Forty years is a long time to guess. Still, when a national treasury sits down and names the forces that will reshape an economy through the mid-2060s, people listen. This week the latest long-range outlook did something neat and a little awkward at the same time. It put artificial intelligence in the main frame. It left digital assets off the five headline themes. That contrast is the story, not a footnote.
I’ve found that these reports rarely capture every live market. They pick a small set of transitions and then build a budget story around them. Fair enough. The problem is timing. AI agents are already being discussed as economic actors. Payments teams are already testing programmable settlement. If you only read the five-theme list, you would think crypto sits in a side drawer. It does not. It just lives in a different document.
What The New 40-Year Map Actually Says
The seventh long-range report landed on September 21 and looks out to 2065–66. Five transitions sit at the center: the AI revolution, geopolitical fragmentation, the energy shift, population aging, and the continued move toward a services economy. That is a clean list. Clean lists sell. They also leave things out.
Officials expect the economy to more than double in size over the horizon. Real income per person is projected to rise about 55%. Average annual growth still slows, because population growth and workforce expansion lose pace. Productivity then does more of the heavy lifting. That last point is why AI got a dedicated seat this time. The previous 2023 version did not treat it as a core transition in the same way.
The treasurer called AI the most dramatic technological change since the last report and the biggest economic transformation of the current era. Strong language. Useful language, too, if you want investment and skills policy to follow. Treasury’s own write-up covers automation, changing skill needs, and the chance that existing jobs get rewritten rather than simply deleted. It does not pretend every productivity gain arrives on autopilot. Investment, adoption, training, and rules still sit inside the assumptions.
The long-run productivity assumption stayed at 1.2% a year, even while officials admitted AI could move investment and labor outcomes in ways that are hard to price decades ahead.
That 1.2% figure is the quiet hinge. Keep it and the budget story holds. Raise it because machines get better, and the fiscal picture improves. Miss the adoption curve, and the aging and energy transitions look heavier. In my experience, this is where political documents get careful. They name the revolution. They do not bank the upside.
Why Agentic Systems Got Special Attention
The report does not stop at chatbots. It spends time on agentic AI: software that can act with less direct human steering. Officials say these systems are becoming more autonomous, more capable, and more widely used. Performance has already beaten human benchmarks on several tasks. Separate financial-innovation work defines agentic systems as software that can take actions, make recommendations, or perform tasks for people and firms.
That definition matters. An assistant that drafts a memo is one thing. An agent that books a supplier, pays an invoice, and rolls the leftover cash into a short-term instrument is another. The second version needs rails that settle fast, talk to other systems, and accept conditions written into the payment itself. You can build some of that on legacy stacks. You can also see why tokenized money keeps showing up in the side papers.
Perhaps the most interesting aspect is the caution baked into the same files. Fully autonomous financial agents are still described as largely experimental. Reliability is not yet good enough for most real-world money movement. That is an honest line. It also explains the split personality of policy this month: AI is a 40-year theme; crypto-style plumbing is a near-term infrastructure file.
The Five Transitions In Plain Language
Lists can blur. Here is the same set without the brochure tone.
- AI changes how work gets done and how output per hour might rise.
- Geopolitics fragments trade, capital, and technology supply.
- Energy systems shift, with cost and reliability still in play.
- An older population slows labor-force growth and lifts care costs.
- Services keep taking a larger share of activity and jobs.
Crypto is not on that card. That does not make digital markets imaginary. It means the authors organized the next four decades around labor, energy, demography, and machines, not around a new asset class. I think that choice is defensible as a framing device. I also think it underplays the payment layer those machines will need.
Crypto Is Missing From The Poster, Not From Policy
Industry voices noticed the gap immediately. A country director at a major exchange argued that future agents will need financial systems that can support automated activity. If prosperity over 40 years depends on technology adoption and productivity, leaving the money rails off the main chart looks incomplete. That critique is industry-colored. It is not empty.
