Jobs in the AI Era

Building what AI cannot. Powering what AI needs. Owning what gets built. The labour-market case for the SBC: Memo 19 costs the build, Memo 20 captures the return, Memo 21 captures the counterfactual, this memo captures the workforce.

1. The question this memo answers

Artificial intelligence and robotics will displace a material share of Australian work over the next 10–15 years. The current external projections — from Goldman Sachs, the World Economic Forum, McKinsey, and Jobs and Skills Australia — converge on a single direction even where they disagree on magnitude: a substantial fraction of the existing Australian workforce, particularly in clerical, administrative, and routine cognitive roles, will see their work either automated outright or fundamentally restructured. The honest assessment is that this is happening already and will accelerate.

The question this memo answers is not whether AI will displace work — it is what Australia is going to do about it.

The answer this memo develops is that the Sovereign Build Corporation programme is the largest available Australian response. It is a continental-scale public-works programme deliberately concentrated in the categories of work that AI and robotics cannot perform — construction, civil engineering, trades, manufacturing operations, agricultural production, and infrastructure operations and maintenance. The programme employs hundreds of thousands of Australians in AI-resistant roles for two decades. At the same time, the same infrastructure delivers the cheap renewable electricity, the sovereign AI compute campus capacity, and the export-grade renewable supply that make Australia the natural destination for the AI economy the displaced workers' children will work in.

Build what AI cannot. Power what AI needs. Own what gets built.

This memo sets out the displacement projections, the SBC workforce arithmetic, the AI-energy-infrastructure synergy, the critical fiscal window, the sovereign asset ownership argument, and the New Deal historical analogue. The methodology stance is the same as Memos 19, 20, and 21: every assumption named, every confidence grade explicit, no macro multipliers applied.

1.1 Confidence grades

Same four-tier system as the rest of the memo series:

• Locked — figure sourced from current government commitments or published data
• Working estimate — figure derived from a defensible analogue or external analysis
• Under MMA scoping — flagged as a working assumption for this memo
• Strategic risk (not monetised) — items where a dollar value would mislead more than inform

1.2 What this memo does not claim

This memo does not claim that AI will eliminate most Australian work. It does not claim the SBC programme will employ every displaced worker. It does not apply macro multipliers to inflate the workforce numbers. It does not claim that physical infrastructure work is permanently immune to automation. It claims something narrower and more defensible: a substantial AI displacement is coming on a 10–15 year horizon; the SBC programme is the largest Australian counterweight available at the scale and on the timeline the displacement requires; and the labour-market case for building the SBC is on top of every other case the trilogy of Memos 19, 20, and 21 has already made.

2. The displacement projections — what the evidence says

External projections of AI workforce impact vary widely on magnitude and on timing. The honest summary is that the range of estimates is wide, the methodologies are different, and the actual outcome will depend on AI capability trajectory, adoption rates, and policy responses none of which are knowable in 2026. What the projections agree on is direction: a material share of existing work will be either automated or restructured by 2035.

2.1 Headline projections

2.2 What the projections agree on

Across this range of methodologies, four claims hold up consistently:

1. The displacement is real and material. Even the most conservative estimates (JSA's 13% fully automatable by 2050) imply hundreds of thousands of Australian jobs structurally affected. The higher estimates imply over a million.

2. Clerical, administrative, and routine cognitive work is most exposed. Bookkeeping, basic legal and accounting tasks, customer service, data entry, basic coding, marketing operations, administrative coordination. These are concentrated in coastal capital cities.

3. Construction, trades, manufacturing operations, agriculture, healthcare, and infrastructure roles are materially less exposed. Robotics improves productivity in some of these areas — warehouse logistics, assembly — but cannot replace civil engineering, structural fitting, electrical trades, plumbing, agriculture in variable conditions, healthcare delivery, or large-scale infrastructure operations.

4. The transition window is 5–15 years. The displacement is happening already (tech layoffs explicitly cite AI; entry-level white-collar hiring is stagnating). The full impact accelerates through 2030 and beyond.

2.3 The Australian-specific picture

Australia enters the AI displacement period with a workforce of approximately 14.5 million people. The McKinsey Australia estimate of 1.3 million workers needing transition by 2030 represents approximately 9% of the workforce on a five-year horizon. This is a manageable number with deliberate policy, and an unmanageable one without it.

The exposure is geographically concentrated. Sydney, Melbourne, and Brisbane carry approximately 60% of Australia's white-collar clerical and administrative employment. These are the same coastal capital concentrations facing simultaneous housing affordability stress, congestion, and the climate and energy cost pressures Memo 21 catalogues. The without-SBC counterfactual places these workers in cities that are simultaneously becoming more expensive to live in and offering fewer secure jobs.

The opportunity is regional. Construction, manufacturing, agriculture, and infrastructure operations are activities that scale more naturally in regional Australia than in coastal capitals. The SBC programme — 200+ corridor towns, 11 intersection cities, continental construction across 23,200 km — is geographically aligned with where the AI-resistant work needs to happen. The same programme that builds Australia's renewable energy backbone builds Australia's regional labour market.

3. The SBC workforce — what the programme actually employs

The SBC programme employs Australians across four distinct categories: construction, manufacturing, operations, and apprenticeships. The figures in this section are working estimates anchored against the locked capex schedule in Memo 19 and the manufacturing revival industry revenue in Memo 20 §3.1. Every figure is named at the working-estimate confidence grade.

3.1 Construction workforce — building the corridor

The SBC programme builds approximately 23,200 km of integrated continental infrastructure (corridor + Alice Hub) at a programme capex of $750–1,340 billion over 20+ years (Memo 19). Construction labour share of major Australian civil infrastructure projects typically sits in the 30–40% range. Average loaded labour cost is approximately $130,000 per worker-year on programmes of this scale.

