Market and industryNovember 19, 2026Редакция 105.kz
Nadella: Critical infrastructure has many unpatchable devices
Microsoft CEO Satya Nadella said a significant part of critical infrastructure — firmware, chillers, HVAC systems — is connected to the internet and cannot be patched. Against the backdrop of AI attacks, this requires revising security strategies and modernization.
Key takeaway. Satya Nadella said critical infrastructure has many unpatchable devices — firmware, chillers, HVAC — connected to the internet. AI attacks shorten the time from vulnerability discovery to exploitation, requiring immediate patches, which creates instability. For Kazakhstan, this is a signal to conduct OT/SCADA inventory, network segmentation, and modernization plans for legacy systems (assessment).
What happened
On November 18, 2026, at The New York Times DealBook summit, Microsoft CEO Satya Nadella said a significant part of critical infrastructure contains software vulnerabilities that cannot be fixed with patches. According to him, this includes firmware, chillers, HVAC systems, and similar devices connected to the internet. "Remember, our critical infrastructure has a lot of unpatchable: firmware, chillers, and HVAC systems everywhere. And all of it, by the way, is on the internet," Nadella said.
He noted that AI accelerates both the discovery and exploitation of vulnerabilities, so patches now need to be applied almost instantly, without the previous time for testing, which creates instability. The talk came amid recent large-scale outages, including an Outlook incident that Nadella attributed to a "latent error" and a "perfect storm." Officials denied that these outages were cyberattacks, but concerns about AI attacks are growing.
What this means for Kazakhstan enterprises
For Kazakhstani oil and gas, energy, and manufacturing facilities, the principle applies directly: SCADA servers, controllers, field devices, dispatch systems, chillers, and HVAC often run for years without firmware updates (common engineering practice). If such devices have access to the corporate network or the internet, they become a potential attack entry point (assessment).
In oil and gas, typical nodes are process control systems at oil pumping stations, compressor stations, tank farms, where SCADA and programmable logic controllers manage processes. In energy, these are turbines, boiler units, substations, relay protection systems, as well as chillers and HVAC for equipment cooling. In factories, there are production line controllers, MES systems, and industrial refrigeration units. For all of them, the same applies: if a firmware vulnerability cannot be closed with a patch, the options are to isolate the device from the network, replace it, or compensate for the risk through monitoring.
Given data localization in Kazakhstan and national companies' information security requirements, hosting monitoring and management systems inside the enterprise perimeter (on-premise) remains the norm. This limits external access but does not eliminate the risk of internal anomalies and misconfiguration.
How to apply: 8–12 week pilot
- OT asset inventory. Compile a register of all devices connected to the process network: SCADA, controllers, engineering stations, chillers, HVAC, sensors. For each, record firmware version, internet accessibility, and updatability. You can start with oil and gas or energy facilities.
- Assess network segmentation. Check whether process segments are separated from the corporate network and the internet. Where segmentation is absent, identify critical nodes for priority isolation.
- Anomaly monitoring. Deploy telemetry collection and network traffic analysis in the pilot zone to identify suspicious activity and attempts to exploit vulnerabilities.
- Pilot on a single unit. Select one process unit or substation, conduct an audit, and apply isolation and monitoring measures. Measure anomaly detection time and the number of events before and after (assessment).
- Prioritize modernization. Based on the register, create a plan to replace or isolate unpatchable devices considering criticality and residual risk.
These steps are managerial, without engineering implementation details; deployment will require integration with existing process control systems and compliance with information security requirements.
Limitations and risks
Nadella's statement is the opinion of a Microsoft executive, not measured data for Kazakhstan. The source does not provide statistics on the share of unpatchable devices in critical infrastructure, and there is no information about specific facilities in the Republic of Kazakhstan.
The recent outages mentioned in the article were not confirmed by officials as cyberattacks, so a causal relationship between AI attacks and these incidents has not been established.
Patches for process control systems and embedded software require mandatory bench testing, as updates can disrupt the technological process. Immediate patch application, as Nadella describes, may be unacceptable for industrial systems.
The effectiveness of AI in discovering vulnerabilities does not by itself mean automatic improvement in security: response and prioritization processes must be established.
There are no open-source data on how many devices at Kazakhstani enterprises have internet access, so all conclusions for the Republic of Kazakhstan should be considered a hypothesis requiring pilot verification.
Sources
- Amidst fears of AI cyberattacks, Satya Nadella says lots of critical infrastructure is ‘unpatchable’ · 2026-11-18T23:45:00.000Z
Remember, there is a lot of stuff in our critical infrastructure that’s unpatchable... think about all the firmware and the chillers and the HVAC systems everywhere else. So all of that, by the way, they’re on the internet.
- Фото: Linda Spashett Storye_book — Bradford Industrial Museum 030.jpg (CC BY 3.0, Wikimedia Commons)
This material was prepared with AI from the sources listed above. Spotted an inaccuracy? Write to info@105.kz.