Look at the calendar. A Financial Innovation Strategy arrived on September 3. It talks about digital assets, stablecoins, tokenization, digital money, and AI-enabled financial services. It is framed as a coordination tool for government, regulators, and firms. That is not the language of a topic that has been dismissed. It is the language of a topic parked in the specialist lane.
Licensing for digital asset platforms and tokenized custody platforms is scheduled to start on April 9, 2027. Payment reforms are expected to treat payment stablecoins as tokenized stored-value facilities. Legal and operational questions around tokenized markets stay under review as commercial use comes closer. Firms already covered by existing financial-services rules face separate obligations now. The dual track is messy. Dual tracks often are.
So the omission is real in the 40-year poster and overstated if you stop at the poster. One document sells a national story. The other builds the pipes. Readers who only scan headlines will miss the pipes. That is how narratives get stuck.
Where AI Agents Meet Programmable Payments
The innovation strategy makes a link the intergenerational report does not put front and center. Advances in agentic systems could lift automated and machine-to-machine transactions. That activity could create demand for payment and settlement systems that execute in real time, interoperate, and support programmable services. In other words, the machine story eventually becomes a money-movement story.
Stablecoins and tokenized markets are one proposed set of rails for that activity. Clearer rules for tokenized stored-value facilities would help, according to market participants. That is a policy ask, not a forecast that agents will necessarily spend crypto. Keep the distinction. Hype collapses those two ideas. Good analysis does not.
Work already underway in payments shows the model in miniature. Tests have paired AI agents with stablecoins for automated payments. Other projects have tried programmatic machine payments on blockchain infrastructure. Domestic payment planning has moved in a similar technical direction, including work on whether existing rails can interoperate with tokenized value and bank money.
I’ve sat with enough product people to know the pattern. First you get a demo that looks slick. Then you hit legal certainty, failure handling, identity, and what happens when the agent is wrong. The strategy paper flags that last problem. Experimental is the right word until those pieces harden.
Central Bank Work On Tokenized Markets
The central bank has kept tokenized finance on its own agenda, separate from the 40-year narrative. In May it released final findings from a wholesale project that tested 20 tokenized asset use cases across issuance, trading, and settlement. The tests used different settlement assets: existing central-bank balances, a pilot wholesale CBDC, commercial-bank deposit tokens, and stablecoins.
Results pointed to possible gains in efficiency, functionality, and resilience. They also left open questions on legal certainty, settlement design, infrastructure, and regulation. That mix should sound familiar. Every serious tokenization file ends in the same four buckets. The romance is the atomic settlement. The work is the law and the plumbing.
Research attached to that program estimates that full adoption of digital finance could generate as much as A$24 billion in annual economic gains. The figure covers tokenized real-world assets, payments, and other applications. It is a modeled estimate, not money already booked. Treat it that way. Modeled billions are useful for sequencing policy. They are not a quarterly earnings beat.
In September the central bank opened another consultation on how its high-value transfer system and fast settlement service could support tokenized markets. Officials are looking at settlement between traditional accounts and tokenized private money, possible stablecoin access to central-bank reserves, and future designs for tokenized reserves. Submissions close on October 30, 2026. Those responses will feed decisions on future system capabilities. A digital financial market infrastructure sandbox is scheduled for the second half of 2027.
| Track | What It Covers | Timing Signal |
| 40-year outlook | AI as a core transition | Through 2065–66 |
| Innovation strategy | Digital assets and AI finance | Coordination now |
| Platform licensing | Digital asset and custody platforms | April 2027 start |
| Market infrastructure | Sandbox for digital markets | Second half 2027 |
| Settlement consult | Tokenized money and RITS | Comments by Oct 30, 2026 |
Read that table twice. The long report names machines. The operational calendar names tokens, licenses, and settlement. If you invest or build, the calendar is the document that pays the bills.