This is the headline jobs number. Approximately 150,000 to 300,000 Australians employed in construction at the programme's peak intensity in the late 2030s. These are AI-resistant roles in their entirety: structural concrete, civil engineering, electrical trades, plumbing, surveying, project management, occupational health and safety, logistics coordination on construction sites, equipment operation. AI can plan, optimise, and schedule construction work. AI cannot pour concrete, lay steel, or commission a substation.

3.2 Manufacturing workforce — building the sovereign industrial base

Memo 20 §3.1 establishes the manufacturing revival industry revenue at $93–162 billion per year at maturity, distributed across aluminium smelting return, green steel, chemicals and fertilisers, cement and glass and ceramics, battery cell manufacturing, and other heavy industry. Australian manufacturing output per worker sits in the range of $300,000–$500,000 per year depending on sub-sector.

This is the largest single category in the SBC employment profile. Hundreds of thousands of Australians employed in manufacturing across sovereign rail mill, OCTG tubular mill, precast megafactory, HVDC component manufacturing, cable-laying vessels, battery cell production, processed minerals, and the sovereign defence and space supply chains. These are AI-resistant in the engineering and operational layers — robotics increases productivity, but the design, the supervision, the quality assurance, the maintenance, and the iterative improvement require Australian engineers and operators.

3.3 Operations and maintenance workforce — running the network

Once the corridor and the Alice Hub are operational, the SBC requires a permanent operations and maintenance workforce: rail and maglev operators, water network operators, HVDC transmission technicians, Alice Hub PHES operators, AI campus utility operators, agrivoltaic farm operators, and supporting administrative roles.

These are permanent, AI-resistant, regionally distributed roles. Many of them are skilled-trade and engineering roles that pay well and that exist in regional locations near the corridor towns rather than concentrated in coastal capitals.

3.4 Apprenticeships and training pipeline

A programme of this scale cannot be staffed from the existing trained Australian workforce alone. The SBC requires a substantial apprenticeship and training pipeline activated from Phase 0 onwards.

Over a 20-year programme this is approximately 400,000–700,000 Australians passing through a structured trade or engineering apprenticeship — many of them in regional locations. This is the largest deliberate apprenticeship expansion in Australian history. The civil engineering and trades workforce in Australia has not been at the scale required since the post-war infrastructure boom. The SBC is what rebuilds it.

3.5 Local service workforce in corridor towns and intersection cities

The SBC corridor activates 200+ corridor towns and 11 intersection cities. As those settlements populate from current levels to mature populations (corridor towns averaging ~5,000 residents, intersection cities averaging ~80,000), they require the local service workforce every Australian town requires: schoolteachers, nurses and GPs, bakers and butchers, cooks and café owners, plumbers and electricians and mechanics, childcare and aged-care workers, retail and hospitality, local government workers, police and fire and ambulance, postmasters, librarians, sports coaches, allied health professionals.

These workers are enabled by the SBC programme but are not directly employed by the SBC. The framing mirrors Memo 20 §3: the SBC creates the corridor, the corridor activates the towns, the towns generate the local employment. The workforce sits in the same Tier 2 / enabled-industry category as the manufacturing revival (Memo 20 §3.1) and the agricultural production uplift (Memo 20 §3.2).

These are AI-resistant roles for the same reasons construction is. Teaching requires physical classroom presence and judgment about child development. Nursing requires physical care under variable conditions. Café work, retail, trades, childcare, aged care — all require physical presence, integration of multiple skills, and accountability at the point of service delivery. Robotics will improve productivity in some of these roles; none of them automate to zero on the 20-30 year horizon. The corridor towns become the natural home of the AI-resistant local economy.

3.6 Direct agricultural workforce

Memo 20 §3.2 establishes the agricultural production uplift at $39–77 B/yr industry revenue at maturity, distributed across Murray-Darling drought-proofing ($6–9 B/yr), new inland agricultural production ($10–20 B/yr), agrivoltaic productive country ($20–40 B/yr across 13.4 M hectares), and northern catchment co-benefits ($3–8 B/yr). Australian agriculture currently generates approximately $80 B/yr at farmgate with approximately 330,000 direct workers, giving an average of approximately $240,000 farmgate revenue per direct worker.

The agricultural workforce is materially AI-resistant. Australian agriculture has automated extensively over the last century — broadacre cropping is largely mechanised, livestock systems have adopted electronic ID and digital management — and the workforce has still stabilised at approximately 330,000 because the work that remains is the work that cannot automate easily: livestock husbandry under variable conditions, horticulture, viticulture, market gardening, equipment operation in variable terrain, and the on-farm decision-making that integrates weather, soil, market, and livestock condition. Agrivoltaic horticulture in particular is labour-intensive — vegetables, fruit, and intensive systems under PV panels can run at 0.02-0.03 workers per hectare versus 0.005-0.01 for dryland grazing — so the agrivoltaic workforce could run materially higher than the average ratio suggests. The working estimate above uses the average ratio for defensibility.

3.7 Food processing and agricultural supply chain workforce

Agricultural production at $39–77 B/yr industry revenue activates a downstream food processing and supply chain workforce. Australian food and agricultural manufacturing currently employs approximately 270,000 workers downstream of farming. Applying a similar downstream-to-direct ratio (approximately 0.5-0.8 downstream workers per direct ag worker) to the SBC-enabled agricultural activity:

This workforce sits primarily in corridor towns and intersection cities — food processing follows agricultural production geographically, and the SBC corridor's freight and refrigerated rail capacity (Memo 20 §2.3) makes inland food processing competitive with coastal alternatives for the first time in decades.

