Palisade Engineering

ChatGPT, Claude, and Gemini is doing more structural calcs in 2026 than the industry will admit.

Where it's wrong: lightweight-concrete anchorage, wind on irregular roofs, anything with SDC ≥ D.

Where it's surprisingly correct: routine moment-frame sizing, bolt group capacity, client-email code summaries.

The seal still belongs to the engineer.

What grain silos in Iowa teach the data center industry.

Both industries assumed the design load was fixed for the life of the building. In both cases, actual usage outgrew design within 10 years.

Slabs designed for 8kW racks. Now carrying 40kW. The math was fine; reality changed.

Unpopular opinion: the era of the big-name engineering firm is over.

The next decade belongs to specialists. AI compressed the bottom of the practice.
Hyperscale clients want depth, not letterhead. Insurance carriers price specialists
separately now.

The partners getting defensive about this — that's the signal.

Five opinions that get you side-eyed at engineering conferences. We think all five are right:

1. PDH hours don't make better engineers — forensics exposure does
2. A lot of what gets called "peer review" isn't
3. Clash detection isn't coordination
4. Junior engineers should watch a steel crew install a connection before they detail one
5. Seismic provisions encode judgment, not just physics — the code is a floor, not an answer key

Disagree with any of these? That's what the comments are for.

Quick observation as the world's biggest tournament kicks off today:

The cameras will be on the players. Ours are on the ceiling.

Every broadcast gantry, scoreboard support, and equipment rack in those stadiums sits on nonstructural seismic anchorage — the same code framework (ASCE 7 Ch 13, Ip = 1.5) we apply to mission-critical data centers.

Mission-critical operation, just with a louder audience. Enjoy the matches.

Western Nevada Supply's Sparks campus has been "almost done" for about ten years now.

It's not one project. It started as the original build, then an addition, then a seismic retrofit on the existing PEMB, then a mezzanine, then a TI. Most of those overlap. Each has its own OAC series, and the meeting count across the campus is well past anyone's ability to count.

Paul worked the original build years ago at Gabbart & Woods. When G&W wound down he kept the EOR seat — finished the PEMB retrofit and is now stamping work across several concurrent projects on the same site. The current TI in Area 3 involves opening walls in the South counter addition, checking whether the existing structure can take each opening, and reinforcing where it can't.

The thing that doesn't show up in a portfolio shot is institutional memory. Why a column is offset at that grid. What the original lateral system was sized for. What got added back in the retrofit. Stuff you just know if you were on the original drawings, instead of trying to reconstruct it from a 15-year-old set.

Out at WNS with Western Nevada Supply, Plenium Builders, and Don Mackey Architect — the campus that's been "almost finished" for ten years and counting.

👉 https://pe-se.com/contact?utm_source=linkedin&utm_medium=social&utm_campaign=2026-06-linkedin&utm_content=tagged-wns

The Francis Scott Key Bridge collapsed in roughly 7 seconds.

Designed in the 1970s for ships ~1/8 the size of the one that hit it. Modern ships outgrew the bridges decades ago. Nobody retrofit the bridges.

Infrastructure inherited from a previous era is sized for a previous era.

What gets photographed: $4M of glass and a perfect 4pm sun angle.
What keeps it standing: 6 months of someone's spreadsheets.

The award goes to the elevation. The lawsuit goes to the connection detail.

Tag the engineer who hasn't been thanked this month.

Friday afternoon at the desk.

The 11th draft of a coordination markup. The third version of the same calc because the vendor changed the cut sheet again.

The work is good. The week is over. Have a weekend.

Unpopular opinion:
⠀
AI isn't going to make construction interesting. It's going to make it boring again.
⠀
The firms that win the next decade aren't building the "AI-designed building." They're the ones who get back to looking boring while their P&L doubles.
⠀
Tell me I'm wrong.
⠀