One of my salespeople came to me last week with a question. A customer building a brand-new facility asked him how they should approach OT security, and he wanted to know what to say.
In existing networks, the challenge is always protecting legacy devices, which is one of the big advantages of network cloaking. My answer surprised him: “Why don’t you start by saying that building with brand-new technology can be even more challenging in today’s OT security environment?”
He expected me to say the new plant was the easy one. Most people would. But that isn’t the case IMHO.
Old plants got connected by accident. New ones are connected by design.
A controller installed twenty years ago was never meant to talk to anything outside its cell. When one ends up on the internet, it’s because somebody added a path later: a vendor’s remote access, a modem, a shortcut that outlived the project that needed it. CISA saw exactly that this year. In July 2026, it observed malicious activity targeting internet-exposed systems at more than 100 water and wastewater entities, “commonly via programmable logic controllers (PLCs) connected directly to a cellular modem.”
That’s a serious problem, and protecting those legacy devices is still the hardest work in most plants. But it’s an accident. The connection wasn’t the system's point, so it can be removed without breaking the reason the plant exists.
A greenfield site doesn’t get that option. The connectivity is in the spec. Remote monitoring, vendor support, cloud analytics, building management, EV chargers that report home. The owner, the integrator, and the equipment maker all expect to reach the system from somewhere else, and the business case was written on that assumption. Nobody is going to unplug it.
Greenfield isn’t a clean slate. It’s a shared one.
The second reason is what the new systems are made of. Old controllers ran proprietary firmware that almost nobody outside the vendor ever read. New ones run on the same open-source libraries and Linux kernels as everything else.
That’s efficient. It’s also why one flaw can travel a long way.
On November 12, CISA republished Siemens advisories for Node2Shell, a remote code execution flaw (CVE-2026-111497, CVSS 9.8) in the open-source Brotli decoder. It reaches Desigo CC building management, Gridscale X, SICHARGE and VersiCharge EV charging, Industrial Edge, and RUGGEDCOM APE1808. CISA added it to its Known Exploited Vulnerabilities catalog five days after the CVE was published. For some Industrial Edge components, the advisory reads “Currently no fix is available.” Nobody has reported an attack aimed at those Siemens products, and this isn’t a Siemens story. It’s what modern equipment is built from.
And people aren’t the only ones reading that code anymore. The Hacker News reported that the agents OpenAI used to red-team its own training infrastructure were the ones that found the Brotli flaw. The shared codebase is now being read by machines, and anything built on it inherits whatever they find.
We aren’t exempt. In October, BlastWave published its first security advisory, BW-SA-2026-001, for a use-after-free affecting our Orchestrator (CVE-2026-98432). In the advisory’s words, “The flaw is in the upstream Linux kernel, not in BlastWave code.” Same kernel, same exposure. We fixed it and published it, which is what you should expect from every vendor whose equipment goes into your new plant.
Isolated, yet still connected and available for remote access
Put those two together, and you get the actual design requirement for a new facility. It has to be isolated, yet still connected and available for remote access.
That sentence is why “just air-gap it” stopped being an answer. An air gap fails the moment remote access is written into the contract, and for greenfield, it’s written in before the concrete is poured. The old answer to connectivity was a perimeter, and any perimeter that has to be reachable can be found.
JP Morgan’s industrial research put it plainly: “Internet-facing control system components, insecure remote access pathways, default credentials and poorly protected boundary devices continue to create direct routes into industrial environments.”
A new plant can start with none of those. Not because its code is better than the old plant’s. It isn’t, and it’ll keep getting flaws found in it for as long as it runs. It can start clean because the architecture decision hasn’t been made yet. Systems that don’t answer anyone they don’t know. People who need access get it without anyone opening a port. A shared codebase that nobody outside can reach to test.
That’s a decision you can only make once cheaply, and it’s on day one.
What I told him to say
So here’s what I told my salesperson to say to the customer. Don’t treat the new plant as the easy one. Assume it’ll be connected, because it has to be. Assume its code will have flaws, because it’s the same code as everyone else’s. Then design it so neither assumption matters to someone scanning from the outside.
The old plant is hard to protect because nobody planned for the internet. The new one is hard because everybody did.
– Cam Cullen, CMO, BlastWave
American Water’s cyber incident shows why water utilities need stronger OT isolation to stop IT breaches from threatening critical infrastructure operations and availability.
Explore the complete analysis of 23 OT attacks that defeated firewalls, VPNs, and air gaps.