3.8 Town construction and amenity build workforce

A point that the earlier sections did not count, and that requires explicit treatment: the corridor towns and intersection cities have to be built. The SBC capex in Memo 19 covers the corridor structure, Alice Hub, the trunk water and energy infrastructure, the rail and HVDC and fibre on the corridor. It does not cover the residential construction, the schools, hospitals, retail centres, community facilities, local roads, and amenity infrastructure that the corridor towns require as they populate. That construction is delivered by state government, local government, private developers, and individual builders — enabled by the SBC corridor but funded and built separately.

This is a second wave of construction employment running parallel to the SBC corridor build. Where the SBC corridor workforce is concentrated on the spine and the spurs, the town construction workforce is geographically distributed across all 211 settlements. Where the SBC corridor build winds down around year 18-20 of the programme, the town construction build continues — the towns keep growing for decades after the corridor is commissioned. Both workforces are AI-resistant for the same reasons §4 develops.

3.9 Skilled migration as scale supplement

A programme of this scale cannot be staffed from the existing Australian workforce alone, even with the apprenticeship pipeline of §3.4 running at peak intensity. The arithmetic is direct: Australia currently has approximately 1.3 million workers in construction. Adding 150,000-300,000 peak SBC corridor construction workers plus 80,000-180,000 peak town construction workers represents a workforce expansion of approximately 20-40% over current levels. Skilled migration is the supplement that closes the gap during peak build years.

The Snowy 1.0 analogue is exact. Snowy 1.0 employed approximately 100,000 workers cumulatively over its 25-year build, of whom roughly 70% were migrants from 30+ countries — post-war displaced Europeans, mostly. They built the infrastructure, married Australian women and other migrant women, raised families in Khancoban, Cabramurra, Talbingo and Adaminaby, became Australian. It was the largest single integration of new Australians into the workforce in the country's history. It worked.

The SBC at continental scale will require comparable skilled migration to staff the peak build period. The honest framing is that Australia takes the best displaced workers globally — AI-displaced skilled workers from peer economies, climate-displaced skilled workers from Pacific island nations and similar geographies, the high-skilled global construction and engineering workforce — and integrates them into the SBC programme. They contribute their labour to building Australia, raise families in the corridor towns, become Australians. Their integration is morally clean because the source economies are themselves under AI/climate/political pressure; the SBC offers opportunity, not predation. It is selective on contribution capacity because Australia's existing immigration framework is applied at scale; merit-based skilled migration is what Australia already does well. It expands Australia's tax base and broadens cultural and economic capability in ways the country has not deliberately attempted since the 1950s-1970s.

The skilled migration discussion is treated qualitatively rather than as a separate workforce line because the migrants are not additional to the workforce figures in §3.1-§3.8 — they are how those figures are filled at peak. The structural significance is in the population expansion that supports the corridor towns and the new tax base, not as a separate jobs category.

3.10 SBC workforce summary

Approximately 1.25 to 2.36 million Australian jobs at programme maturity. Directly employed by the SBC: the corridor + Alice Hub construction workforce, the operations and maintenance workforce, and the apprenticeships flowing through the SBC's training pipeline. Enabled by the SBC: the manufacturing revival, the agricultural production uplift, the food processing and supply chain, the local services in corridor towns, and the town construction itself. The figure comfortably exceeds the McKinsey Australia displacement projection of 1.3 million workers needing transition by 2030. The SBC is at the scale the displacement requires, with deliberate room for the population to expand.

4. Why these jobs are AI-resistant

Not every job survives every technology shift. Australian agricultural employment has declined from 25% of the workforce in 1900 to under 3% today because mechanisation, fertiliser, and selective breeding made one farmer as productive as ten previously had been. The jobs that survive technology shifts are jobs where physical presence, judgment under variable conditions, integration of multiple skills, and accountability for outcomes are required at the point of work. The SBC programme's workforce is concentrated in exactly those categories.

4.1 Why construction is AI-resistant

A construction site is a variable, contested, physical environment. No two pours are the same. No two sites have the same geotechnical profile, the same weather window, the same logistics access. AI can optimise scheduling. Robotics can perform repetitive sub-tasks (bricklaying, rebar tying) in controlled settings. What AI and robotics together cannot do is integrate the dozen different trades, the variable site conditions, the regulatory requirements, the safety judgments, and the accountability for delivery into a constructed asset.

Civil engineering execution is the canonical example of physical work that AI assists but does not replace. The SBC programme's 23,200 km of construction sits squarely in this category.

4.2 Why manufacturing operations are AI-resistant at scale

Australian manufacturing has been substantially offshored over four decades because the cost structure made it uncompetitive. The thing that displaced Australian manufacturing workers was not robots — it was cheaper labour and cheaper energy in other countries. The SBC programme reverses both of those drivers: sub-10c/kWh electricity at scale, and a sovereign-content mandate that creates the industrial demand for Australian production. With those drivers in place, manufacturing employment grows.

Robotics is part of the production process. It always has been. But the design of the production line, the iteration on yield improvement, the supervision of quality, the handling of variable inputs, and the maintenance of the equipment all require Australian engineers and operators. The category that AI most automates in manufacturing is the administrative layer — order processing, inventory management, scheduling — not the physical layer.

4.3 Why operations and maintenance are AI-resistant

Network operations require humans because networks fail, and when they fail, the response requires physical presence with judgment. A transmission line goes down — it gets repaired by linesworkers in trucks. A pump fails at Alice Hub — it gets fixed by mechanical fitters with spare parts. A maglev guideway needs scheduled inspection — it gets walked by qualified inspection technicians. AI and remote sensing improve fault detection and predictive maintenance, but the response is physical and human.

4.4 What the SBC programme does NOT employ in scale

The memo's argument has to be honest about what the SBC programme does not employ at large scale. The programme does not employ large clerical, customer service, administrative coordination, or basic professional services workforces. These are exactly the categories most exposed to AI. The SBC programme is not a direct replacement employer for the workers AI displaces from their current roles. Workers in displaced categories will need to retrain into one of the SBC's employment categories — construction trades, manufacturing operations, infrastructure operations, engineering — for the programme to be their employment path.

