The PNW AI Desert

If you're an operator in the Vancouver-Camas-Portland-Tigard corridor and you've thought about hiring an AI engineer, the data tells you something the local conversation hasn't yet caught up to: the AI engineer market does not exist here at the size you'd need to run a normal hiring search.

The data

aidevboard.com indexes AI/ML engineering roles across 489 companies. As of April 2026, the index lists 8,400+ open positions and surfaces 25 cities as named hiring hubs — the cities with enough role density to justify a dedicated landing page. The full list:

San Francisco · New York · London · Mountain View · Sunnyvale · Bangalore · Palo Alto · Boston · Paris · Toronto · San Jose · Austin · Seattle · Singapore · Amsterdam · Costa Mesa · San Mateo · Foster City · Munich · Redwood City · Irvine · Stockholm · Menlo Park · Berlin · Tokyo

One Pacific Northwest city. Seattle. No Portland. No Vancouver WA. No Camas. No Tigard. No Beaverton. No Hillsboro. No Lake Oswego.

What this means for SMB operators

If you run a small business in Vancouver WA, Camas, Portland OR, or Tigard, and you've been quietly assuming you can solve "we should be using AI" by hiring someone — the labor market you're imagining doesn't exist here at the size you'd need.

The numbers underneath this:

• The average AI/ML engineering salary across the index is in the $200k–$280k range. SMBs in operating-business categories (logistics, manufacturing, healthcare practices, professional services) cannot competitively bid against San Francisco and Seattle for the small pool of qualified candidates who would relocate.
• The candidates who already live in the Vancouver-Camas-Portland-Tigard corridor and have AI engineering skills are, in most cases, already employed remotely by SF or Seattle companies who pay above-market for the privilege of keeping them out of the local recruiting pool.
• Even when you do find a candidate, the cost of one senior AI hire is roughly the same as a full embedded engagement that ships three to five working AI workflows alongside your existing team and leaves the runbooks behind.

The hiring path is structurally closed. The build-it-yourself path requires an AI engineer you can't hire. Which leaves a third path — one that the operating businesses doing this well in the PNW are already running.

The third path

Install AI workflows on top of the team you already have. Not by hiring. Not by sending your operations manager to a four-week course. By bringing in a small embedded team for four to twelve weeks — the people who have shipped AI deployments before and have the systems-design judgment that doesn't transfer through documentation — and having them build the workflow alongside the operators who actually run the business.

This works in the PNW for a specific reason that doesn't apply in Bay Area: your operators have institutional knowledge that no AI hire could acquire, and the hiring market for AI engineers locally is so thin that the alternative isn't competitive even on cost. The math that recommends "hire" in San Francisco recommends "embed" in Vancouver WA, and the difference is not subtle.

The pattern across the four cities

The shape of this opportunity differs slightly by city:

City	Highest-density SMB sectors	Local AI hire viability
Vancouver, WA	Logistics & 3PL, manufacturing, healthcare practices, professional services	Effectively zero local candidate pool; embed is the only viable path
Camas, WA	Specialty manufacturing, family-owned trades, niche professional services	Effectively zero local; embed is the only viable path
Portland, OR	F&B production, creative agencies, healthcare practices, specialty retail	Thin local pool; mostly remote-employed by Bay Area companies
Tigard, OR	Wholesale distribution, professional services, specialty retail, healthcare	Effectively zero local; embed is the only viable path

The implication is the same in all four: operating businesses that wait to "hire someone good" before moving on AI will be waiting indefinitely. The market doesn't reward patience here. It rewards finding the alternate path and running it.