Crypto Steps Up Defense as AI Data Centers Become Military Targets
 |
| By Bob Czeschin |
Data centers have been sold as the keystone of the AI revolution.
But the reality is simpler than that: They’ve evolved to monetize the massive economies of scale inherent in the computer hardware business.
With huge server farms delivering services to customers via the internet, it is possible to process data at a tiny fraction of the cost of running thousands of dispersed stand-alone units.
And that includes military applications.
Of course, the Pentagon prefers the robust command and control that comes with running their own artificial intelligence systems.
But in many areas, that’s impractical and cost prohibitive.
As a result, military AI shares the same cloud computing infrastructure as retail banking, healthcare, communication networks, and countless other platforms powering civilian economies.
Defense of the Fleet
Here’s the problem with the military sharing data centers: They’re not like other types infrastructure that can sustain partial damage but keep on running, albeit at reduced capacity.
Data centers either run, or they go dark. There’s no in between.
The first known military attack on a civilian data center took place on Sunday, March 1. An Iranian Shahed-136 drone struck an Amazon Web Services (AWS) data center in United Arab Emirates. Followed by another in Bahrain.
That’s most likely because of mounting Iranian frustration at how critically successful AI has been helping defending U.S. Naval forces.
Specifically in protecting the Nimitz-class aircraft-carriers.
These cost a whopping $13 to $15 billion each. So, operational doctrine almost obsessively revolves around protecting these capital ships.
For example, they rarely leave the center of a group of frigates and cruisers that protect against surface attack. And submarines that protect from beneath.
In addition, fighters on combat air patrol police the airspace above the fleet 24/7.
While at high altitude, surveillance and electronic warfare aircraft provide a God’s eye view of the battle space.
There are also multiple layers of missile defense. Aimed at everything from hyper-velocity anti-ship missiles to sea-skimmers designed to punch holes in the target vessel right at the waterline.
All this worked well enough in peacetime. And in the open ocean, with ample room for maneuver.
But it was far less effective in the crowded waters of the Persian Gulf — where projectiles fired from land may require only seconds to strike.
It also didn’t hold up well against Iran’s choreographed swarm-attacks. Simultaneous assault by dozens (if not hundreds) of suicide drones and high-speed, explosive-laden kamikaze motorboats.
That’s where AI comes in.
Virtually every bulkhead, person, or vehicle in the U.S. fleet carries sensors. AI helps collect, collate and disaggregate this high-volume information stream.
In short, it makes sure that every data-consuming war-fighter on the network gets what gets what he needs. In real time.
In this way AI transforms “the fog of war” into a powerful informational advantage favoring the U.S. side.
It’s also a key reason, why after hundreds of attacks, the Iranians haven’t sunk a single U.S. warship.
Russia and Ukraine
Ukraine’s ability to destroy the array of forces sent against it by a vastly larger enemy is itself an amazing story.
At the heart of it was a cadre of young engineers. Who saw a way to build a deadly domestic drone industry using cheap, commercially available components and off-the-shelf parts.
To accomplish this, however, they had to rely heavily on AI for real-time image recognition, machine learning, and targeting.
U.S. tech companies such as Palantir (PLTR) are reported to be working closely with them.
Since the Russians were unable to locate Ukraine’s underground drone workshops, it started instead to target the data center-hosted AI apps last month.
Related Story: Palantir Just Made the Case for Crypto with AI
Ukraine hit back with drone strikes against three data centers owned and run by Yandex (popularly known as Russia’s Google).
Unless you think peace is about to break out in Ukraine and the Mideast, I think it’s abundantly clear the war on data centers is just getting started.
Investment Implications
Defending data centers is rapidly becoming one of the toughest tasks in 21st-century warfare. That’s because their massive footprints and fixed locations given their enemies such a target-rich environment.
So, one elemental way for AI to enhance defense … is to bring on more of the radical decentralization crypto is famous for.
Here are three decentralized AI plays that could benefit …
1. Filecoin (FIL).
Very few people use their computers 24/7. Most sit idle at least a few hours every day.
Filecoin will pay you (in crypto) to rent your idle bandwidth for data stortage. Which the platform then aggregates, marks up, and resells to AI companies in the large amounts required to build out their neural nets.
In the absence of decentralized providers like Filecoin, AI developers would have no alternative but to rely on cloud computing services … supplied by attack-prone data centers.
2. RENDER (RENDER)
This is a decentralized service provider with roots in the entertainment industry.
Animated epics like James Cameron’s Avatar were so computationally intensive to produce, studios had to turn to cloud computing.
Or built their own basement of super-computers.
So, RENDER’s initial mission was providing decentralized video rendering to studios.
But then it discovered the NVIDIA graphics chips used for rendering were similarly well-suited to the equally computer-intensive task of training AI models.
3. GRASS (GRASS)
This is another platform allows users to share their unused internet bandwidth. Which it uses to harvest and structure the very latest website data to keep AI models up to date.
My colleague, Mark Gough, just analyzed GRASS’ promising first-half financial results a few weeks ago.
These three cryptos are not intended to be buy recommendations. Merely starting points for doing your own research.
Because one thing is already quite clear: A lot of folks are going to make a lot money, going forward, if they find ways to protect AI infrastructure.
We may as well be among them!
Best,
Bob Czeschin