The apprenticeship and training pipeline in §3.4 is the mechanism that enables this transition. A 35-year-old administrative coordinator in Western Sydney does not become a fitter and turner by signing up at a recruitment website. They become one by enrolling in a structured 3-4 year apprenticeship programme with paid training, regional placement, and a credible career pathway at the end. The SBC is the entity that has the scale, the programme demand, and the timeline to commit to that pipeline.

4.5 The retooling pathway — more towns, more farmland, more options

The argument above describes the trades and engineering apprenticeship pathway. But it is not the only pathway, and for many displaced workers it is not the most realistic one. A 45-year-old administrative coordinator with a mortgage and two teenagers in school cannot easily commit to a 3-4 year fitter-and-turner apprenticeship with paid-but-reduced income. A 38-year-old customer service team leader with a young family has skills that translate poorly to a construction site. The trades-apprenticeship pathway works for some displaced workers, particularly younger ones, but not for the broader displaced workforce.

The corridor towns and the new agricultural areas open a much wider set of retooling pathways. Every one of the 200+ corridor towns and 11 intersection cities needs the local service workforce of §3.5: schoolteachers, nurses, baristas, café owners, retail managers, hairdressers, mechanics, small-business operators, real-estate agents, financial advisers, child-care workers, allied health professionals, hospitality workers, local government officers. These are jobs that many displaced workers can move into directly with existing skills or short-course retraining, not 3-4 year apprenticeships. The administrative coordinator becomes a school administrator. The customer service team leader becomes a real estate agent or a café owner. The bookkeeper becomes a small-town accountant. The IT operations specialist becomes a regional council coordinator. The pathway widens from "become a tradie" to "rebuild a life in a regional town that needs your skills."

The agricultural workforce of §3.6 opens a parallel set of pathways. Agrivoltaic farm management, irrigation system management, agricultural agronomy, livestock systems, horticulture, viticulture — these are categories where mid-career retraining works. The Murray-Darling drought-proofing in §3.6 brings back into production agricultural areas that have been contracting for decades; the agrivoltaic productive country of 13.4 M hectares is essentially new agricultural land being created at continental scale. Both are absorbing workforce, not shedding it.

The maglev lifestyle unlock is the structural enabler that makes these regional pathways actually work as life choices. Phase 0 maglev at 600 km/h delivers Newcastle-Sydney in 15 minutes, Sydney-Melbourne in 90 minutes, Melbourne-Brisbane in 3h 50min direct (Memo 20 §2.4). A corridor town in the Hunter Valley is 30 minutes from Newcastle beaches. An inland NSW corridor town is 1.5-2 hours from Sydney. An Albury corridor town is 45 minutes from Melbourne. A weekend at the beach from an inland corridor town becomes the same kind of trip as a weekend in the Blue Mountains from Sydney is today — cheap, fast, routine. The corridor towns stop being isolated regional outposts and become connected regional towns where you live cheaper, raise kids in safer streets, work in well-paid SBC-enabled employment, and access coastal lifestyle on weekends. That changes the proposition of regional life decisively from what it has been for the last fifty years.

The combined effect is that the SBC programme widens the retraining pathway from "narrow apprenticeship to a small set of trades" to "broad relocation to a corridor town with connected coastal lifestyle, where many existing skills translate, and many short retraining pathways are available." This is what makes the AI-displacement-to-AI-resistant-employment transition realistic at population scale, rather than a policy ambition that only works for some subset of displaced workers.

5. The AI–energy–infrastructure synergy

The SBC programme employs the workforce that AI cannot replace. The same programme also builds the infrastructure that the AI economy requires. This is the strategic logic of the memo: Australia simultaneously builds what AI cannot, and powers what AI needs.

5.1 What the AI economy requires

The AI economy — training large models, operating inference at scale, supporting autonomous systems, providing the compute backbone for the world's productivity infrastructure — requires four inputs that are increasingly the binding constraint:

1. Cheap reliable electricity at scale. Frontier AI compute demand is growing at approximately 25-40% per year. Hyperscaler data centre projections require approximately 150-300 GW of additional global compute capacity by 2030. Electricity availability and cost are the binding constraint on where that compute is built. 2. Water for cooling. Modern AI compute campuses use evaporative cooling or closed-loop water systems requiring substantial water supply. 3. Low-latency fibre connectivity. AI services require sub-50ms latency to major user populations. 4. Political stability and aligned regulatory frameworks. AI compute is increasingly subject to national-security export controls; the jurisdiction matters.

Australia has exactly one of these four inputs at scale today (political stability). The SBC delivers all four.

5.2 What the SBC delivers to the AI economy

• Sub-10c/kWh consumer electricity (Memo 20 §2.1, §2.2). Continental desert PV at 1,000+ GW potential, Alice Hub storage at $1.33/kWh storage cost vs Snowy 2.0 at $34/kWh, HVDC backbone integrated into the corridor. The cheapest large-scale grid electricity in the developed world.
• Cooling water infrastructure (Memo 5 §11.5). Alice Hub aqueduct co-location provides AI campus cooling water interfaces at marginal cost.
• Fibre backbone on the corridor. The Phase 0 spine and continental corridors carry integrated fibre, providing terrestrial low-latency connection to all major Australian population centres and onward subsea connections to Singapore, Indonesia, Japan, and India.
• Australian sovereign-aligned regulatory framework. The SBC compute campus operates under Australian law, Australian privacy framework, and Australian export-control alignment with allies.