Productivity Math Without The Slogan
Growth slowing because people age is not new. What is new is the hope that software can offset some of that drag. Treasury is not writing a science-fiction budget. It keeps 1.2% productivity as the baseline and then admits AI could change the path. That is conservative. Conservative can be wise. It can also understate a step-change if agents start handling whole workflows.
Skills sit in the middle. If firms adopt tools but workers cannot supervise them, you get expensive toys. If workers can supervise them but systems cannot settle value safely, you get stalled pilots. The missing crypto theme is really a missing settlement theme. Call it tokens or call it upgraded bank money. The economic question is the same: can value move at the speed of the agent?
Geopolitical fragmentation makes that question sharper. Supply chains for chips, models, and cloud capacity are not frictionless. Payment systems that depend on a single overseas stack carry their own risk. Domestic work on interoperable rails is, in part, an insurance policy. Not everyone frames it that way. They should.
What Industry Wants And What Officials Will Give
Market firms want the AI story and the asset story bound together. Officials prefer two folders. One folder is the national narrative: aging, energy, services, machines, fractured geopolitics. The other folder is licensing, stored-value rules, sandbox design, and settlement access. Binding them in one speech would make a louder press cycle. Splitting them reduces political heat around speculative tokens while still moving the rails.
That split has a cost. Talent and capital read headlines. A 40-year report that never says “digital assets” can make a jurisdiction look late even when the specialist papers are not late. Communication is part of industrial strategy, whether finance ministries like that sentence or not.
- Keep the productivity baseline honest and show the AI upside as a range, not a promise.
- Write the payment layer into the same public story as the agents.
- Finish legal certainty work before the 2027 licenses go live.
- Test failure cases for autonomous payments, not only happy-path demos.
- Measure adoption in real firms, not only in lab settlements.
None of that requires a slogan. It requires sequencing. Licenses without settlement design create empty shops. Settlement design without licenses creates gray markets. Agents without either create slide decks.
A Closer Look At The A$24 Billion Claim
Large modeled gains get recycled fast. Pause on this one. The estimate assumes broad adoption of digital finance, not a single coin rally. Tokenized real-world assets, cheaper payments, and cleaner post-trade processes do the work in the model. If legal title stays fuzzy, the number shrinks. If only wholesale desks adopt, the number shrinks. If retail stablecoin use stays thin, the number shrinks.
I would rather see a range with adoption scenarios than a single headline total. Still, the direction is plausible. Markets that settle faster tie up less capital. Assets that can be programmed can be used as collateral with fewer faxes. Those are boring gains. Boring gains compound.
Compare that with the intergenerational growth story. An economy that more than doubles over 40 years while income per person rises 55% is a story about living standards. Tokenization does not need to be a sixth official transition to matter inside that story. It only needs to be part of how investment and payments actually happen.
Risks That Do Not Fit On A Poster
Agentic finance fails in specific ways. An agent can pay the wrong party. It can loop a trade. It can leak data while chasing a cheaper quote. Legacy fraud teams are built for human hours. Machine hours are different. The experimental label in the strategy paper is not decoration. It is a warning about operational risk dressed as product vision.
There is also a concentration risk. If a handful of model providers and a handful of stablecoin issuers become the default pair, resilience claims get thinner. Interoperability language in official papers is the right instinct. Interoperability in production is harder. Standards fights are slow. That slowness is why 2027 dates keep appearing.
Aging and energy already strain the budget. If AI lifts productivity only in a few coastal firms, regional gaps widen. Digital finance will not fix that by itself. It can, at best, lower the cost of moving value into smaller markets. Overpromising here helps no one.
Name the machine revolution in the long report. Fund the settlement work in the short report. Then admit the two files are one system.
How Readers Should Use The Two Documents
If you build products, do not wait for crypto to appear in the next 40-year edition. Watch the April 2027 licensing start, the 2027 sandbox window, and the October 2026 settlement consultation. Those dates change roadmaps. The thematic list changes speeches.
