Every year-end brings a flood of predictions about where technology is heading. Most are either obvious (yes, AI will keep developing) or wrong (no, factories won’t be fully autonomous by next quarter).

Rather than grand pronouncements, let me share what I’m actually seeing in the field and where I expect things to move in the coming year.

Trend 1: Generative AI finds manufacturing-specific applications

2025 saw generative AI (ChatGPT and friends) explode into mainstream awareness. Most manufacturing applications so far have been generic—using ChatGPT for documentation, emails, reports.

In 2026, I expect more manufacturing-specific generative AI applications:

Technical documentation assistants: AI that understands your specific equipment and can answer technical questions, generate procedures, and assist with troubleshooting.

Design assistance: Generative AI helping with CAD, suggesting design improvements, and automating routine engineering tasks.

Planning and scheduling interfaces: Natural language interfaces for complex planning systems, making them accessible to more users.

Training content generation: Creating customised training materials, simulations, and assessments.

These won’t replace specialised manufacturing AI (vision systems, predictive maintenance), but they’ll fill gaps where generic AI can add value to industrial contexts.

Trend 2: Edge AI gets easier

Running AI on edge devices—industrial PCs, gateways, embedded systems—has required significant expertise. In 2026, I expect this to become more accessible:

Pre-packaged solutions: More “AI in a box” products from automation vendors, with models ready for common manufacturing applications.

Easier model deployment: Better tools for moving models from development to edge deployment without deep expertise.

Hardware commoditisation: Edge AI hardware becoming cheaper and more standardised.

This matters because edge deployment is necessary for many manufacturing applications (latency, reliability, data sensitivity), but the barrier has been too high for many organisations.

Trend 3: Integration platforms mature

A major challenge for manufacturing AI has been integration—connecting AI systems to the equipment, data sources, and enterprise systems they need.

I’m seeing integration platforms mature:

Industrial data platforms: Unified data layers that collect and normalise data from diverse sources (Siemens Insights Hub, PTC ThingWorx, AWS IoT SiteWise, and others).

Pre-built connectors: More out-of-the-box integration with common equipment and systems, reducing custom development.

API standardisation: Gradual (though still slow) progress toward common interfaces for industrial data.

This won’t make integration easy—it’s still a major undertaking—but it should become less of a barrier.

Trend 4: The skilled-worker crunch drives adoption

Australia’s manufacturing skills shortage is well-documented. Finding skilled trades, technicians, and engineers is increasingly difficult.

This pressure is pushing AI adoption as a coping mechanism:

Knowledge capture: AI systems that capture expert knowledge before experienced workers retire.

Skill augmentation: AI that helps less experienced workers perform at higher levels.

Automation of shortage areas: Where workers can’t be found, automation becomes the only option.

I expect skills pressure to be a stronger AI adoption driver in 2026 than pure efficiency gains.

Trend 5: Focus shifts from pilots to production

Manufacturing AI has been in “pilot phase” for years. Many organisations have completed successful pilots but haven’t scaled to production.

In 2026, I expect more focus on:

Scaling proven applications: Moving from “it works on one machine” to “it works across the operation.”

Operationalisation: Building the processes, roles, and infrastructure to run AI in production rather than as experiments.

Value realisation: Accountability for actually delivering the benefits that justified investment.

This shift is healthy. Pilots are necessary, but value comes from production deployment.

Trend 6: Cybersecurity becomes a bigger AI consideration

Industrial cybersecurity has been a concern for years. AI adds new dimensions:

AI systems as attack targets: AI systems in operations create new attack surfaces.

AI for security: Using AI to detect industrial network anomalies and potential attacks.

Secure AI development: Ensuring AI supply chains (training data, models, platforms) are secure.

I expect security considerations to become more prominent in manufacturing AI discussions in 2026, driven by both regulation (SOCI Act) and high-profile incidents in other industries.

Trend 7: Sustainability applications grow

Environmental pressure on manufacturing is increasing from regulators, customers, and society. AI can help:

Energy optimisation: AI reducing energy consumption, which I’ve discussed before.

Emissions tracking and reduction: AI monitoring and optimising processes to reduce carbon footprint.

Waste reduction: AI improving yield and reducing scrap.

Circular economy: AI enabling recycling, remanufacturing, and product lifecycle management.

These applications often have strong business cases (energy savings, material savings) as well as sustainability benefits. I expect them to grow.

What I don’t expect

A few things I don’t see happening in 2026:

Autonomous factories: Human oversight will remain essential. Full autonomy is still years (probably decades) away.

Massive job displacement: AI will change jobs more than eliminate them in the near term. The skills crunch limits how fast change can happen anyway.

Universal standards: The industrial standards landscape will remain fragmented. Don’t expect a single standard that solves interoperability.

AI solving everything: There will still be plenty of problems where traditional approaches work better than AI.

What this means for Australian manufacturers

If you’re planning for 2026:

Evaluate generative AI: Explore manufacturing-specific applications beyond generic chat assistance.

Consider edge: If you’ve been waiting for edge AI to mature, revisit the options—they’re improving.

Plan to scale: If you’ve got working pilots, what would it take to expand them? That planning is worth doing now.

Address security: Make sure AI systems are included in your cybersecurity thinking.

Link AI to sustainability: If you have sustainability commitments, consider how AI might help achieve them.

Build skills: The people-side investment is as important as the technology. Skills developed in 2026 enable progress in 2027 and beyond.

A measured outlook

I’m cautiously optimistic about manufacturing AI progress in 2026. The technology continues to mature. Practical applications are proven. The challenges (skills, integration, change management) are real but not insurmountable.

Progress will be incremental rather than revolutionary. That’s fine. Incremental progress, compounded over years, adds up to significant transformation.

Australian manufacturing won’t be unrecognisable in a year. But manufacturers who are thoughtfully adopting AI will continue to pull ahead of those who aren’t. That gap will widen.

The question for each organisation is where you want to be when you look back at 2026.