Trump Says AI Will Create ‘Millions and Millions’ of Jobs. Here Is What Canadian Business Owners Should Actually Take From That Claim.
President Trump recently told reporters that any governor or mayor should welcome AI data centres because they will create “millions and millions” of jobs. That quote is accurate — he said it. Whether it holds up as a net jobs forecast for AI broadly is a different question, and the evidence does not settle it. For Canadian business owners watching the AI conversation from a distance, the more useful question is this: construction jobs around a facility are real and time-limited; operating a trained model is not the same as running a business. Your firm still needs people doing revenue-generating work, and that is exactly where an agent system is designed to help — not by replacing your staff, but by clearing the operational noise that buries them.
What This Means for Canadian Business Owners
The data centre boom Trump is describing is real. Billions of dollars in capital expenditure are flowing into large-scale AI infrastructure across North America. Construction workers, electricians, HVAC technicians, and civil engineers are being hired in volume to build those facilities. That is a legitimate economic activity and those jobs are tangible.
What those jobs are not is a forecast for what happens to your accounting firm, your law practice, or your trades business once the buildings are finished and the servers are running. A data centre employs somewhere between 100 and 200 people in its immediate area once operational, according to Brookings Institution research on data centre employment effects — a fraction of the construction workforce that built it. The facility runs largely with a small technical team. The downstream economic activity in surrounding sectors is real but modest and localized.
This is not an argument against AI investment. It is a reminder that infrastructure-level job creation and firm-level workforce strategy are two separate conversations. Canadian business owners do not need to form a view on US energy policy or federal data centre subsidies. They need to understand what agent systems do inside a working firm — and what they do not do.
The Real Problem with the AI Jobs Narrative
The public AI jobs conversation has two failure modes that feed on each other. The first is uncritical optimism: AI will create millions of jobs, transformation is everywhere, the economy will expand. The second is uncritical fear: AI will eliminate millions of jobs, displacement is everywhere, your staff is at risk. Both of these positions are driven more by political incentive than by operational evidence.
For a Canadian business owner — a CPA running a six-person firm in Hamilton, a paralegal practice in Ottawa, an electrical contractor in the GTA — neither of those narratives maps onto the daily operational reality. The real problem in most of these businesses is not a shortage of AI infrastructure. It is a mismatch between where your most experienced people spend their time and where your revenue actually comes from.
We use a specific vocabulary at TAS to describe this. Cost Centers are operational tasks that consume time but do not directly generate revenue: intake processing, scheduling, document routing, follow-up emails, compliance tracking. Income-Generating Activities are the billable, client-facing, relationship-driven work that your senior staff were hired to do. Strategic Reallocation is the process of using agent systems to shift time away from Cost Centers so your people can do more Income-Generating Activities. Human Middleware is what your senior staff become when they spend most of their day routing information between systems that should not require a human to route it at all.
None of this is about replacing people. It is about removing the administrative weight that buries them.
You can read more about the full range of TAS engagement models and custom system builds to understand what this looks like in practice across different industries.
What Strategic Reallocation Looks Like in Practice
The following is a representative scenario, not a documented client case study. Details are illustrative and intended to show how this framework applies in a typical professional services context.
Consider a mid-sized accounting firm in Ontario. The firm has three senior accountants and two junior staff. The senior accountants each spend several hours per week handling intake questions, routing documents to the right files, chasing missing information from clients, and managing scheduling conflicts. None of this work requires their credentials. All of it takes time away from billable client work and from mentoring the junior staff who need their guidance to develop.
In a representative Digital Landlord engagement, TAS would build an agent system designed to handle that intake and document routing layer. The system would collect information from clients through a structured interface, route documents to the correct file automatically, flag incomplete submissions, and queue follow-up outreach without requiring a senior staff member to initiate it. The senior accountants would receive a clean, organized queue of work that actually requires their expertise — not a pile of administrative tasks that arrived in their inbox over the weekend.
The goal of a system like this is to return several hours per week of senior staff time to Income-Generating Activities. In many firms, the downstream effect is not just efficiency — it is retention. Senior staff who are not buried in administrative noise are less likely to burn out. Junior staff who have access to senior mentors are less likely to panic about their own trajectory and leave. The organizational cascade that begins with one senior departure — and often ends with two or three junior departures within the following months — is the risk this kind of system is designed to interrupt.