5.3 What this is worth — the AI compute industry

Memo 20 §2.6 establishes AI compute and data services revenue at $15–20 B/yr at maturity as a working figure, assuming Australia captures 10–15% of regional AI compute capacity. This is not "AI superpower" hyperbole — it is a defensible working figure for what Australia can credibly deliver. The conservative case at 5% regional share is still $5 B/yr in direct SBC revenue, before counting the indirect activation effects on the broader Australian technology sector.

The strategic outcome is bigger than the dollar figure. Australia becomes a destination for the AI economy's compute infrastructure rather than a consumer of services hosted elsewhere. The workforce that operates and maintains the AI campus capacity is Australian. The data centres and the cooling water and the cable landings are on Australian soil under Australian law. The workforce that builds and operates the supporting infrastructure — the energy generation, the transmission backbone, the water systems — is Australian.

5.4 The ironic synergy, stated honestly

The same SBC programme that creates 1.25 million+ Australian jobs in the categories AI cannot perform also creates the physical infrastructure that the AI economy requires to operate. The workforce that builds the SBC infrastructure today is the workforce whose children work in the AI economy the SBC infrastructure makes possible. The displacement and the response are integrated into a single national programme. This is the kind of nation-building integration the country has not attempted since Snowy 1.0.

5.5 The maglev lifestyle unlock

A point that makes the corridor town economy actually viable as a place people choose to live: cheap, fast maglev passenger service. Phase 0 maglev at 600 km/h sustained delivers Newcastle-Sydney in 15 minutes, Sydney-Melbourne in 90 minutes, Melbourne-Brisbane in 3h 50min direct (Memo 20 §2.4). Target fare $10-20 per segment versus the High Speed Rail Authority's confirmed $31 per segment for slower conventional rail. An inland corridor town in NSW is 1.5-2 hours from Sydney; a corridor town in the Hunter Valley is 30 minutes from Newcastle beaches; an Albury corridor town is 45 minutes from Melbourne.

This changes what a corridor town can be. Through the post-Whitlam era, "regional Australia" has been associated with limited employment, limited services, limited cultural and lifestyle access, and difficulty of physical movement to and from coastal capitals. The SBC maglev unlock breaks that pattern. A weekend at the beach from an inland corridor town becomes the same kind of trip as a weekend in the Blue Mountains from Sydney is today — cheap, fast, routine. Working professionals can live in a corridor town and visit family in Sydney every weekend without it being a planned event. Children growing up in corridor towns access coastal lifestyle and cultural amenity as a routine part of life, not as an annual holiday.

This is the structural innovation that makes the labour-market argument of §3 and §4.5 actually work at scale. Without cheap fast inland-to-coast connectivity, corridor towns reach a certain size and stall — Albury, Wagga Wagga, Bathurst illustrate the historical pattern. With cheap fast maglev connectivity, corridor towns can grow to substantial populations because the lifestyle proposition is competitive with coastal capital living: cheaper housing, safer streets, well-paid SBC-enabled employment, and coastal lifestyle on the weekends. The corridor towns become the natural home of the AI-displaced workforce that retools into the regional economy.

6. The critical window — why now matters

Timing matters. The fiscal capacity to build the SBC exists today and erodes as AI displacement accelerates. This section sets out why the window for action is the next five years, not the next twenty.

6.1 The current fiscal position

Australia enters the AI displacement period with substantial fiscal capacity by developed-economy standards:

• Federal deficit projected at approximately $28–34 billion for 2026-27 — manageable at approximately 1.0-1.2% of GDP
• Unemployment at approximately 4.2-4.5% — within the technical full-employment range
• Federal debt at approximately 35% of GDP — substantially below most peer economies
• Inflation moderated to approximately 2.5-3.5% range
• Australian superannuation pool approximately $4.5 trillion AUM, with substantial unallocated infrastructure capacity
• Strong sovereign credit rating with substantial issuance capacity at competitive rates

This is the strongest fiscal starting position any major OECD economy has for a programme of this scale. It is also a position that depends on a high-employment labour market generating strong tax revenue. If unemployment moves to 6-8% under AI displacement pressure, every line above deteriorates simultaneously.

6.2 The deterioration trajectory under AI displacement without SBC

Without the SBC programme, the AI displacement scenario plays out as follows:

These are working estimates at the under-MMA-scoping confidence grade — but the direction is the consensus view among Australian economists who have modelled AI displacement scenarios. Every dollar of additional welfare spending and every dollar of lost tax revenue is a dollar that cannot be allocated to the infrastructure programme that would have prevented the displacement spiral in the first place.

6.3 The deterioration trajectory under AI displacement with SBC

With the SBC programme in place, the same period plays out very differently:

• Construction workforce ramps from approximately 50,000 in 2028 to 200,000+ by 2035 — absorbing a substantial portion of displaced clerical and administrative workers through the apprenticeship pipeline
• Manufacturing workforce ramps from current levels to approximately 380,000-700,000 by 2045 as the sovereign-content mandate activates the rail mill, OCTG mill, precast megafactory, and downstream manufacturers
• Apprenticeship throughput at peak approximately 30,000/yr provides the structured retraining pathway
• Operations workforce ramps as network commissioning completes, locking in approximately 100,000 permanent positions across the corridor and Alice Hub
• Direct SBC revenue from Phase 1 onwards covers operations and progressively offsets capex (Memo 19 §11, Memo 20 §6)

The labour-market case for the SBC is that it is the only Australian programme at the scale and timeline to actually counterweight the AI displacement trajectory.

6.4 The window closes

The arithmetic above is the case for acting now rather than later. If the SBC programme is not committed in approximately the 2027-2030 period, the fiscal capacity to commit it later will not exist at the scale required. The deteriorating welfare and tax position under AI displacement narrows the fiscal envelope. The displaced workforce, without the SBC apprenticeship pipeline, has not been retrained into the construction and manufacturing categories the programme requires. The political environment, under sustained labour-market stress, becomes increasingly oriented toward short-term relief rather than long-term capability building.