If you allocate capital, separate token beta from infrastructure beta. A modeled A$24 billion opportunity is not a ticker. It is a reason to study custody rules, stored-value treatment, and access to fast settlement. The firms that survive the rulebook will matter more than the first demo video.
If you follow public policy, judge coherence. Is the AI transition funded in skills and compute, or only named? Are payment stablecoins regulated as stored value in a way that banks and non-banks can both live with? Does the high-value system actually accept tokenized private money, or only talk about it? Coherence is the test. Posters are not.
The Human Layer Under The Machine Layer
It is easy to write as if agents replace people. The report is more careful. It talks about jobs being performed differently. That is the version I buy. Supervisors, auditors, and exception-handlers become more valuable when software can move money. The scarce skill is judgment under automation, not nostalgia for paper.
Services already dominate the employment mix. AI lands hardest there, because so much services work is language, scheduling, and reconciliation. Tokenized payments land in the same zone: invoices, escrow, marketplace splits, machine usage fees. Put those two sentences together and the omitted theme looks less like a snub and more like a missed cross-reference.
People will still want simple money. They will not want to know whether the agent used a deposit token or a payment stablecoin. They will want the coffee paid and the refund instant. The policy job is to make the exotic layer invisible and safe. Visibility can wait for wholesale desks.
What I Think Gets Misread Next
Some readers will say the country “ignored crypto.” That is lazy. Some officials will say the theme list already covers technology, so nothing is missing. That is also lazy. The accurate read is narrower. The long-range fiscal story is organized around labor, energy, demography, geopolitics, and machines. The money architecture for those machines is being written next door.
Will the next edition fold digital finance into the AI chapter? Maybe. It should at least mention programmable settlement when it talks about autonomous systems. Leaving that sentence out makes the AI chapter feel unfinished. Including it does not require a coin endorsement. It requires describing how value moves when software starts acting.
I’ve found that markets punish vagueness later, not on release day. Release day is narrative. Two years later, licenses either work or they do not. Settlement either opens or it stays a consultation. That is when the omitted theme becomes measurable.
A Practical Scorecard Through 2027
Here is a simple way to track whether the two files converge.
- Do skills programs mention supervision of automated payments, not only model use?
- Do stored-value rules let payment stablecoins operate without killing bank deposit tokens?
- Does the high-value system publish a real design for tokenized private money?
- Does the sandbox attract live issuance, or only position papers?
- Do productivity statistics start showing adoption in services, not only anecdotes?
Five yes answers would mean the poster and the plumbing finally match. Three yes answers would mean progress with gaps. One yes answer would mean the AI chapter was mostly rhetoric. Rhetoric has a short half-life in markets.
None of this needs drama. The country can name AI as a 40-year force and still build tokenized rails. Other places will do both out loud. Doing both quietly still counts if the dates hold. Dates slipping would be the real omission.
Closing The Loop Without Forcing A Slogan
So where does that leave a reader who cares about both machines and markets? Hold two thoughts. First, the long outlook is right to treat AI as a defining transition. Aging and slower workforce growth make productivity the main lever. Tools that change how work is done belong on that list. Second, autonomous tools that buy, sell, and settle need financial infrastructure that the five-theme poster does not describe. That infrastructure is already in draft across licensing, payments, and central-bank operations.
The gap is editorial more than operational. Editorial gaps still matter because they shape what ministers defend in public. If digital finance stays a specialist file, it can be delayed when attention moves. If it is named as part of the AI transition, delay becomes harder to hide. I would rather see the second version in the next cycle. Not because tokens need a cheer. Because agents without rails are a half-built story.
Forty years from now, nobody will remember the exact five labels. They will remember whether income per person actually rose, whether services work got better or just faster, and whether money moved when software asked it to. The report released this week answers the first questions with projections. The second set of papers tries to answer the last one. Read them together. That is the only way the omission makes sense, and the only way the AI chapter becomes more than a well-written heading.