This is a completely different conversation from Trump’s data centres. Those jobs are in construction and infrastructure. These jobs are in your firm, and they are worth protecting.
How to Know If Your Business Is Ready
A few honest indicators that your business is ready for a Systems Assessment:
- Your senior staff are regularly finishing administrative tasks outside business hours. If intake, document routing, or scheduling is bleeding into evenings and weekends, that is a Cost Center absorbing time that should belong to billable work — or to the rest of the week.
- Your intake process depends on a person to initiate every step. If a client inquiry sits idle until someone checks their email, you have a Human Middleware problem.
- You have experienced staff turnover in the last 18 months and attributed it to workload. Workload is often a Cost Center problem in disguise. The volume of revenue-generating work is rarely the issue — the administrative overhead on top of it is.
- You have looked at off-the-shelf tools like Zapier or Make and found them too rigid or too complex to maintain. Those tools work for narrow, stable workflows. Custom agent systems are built for the way your business actually operates.
- You handle client data and have questions about where it lives. Canadian businesses operating under PIPEDA have compliance obligations that US-hosted platforms frequently cannot meet. If your current tools route client data through US processors, that is worth a direct conversation before you expand those integrations.
TAS builds all agent systems with Canadian data residency by design. No client data routed through US cloud jurisdiction. If a third-party processor with US infrastructure is the right tool for a specific function, we name it explicitly and document it in the architecture — we do not obscure it. You can review the TAS approach to PIPEDA-compliant AI system architecture if data residency is a concern for your practice.
If you are a verified Chamber of Commerce member, mention it on your intake form. A discount applies to any engagement tier.
Frequently Asked Questions
Will AI actually create jobs, or is it going to eliminate them?
The honest answer is: both, in different categories. Infrastructure-level AI investment — data centres, model training facilities, specialized hardware — does create construction and technical jobs during build phases. Operating staff numbers for those facilities are small once they are live. At the firm level, well-built agent systems are designed to remove administrative tasks from senior staff, not to eliminate staff headcount. The businesses TAS works with are not reducing their teams. They are trying to keep their existing teams from burning out.
Is Trump’s data centre jobs claim accurate?
He said it — that part is accurate. Whether it will materialize as net job creation from AI broadly is disputed. Brookings Institution research on data centre employment effects shows that a large data centre typically generates around 100 to 200 local jobs in related sectors once operational. Construction employment during the build phase is higher but time-limited. Treating a construction boom as a long-run AI employment forecast conflates two separate things.
How is a custom agent system different from off-the-shelf tools like Zapier?
Off-the-shelf tools connect existing applications through predefined triggers. They work well for narrow, stable workflows that fit their template library. A custom agent system is built around your specific operational processes — the way your intake actually works, the way your documents are actually organized, the compliance requirements that actually apply to your firm. When the workflow changes, a custom system adapts. A template does not. TAS builds custom systems; it does not resell or configure generic templates.
What does PIPEDA compliance mean for my AI system?
PIPEDA, Canada’s federal private sector privacy law, requires that personal information collected in the course of commercial activity be handled with appropriate safeguards. When a client intake form, a document management system, or a scheduling tool routes data through a US-based cloud processor, that data may fall under US legal jurisdiction — including potential government access under US law. A PIPEDA-compliant system architecture keeps client data in Canada. For legal, accounting, dental, and medical practices, this is not optional risk management. It is a professional obligation.
How do I know what an agent system build would actually cost for my firm?
Costs vary based on the scope of your workflows, the number of integrations required, whether you want to own the system outright or have it managed on your behalf, and the complexity of your compliance requirements. TAS does not quote engagement costs without a Systems Assessment first. That conversation maps your actual workflows and identifies where an agent system is likely to deliver the most value — before any commitment is made. Book the assessment and you will have a scoped picture of what the engagement looks like for your specific operation.
If this resonates with how your business operates, book a free 30-minute Systems Assessment. We’ll map your workflows and show you exactly where an agent system could help — no commitment required.