The narrow window is now. The 2027-2030 commitment decisions determine whether Australia builds the programme that meets the AI displacement, or watches the displacement deteriorate the fiscal capacity that would have built the programme.

7. The sovereign asset ownership argument

The SBC is not only a jobs programme. It is also Australia's largest opportunity in a generation to build sovereign assets that pay their way for future generations. This section develops that argument.

7.1 What Australia has done with public assets historically

Australia's pattern over the past four decades has been to build national assets through public investment, then sell them to private owners — often foreign — for short-term budget benefit. The ports of Newcastle, Botany, Melbourne, Darwin, and Port of Brisbane all sold. Telstra privatised in three tranches. Electricity generation and distribution privatised across most states. CSL, Qantas, Commonwealth Bank — all originally public assets, now privately owned. The cumulative effect: substantial annual returns from infrastructure that Australia paid to build now flow to private owners, often offshore.

The SBC is the opportunity to reverse the pattern. The Sovereign Build Corporation is — by name and by structure — a government-owned entity that builds, owns, and operates Australian infrastructure for Australian beneficiaries on an indefinite timeline.

7.2 What sovereign ownership delivers over 50 years

Snowy 1.0 is the Australian precedent. Built 1949-1974 as a public works programme. Still government-owned in 2026, fifty years after completion. Still delivering electricity to the National Electricity Market. Still delivering water diversions to the Murray-Darling food bowl. Cumulative return to the Australian people over the operating period: in the order of $200+ billion in direct revenue, plus the cascading uplift across agriculture, regional development, and engineering capability that Memo 20 §1.4 catalogues.

The SBC programme is approximately 50× the scale of Snowy 1.0. At 50× scale and the same structural ownership, the cumulative return to the Australian people over a 50-year operating window is approximately $10 trillion or more in direct revenue. Plus the cascading uplift across the rest of the economy.

This is not a working estimate. This is what arithmetic does to the Snowy 1.0 return profile when applied to a programme 50× the scale. The figure is at high uncertainty over 50 years for obvious reasons — but the direction is robust.

7.3 Sovereign ownership as labour-market guarantee

Sovereign ownership also matters for the labour-market argument this memo makes. A privately-owned SBC could legitimately optimise its workforce composition for shareholder return — outsourcing, offshoring, replacing workers with automation wherever marginal economics allowed. A sovereign-owned SBC operates under a different mandate: Australian content, Australian workforce, Australian apprenticeship pipeline, Australian supply chain. The 1.25-2.36 million jobs in §3.10 are sustainable at that scale because the entity that creates them is structured to sustain them.

7.4 The contrast with privatisation alternatives

Some readers will ask why the SBC must be government-owned rather than delivered through a private consortium or public-private partnership. The answer is in §7.1, §7.2, and §7.3 above. Australia has the historical evidence to know what happens when nation-building infrastructure is delivered by private owners: the assets get built, the workforce gets employed during construction, and then the returns flow to private (often foreign) owners for the operating life of the asset. The SBC is structured to break that pattern. It is named the Sovereign Build Corporation deliberately.

8. The New Deal analogue, honestly

The Grok research memo that informed this analysis invoked Roosevelt's New Deal as historical precedent. The analogue is useful but worth handling carefully. This section sets out what the analogue is and what it isn't.

8.1 What the New Deal actually did

Between 1933 and 1942, the Roosevelt administration in the United States ran a series of large public-works programmes — the Works Progress Administration, the Public Works Administration, the Tennessee Valley Authority, the Civilian Conservation Corps. Cumulative employment across these programmes peaked at approximately 8.5 million workers (~15% of the US workforce). Lasting assets delivered included the Hoover Dam, the TVA hydroelectric system, large portions of the federal highway system, and substantial regional development infrastructure across the south-eastern United States.

The programmes were the response to the Great Depression — a structural economic collapse that left unemployment at approximately 25% and the conventional employment market broken. They worked. Unemployment recovered to single digits by the end of the 1930s. The infrastructure assets remained productive for the following half-century.

8.2 Why the analogue applies to the SBC

The AI displacement scenario is not identical to the 1930s depression — the causes are different, the timeline is different, the economic context is different. What is the same is the structural mismatch between conventional labour market demand and the available workforce, and the role of large-scale public works in bridging that mismatch with lasting national assets. This memo argues that the AI displacement creates an analogous structural mismatch that an analogous large-scale public works response can address.

8.3 Snowy 1.0 — the Australian precedent

The Australian closer analogue is Snowy 1.0 (1949-1974). A government-owned public works programme employing approximately 100,000 workers cumulatively over its 25-year build period, of whom roughly 70% were migrants from 30+ countries — post-war displaced Europeans, mostly. Italians, Germans, Poles, Ukrainians, Yugoslavs, Greeks, Hungarians, Czechs, Dutch. They came to Australia displaced by a war, were integrated into the workforce through structured trade training on the scheme, married Australian women and other migrant women, raised families in the seven new regional towns the scheme created (Khancoban, Cabramurra, Talbingo, Adaminaby, Cooma, Berridale, Jindabyne), became Australian. It was the largest single integration of new Australians into the workforce in the country's history. Lasting national assets: 4 GW of dispatchable hydroelectric capacity, the Murray-Darling water diversion delivering the food bowl that produces $30 B/yr in agriculture today, the engineering profession trained at continental scale, the seven regional towns, and the cultural transformation of Australia from a 7-million-person Anglo-Celtic country in 1947 to a 13-million-person multicultural country by 1971. The scheme didn't just build infrastructure. It built modern Australia.

