A long row of software-defined server racks with glowing indicator lights extends into the distance under a blue sky, illustrating modern cloud computing and network infrastructure.

I recently read how T-Mobilsing tens of thousands of modem failures. CTIA and Ericsson announced AI-native RAN software that they said lifted spectrum efficiency by nearly 10% in U.S. trials and delivered downlink throughput gains of up to 15% versus legacy methods. Impressive numbers.

I don’t share that because of the AI angle, not entirely anyway. Instead, it’s all about speed. AI is only one factor that’s radically driving up the speeds at which satellite must learn to operate.

Here’s another.

I recently attended the Australian Space Forum: 2026 where Paul Solomon, Chief Technology Advisor at KBR, explained the first steps to Russia’s invasion of Ukraine were a line of code. It began with a cyberattack on the Viasat and KA-SAT networks, causing tens of thousands of modem failures and cutting military and government communications.

Necessary reaction times were slashed, a trend that has been ongoing. According to Solomon, citing a Mandiant report, "Time-to-Exploit Trends," the mean number of days to exploit a cyber vulnerability has gone down from 63 in 2018 to just 5 in 2023. Worse yet, Solomon said that starting in 2026 exploitations now typically begin before a patch even exists.

Speed matters.

There are other drivers: cloud enablement, network resiliency and more, all with one thing in common: software. Networks must learn to react at the speed of software. Hardware just can’t keep up, adapt or mend fast enough.

What does this mean for satellite? Well, when your two biggest customer blocs, defense and commercial comms, are both moving in the same direction, following would be a pretty good idea. The further behind we fall, the harder it will be to ever catch up.

According to Nokia CTO Pallavi Mahajan, “Our fundamental thinking behind AI-RAN was let’s decouple hardware from software.” Going on to say that prior to this, “if we were to talk of spectral efficiency, we would have said, ‘I’m going to develop custom silicon which will provide 20%, 30%, 40% spectral efficiency.’ But we are not saying that we will do that. Now we are saying you know what? Nokia today provides you 20 to 25 percent spectral efficiency,” she said, adding that by the end of 2027, “we’ll provide you 50% spectral efficiency. By the end of 2028, we’ll provide you 100% spectral efficiency. And in none of these conversations are we asking our customers to go about and upgrade their hardware.”

The need to move to standard compute for network hardware is no longer a debate. Our customers are already beginning to demand it because they need to move at the speed of software. Can we react at speed?