Every warehouse has cameras. Almost none have early warning.
Eight days to put out a fire
On June 17, 2026, a fire started at a Lineage Logistics cold-storage warehouse in Boyle Heights, east of downtown Los Angeles. Crews fought it for roughly eight days before knocking it down. The building was about 500,000 square feet and held an estimated 85 million pounds of frozen food. Air quality at the nearest monitor spiked to an AQI of 621 (past the "hazardous" line), and the smoke plume put an emergency declaration over a 2.5-mile radius and something like a quarter-million households.
No one was hurt, which matters more than anything else here. But the part that stopped me was a line from LAFD Chief Jaime Moore, explaining why his firefighters couldn't go inside:
I don't know that we'll ever get firefighters inside, because the entire roof has been compromised and it is sitting on top of 65-foot towers.
Read that again. Once a cold-storage facility like this is fully involved, the building itself becomes something you fight from the outside and wait out. The cause is still under investigation, so this isn't about blaming anyone's equipment. It's about the window, and how early you have to be inside it for any of your tools to matter.
Why these buildings don't give you a second window
Cold storage is uniquely brutal to fight, and the reasons are worth understanding because they collapse your response time to almost nothing:
- The walls and roof are packed with dense foam insulation. The same property that keeps 85 million pounds of food frozen also traps heat and defeats ventilation, the standard way crews relieve a fire.
- The interior is 65-foot steel racking, hundreds of feet long. That's a collapse hazard that keeps humans out and a chimney that helps fire climb.
- Many sites run ammonia refrigeration, which forces crews into a defensive posture the moment a line is breached.
A conventional warehouse fire is often out in a day. A cold-storage fire routinely burns for a week or more. So the entire game is decided early (in the minutes when a problem is still small, localized, and visible), long before it becomes a structure you can only watch burn.
And here's the uncomfortable part: facilities like this are covered in cameras.
"Having cameras" is not "having early warning"
Almost every large warehouse already has extensive CCTV. What it usually doesn't have is anything watching those feeds in the moment. The cameras are a recording system, a forensic tool you rewind after an incident to understand what happened. They are not an early-warning system that surfaces the first wisp of smoke while it's still a small thing in one aisle.
That distinction is the whole point of DHI. A recording system answers "what happened yesterday." An early-warning system answers "what is starting right now, and who needs to know in the next ten seconds."
Most warehouse safety conversations are about compliance checklists. The real question is whether your site can surface an early signal and route it before a small problem becomes a multi-day emergency.
Visual smoke and flame detection won't see a fire that starts on an empty rooftop, and I'm not going to pretend otherwise. But across the floor, the racks, the charging areas, and the dock (everywhere a camera already looks), the difference between footage and detection is the difference between reviewing an incident and interrupting one. A model running on the feed can flag smoke or flame the instant it's visible and push an alert to a human, instead of waiting for someone to happen to glance at the right monitor.
What this means for a warehouse safety program
If you run safety for a logistics or industrial site, the takeaway from Los Angeles isn't "buy more cameras." You almost certainly have enough. It's three questions:
- 01Of all the feeds you already record, how many is anything actually watching for an early signal: smoke, flame, a forklift in a walkway, a person down?
- 02When a signal does appear, how fast does it reach a human, and through what workflow?
- 03Does that detection run on-site, so it keeps working in the first chaotic minutes even if the network is degraded?
The cameras you've already paid for can do far more than record the thing you'll later wish you'd caught. The cheapest minute in any incident is the first one. Start by pointing real detection at the single feed where a fire would do the most damage, and see what it catches before anyone else would have.
For details on the Los Angeles warehouse fire, see the AP wire report (June 22, 2026) and NPR's coverage (June 23, 2026).