distance is easy to underestimate from the inside. The Productivity Commission found Australian businesses "may not be aware of how far they lag behind": in 2020, only 2.4% said shortages of management skills affected their core activities, even as measured capability was weak [3]. Eight questions will place you, and name the lever to pull first.
Where does your business sit?
Eight questions · 0/8 answered
Capability
Complements
Connection
Challenge
Closing the distance
The most hopeful line in the evidence is also the simplest: diffusion "has the scope to lift the performance of millions of businesses" [3]. Wherever you landed, three moves hold:
- Start with your biggest lever, not your biggest purchase. A tool lands on top of whatever capability, redesign, connection and challenge already exist. Strengthen the weakest of those first and every later investment works harder.
- Learn from the nearest frontier. You do not have to reach the global leaders to make progress — you have to improve faster than you have been. The quickest lessons usually sit with a peer, a supplier or a customer who already runs your core process better.
- Expect the dip before the payoff. Redesign is an investment that rarely shows up in the numbers at first; the gains are harvested later [6]. Judge it on the end-to-end outcome over a year, not on the first quarter.
The diffusion gap is not a verdict on any business. It is a measure of how much proven improvement is still sitting on the table. So leave with one question: which business near you already does your core work better — and what would it take to learn from it this quarter?
Sources
- Australian Government — The Treasury. Treasury Submission to the Select Committee on Productivity in Australia. March 2026.States that 'For most firms, productivity gains arise primarily through diffusion rather than frontier innovation', defines diffusion as 'the broad adoption and effective use of existing technologies', and describes the upside of technological innovations, 'including artificial intelligence', as 'potentially very significant, but uncertain'.View source
- Australian Government — The Treasury. Reaching for the stars: Australian firms and the global productivity frontier (Treasury Working Paper 2022-01). February 2022.Andrews, Hambur, Hansell & Wheeler. Linking OECD-Orbis and Australian BLADE microdata, the paper finds that while 'productivity of those firms in the global frontier 60 per cent higher in 2016 compared to 2002, Australian frontier firms' productivity is only 25 per cent higher, and other 'laggard' firms 15 per cent higher' (each group indexed to 2002). 'Australian firms catch up to the global frontier more slowly than previously'; the divergence is 'particularly striking in the services sector'; and declines in business dynamism and competitive pressures 'appear to account for half to three-quarters of the slowdown' in catch-up.View source
- Productivity Commission. 5-year Productivity Inquiry: Innovation for the 98% (Inquiry Report No. 100, Volume 5). 7 February 2023.New-to-the-world innovation 'relates to only 1 to 2% of Australian firms'; diffusion 'has the scope to lift the performance of millions of businesses'. Australian businesses 'may not be aware of how far they lag behind', and management capability is 'a critical determinant of adoption of best practice'. A 0.1 increase in the overall management capability score (range 0 to 1) 'is associated with a 6.2% increase in labour productivity', yet in 2020 'only 2.4% of Australian businesses' said management-skill deficiencies affected their core activities. There are 'signs that the 'diffusion machine' has weakened', with slower growth in labour mobility, business exits and entries, and investment in 'capital with brains'.View source
- OECD. AI use by individuals surges across the OECD as adoption by firms continues to expand. 28 January 2026.'In 2025, 20.2% of firms reported using AI, up from 14.2% in 2024 and 8.7% in 2023, meaning adoption has more than doubled over the past two years.' Adoption 'remains highly uneven across firms of different sizes: 52.0% of large firms use AI, compared with 17.4% of small firms.' Growth was strongest in industries that previously lagged: 'Year-on-year growth reached 62.5% in accommodation and food services, and 59.1% in construction.'View source
- National AI Centre — Department of Industry, Science and Resources. AI adoption insights: December 2025 to February 2026. 7 May 2026 (updated 3 June 2026).'Across the December to February quarter, 43% of Australian SMEs reported some level of AI adoption.' Broad adoption, 'where AI is embedded across multiple parts of the business', reached its highest level in seven months while 'Limited, one-off use is declining' — 'those who have committed to AI are doubling down.' Monthly survey of different businesses each month (snapshots, not a panel).View source
- National Bureau of Economic Research. The Productivity J-Curve: How Intangibles Complement General Purpose Technologies (Working Paper 25148). October 2018.Brynjolfsson, Rock & Syverson: 'General purpose technologies (GPTs) such as AI enable and require significant complementary investments, including co-invention of new processes, products, business models and human capital.' Because these investments are intangible and poorly measured, productivity growth is underestimated in a new technology's early years and rises later 'when the benefits of intangible investments are harvested'.View source