The SBC is the modern Snowy 1.0 at continental scale. Same public-works structure. Same sovereign ownership. Same emphasis on lasting national assets. Same regional employment focus. Same integration of skilled trade training. Same role for skilled migration in supplementing the peak-build workforce. Same population-expansion mechanism. Different technology, different geographic scope, different driving need — but the institutional architecture and the demographic logic are recognisably the same.

8.3.1 Population expansion as the deliberate Snowy 1.0 mechanism

Australia in 1947 had a population of approximately 7.6 million. By 1971 the population was approximately 13 million — a 70% increase in 24 years, driven by deliberate skilled migration policy plus natural increase. The infrastructure programmes of the period (Snowy 1.0 first, plus the highway system, the manufacturing base, the suburban expansion) absorbed the migration by providing the employment, the housing, and the regional settlement structure that made the population growth physically and economically viable.

Australia in 2026 has a population of approximately 27 million. The SBC programme is sized for a country of 38-42 million people by 2050 — a similar percentage increase to the post-war period, on a similar timeline, driven by a similar combination of skilled migration and natural increase. The 200+ corridor towns and 11 intersection cities are the regional settlement structure. The 1.25-2.36 million SBC-related jobs are the employment. The 500,000-1,000,000 new dwellings (Memo 20 §4.5) are the housing. The arithmetic is internally consistent: a population of 38-42 million is what the SBC programme is sized to employ, house, and feed.

The honest framing for the international audience is that Australia opens to the world's displaced skilled workers — AI-displaced workers from peer economies, climate-displaced workers from Pacific and Asian geographies, the high-skilled global construction and engineering workforce currently underemployed — and integrates them into the SBC programme the way Australia integrated post-war Europeans into Snowy 1.0. The country expands deliberately, the way Australia expanded deliberately in the 1950s-1970s. The infrastructure programme and the population expansion reinforce each other rather than competing for resources.

8.4 What this memo does not claim about the analogue

This memo does not claim that the SBC replicates the New Deal or Snowy 1.0 line for line. The economic, technological, and geopolitical contexts differ in material ways. What this memo claims is that the structural problem (technological displacement creating labour-market mismatch on a 10-15 year horizon, with a parallel opportunity for deliberate population expansion) has been faced before, and the structural response (large-scale sovereign-owned public works delivering lasting national assets, supported by skilled migration at scale) has been demonstrated to work at the scale and on the timeline the structural problem requires. The SBC is the Australian iteration of that structural response.

9. The counterfactual — Australia in 2045 without SBC, with AI displacement

Memo 21 develops the without-SBC counterfactual on the infrastructure and economic dimensions. This memo extends it to the labour-market dimension.

9.1 The 2045 without-SBC labour market

Without the SBC programme, Australia in 2045 faces:

• AI displacement that has run its course through clerical, administrative, customer service, basic professional services, and entry-level cognitive roles. Approximately 1.5-2.5 million Australian workers displaced from current roles over the period.
• A reduced workforce in coastal capital cities as displaced workers move toward whatever employment opportunities exist
• Construction and manufacturing workforces that have not grown — the without-SBC infrastructure programmes Memo 21 catalogues are smaller and more fragmented
• An ageing population requiring more healthcare workers — many of whom would have come from the now-displaced administrative workforce had retraining infrastructure existed
• Welfare spending materially elevated — approximately $40-80 B/yr above current baseline by 2045
• Tax revenue materially reduced — approximately $25-50 B/yr below current baseline by 2045
• Federal debt accumulated to approximately 65-85% of GDP under the deteriorating fiscal trajectory

9.2 The 2045 with-SBC labour market

With the SBC programme in place, the same period plays out very differently:

• The same AI displacement pressure on clerical and administrative roles — but with a structured apprenticeship pipeline of approximately 20,000-36,000 per year for retraining into construction, manufacturing, engineering, and operations, plus a much wider set of short-retraining pathways into the corridor town and agricultural economies (§4.5)
• Peak SBC corridor construction workforce of approximately 150,000-300,000 providing direct AI-resistant employment through the build period
• Manufacturing workforce of approximately 380,000-700,000 at maturity (including sovereign defence, export, and space industries) providing direct AI-resistant employment through the operating period
• Operations workforce of approximately 60,000-130,000 permanent across the corridor and Alice Hub
• Local service workforce of approximately 340,000-570,000 in corridor towns and intersection cities (schoolteachers, nurses, trades, retail, hospitality, local government)
• Direct agricultural workforce of approximately 160,000-320,000 across Murray-Darling drought-proofing, new inland production, and the agrivoltaic productive country
• Food processing and agricultural supply chain workforce of approximately 80,000-160,000
• Town construction and amenity-build workforce of approximately 80,000-180,000 at peak (separate from SBC corridor capex; built by state and local government, private developers, individual builders)
• Population expansion from approximately 27 million today to approximately 38-42 million by 2050, driven by deliberate skilled migration into the SBC programme and natural increase — the Snowy 1.0 mechanism at continental scale
• Regional employment distributed across 200+ corridor towns and 11 intersection cities — not concentrated in coastal capitals
• Welfare spending at or near current baseline because the labour market has been actively supported through the transition
• Tax revenue at or above current baseline because the manufacturing revival ($93-162 B/yr industry revenue), agricultural production uplift ($39-77 B/yr), and construction sector activity have broadened the tax base
• Federal debt trajectory stable because the SBC programme is structurally self-funding from Phase 1 onwards (Memo 20 §6)

The labour-market case for the SBC is approximately $80-150 B/yr of avoided welfare costs and recovered tax revenue at 2045, on top of every other case the trilogy of memos has made.

10. Honest qualifications

Same methodology stance as Memos 19, 20, and 21.

10.1 The displacement projections are uncertain

The Goldman Sachs, WEF, McKinsey, and JSA projections summarised in §2 are working estimates by responsible analysts. They could be too high or too low by a factor of two in either direction. The case for the SBC programme as an AI displacement response does not depend on any specific projection being correct. It depends on the direction being correct — that material displacement is coming on a 10-15 year horizon — which is the consensus view.

10.2 The workforce numbers are working estimates

The 1.25-2.36 million total SBC-related employment at maturity figure in §3.10 is a working estimate derived from the locked capex schedule (Memo 19) and the locked manufacturing revival and agricultural production revenue figures (Memo 20 §3.1, §3.2, §3.6, §3.7). The local service workforce, food processing supply chain, and town construction figures are at under-SBC-scoping confidence — they require detailed population and settlement modelling to tighten. Detailed workforce modelling by the Sovereign Build Corporation will tighten all the ranges. The figures are honest pre-feasibility working numbers, not detailed industrial-relations forecasts.

10.3 The retraining transition is the hard part

The SBC programme creates employment in AI-resistant categories. It does not automatically retrain displaced clerical workers into those categories. The apprenticeship and training pipeline in §3.4 is the mechanism, but it requires individual decisions by displaced workers to enter retraining, structured income support during the retraining period, and credible career pathways at the other end. Designing that retraining transition well is the policy challenge this memo identifies but does not solve. A subsequent memo on workforce transition policy is forthcoming.

10.4 Some SBC employment will eventually face automation

Robotics and AI will continue improving. Some construction tasks will eventually automate. Some manufacturing operations will eventually move to higher levels of robotics. The SBC programme is not a permanent insulation from technological change — no employment programme can be. What it is is the largest available Australian counterweight on the 20-30 year horizon that matters for the workforce currently in mid-career. The longer-term workforce question is for a different generation of policy.

10.5 The fiscal arithmetic depends on programme execution

The deteriorating-without-SBC and stable-with-SBC fiscal trajectories in §6 depend on the SBC programme being executed on something approximating the Memo 19 capex schedule and the Memo 20 return profile. Execution risk is real. The SBC programme could be approved and then under-delivered, miscost, or politically captured. The methodology stance is to flag this risk explicitly rather than assume frictionless execution.

11. Bottom line

Artificial intelligence and robotics will displace a material share of Australian work over the next 10-15 years. The honest range of external projections sits between approximately 1 million and 2.5 million Australian workers requiring transition by 2045. The displacement is geographically concentrated in coastal capital cities and occupationally concentrated in clerical, administrative, and routine cognitive roles.

The SBC is the largest available Australian response at the scale and on the timeline the displacement requires. The programme employs and enables approximately 1.25 to 2.36 million Australians at maturity — directly through SBC corridor construction, sovereign manufacturing, operations and maintenance, and the apprenticeship pipeline; and indirectly through the agricultural workforce, food processing supply chain, local services in corridor towns and intersection cities, and the town construction itself. The figure comfortably exceeds the McKinsey Australia displacement projection of 1.3 million workers needing transition by 2030. The SBC is at the scale the displacement requires, with deliberate room for the population to expand.

The programme is geographically distributed across 200+ corridor towns and 11 intersection cities — not concentrated in coastal capitals. The corridor town economy is made viable by cheap fast maglev passenger service (Newcastle-Sydney 15 minutes, Sydney-Melbourne 90 minutes, Melbourne-Brisbane 3h 50min direct) that breaks the historical pattern of regional Australian isolation. A weekend at the beach from an inland corridor town becomes the same kind of trip as a weekend in the Blue Mountains from Sydney is today — and the corridor towns become the natural home of the AI-displaced workforce that retools into the regional economy.

The same programme that creates the AI-resistant employment also builds the physical infrastructure that the AI economy requires: sub-10c/kWh renewable electricity, cooling water at scale, sovereign-aligned regulatory framework, low-latency fibre to the Asian community of nations. Australia simultaneously builds what AI cannot, and powers what AI needs.

The programme builds new sovereign assets that pay their way for future generations of Australians — reversing the decades-long pattern of building public infrastructure and then selling it to private (often foreign) owners. The Sovereign Build Corporation is named the way it is named deliberately.

The programme is supported by deliberate skilled migration on the Snowy 1.0 model. Snowy 1.0 employed approximately 100,000 workers cumulatively over 25 years, of whom roughly 70% were post-war displaced European migrants, who became Australian. The SBC opens to the world's displaced skilled workers — AI-displaced workers from peer economies, climate-displaced workers from Pacific and Asian geographies, the high-skilled global construction and engineering workforce — and integrates them into the programme the way Snowy 1.0 integrated post-war Europeans. The country expands deliberately, from approximately 27 million today toward approximately 38-42 million by 2050. The infrastructure programme and the population expansion reinforce each other rather than competing for resources.

The fiscal window for committing to the programme is now, while unemployment remains low, deficits remain manageable, and the tax base remains broad. As AI displacement accelerates, the fiscal capacity to commit to a programme of this scale erodes simultaneously with the labour market that would have made it necessary. The 2027-2030 commitment decisions determine whether Australia builds the programme that meets the AI displacement, or watches the displacement deteriorate the fiscal capacity that would have built the programme.

The historical precedents are honest — Roosevelt's New Deal (Hoover Dam, TVA, federal highway system) and Australia's Snowy 1.0 (4 GW dispatchable hydro, the Murray-Darling food bowl, seven regional towns, the engineering profession at continental scale, and the cultural transformation of Australia into a modern multicultural country). The SBC is the Australian iteration of that institutional architecture at continental scale.

The labour-market case for the SBC sits on top of every other case the trilogy of Memos 19, 20, and 21 has made. The country that builds what AI cannot, powers what AI needs, owns what gets built, and opens to the world's best displaced workers will set the terms on which it enters the AI era. The country that does not, will not.

Build what AI cannot. Power what AI needs. Own what gets built.