Summary Bullets:
• Service providers, tech companies, and enterprises are under extraordinary pressure to develop and deploy AI solutions.
• Enterprises will need to re-evaluate their goals for the rest of the year, considering cost increases and possible delivery delays.
In the late 1930s, Sociologist Robert K. Merton wrote about unintended consequences, where some of the outcomes of purposeful action are not predicted or intentional. He proposed three types of unintended consequences – unexpected benefit, unexpected drawback, and perverse result. In the first, there is a positive benefit, something good happened that wasn’t foreseen. In the second, an unforeseen detriment occurs. In the third, the result makes the problem worse, a contrarian result.
He also noted that there are five broad possible causes of unintended consequences. Of those five, two are applicable to the situation with AI today. First, errors in analysis, assuming that practices and results of the past are applicable to the current situation. The second is immediate interests overriding long-term interests. There is another, but not from Merton – the concept of groupthink that leads to unintended consequences.
Service providers, tech companies, and enterprises are under extraordinary pressure to develop and deploy AI solutions. The dominant narrative says that without AI, businesses are likely to fail or be at a severe disadvantage. It extends nationally as well – the dominant narrative also tells us that nation-states each need to be a leader in AI, lest they be at a disadvantage from a military or economic standpoint. Resources and treasure, the dominant narrative tells us, are necessary to bring AI into being everywhere. Business and technology leaders have nodded vigorously and set forth to bring AI into existence everywhere. But AI is expensive – the hardware is expensive, the power and cooling costs are astronomic, and there are nowhere near enough data centers to house all of it, let alone power to run them. All of this has resulted in the AI boom.
This is the place the enterprise information technology market lives today – in the crater of unintended consequences caused by the AI boom.
Consequences
Setting aside all the social and labor impacts that AI is having, the AI boom is making some basic technology resources very expensive. Basic technology such as RAM and storage, both flash and spinning rust storage (hard drives). The manufacturing capacity for the expensive RAM that is used in AI systems is pre-purchased at least through 2026, likely into 2027 as well. Because the RAM used for AI data centers is much more profitable, there are RAM manufacturers converting production lines away from more common RAM memory for things like desktops and some servers to the more lucrative RAM used in AI systems.
The same thing is happening with storage. Flash-based storage manufacturing capabilities have been sold at least through 2026. The shortage is so acute that it has even revived the legacy market for older storage technology in traditional steel hard disks. Which are now also having shortages. Scarcity means price increases.
The costs for everything that uses RAM and storage have gone up considerably. For enterprises, this means that almost every piece of IT equipment they buy for their own use is getting considerably more expensive and there is a real chance that these manufacturing capacity issues will lead to significant delays. It’s not just servers, networking, and storage in the data center, it’s also desktops, smartphones, printers, and so many other things that use memory and storage to function. All those SaaS services also need these resources just to grow the base services, let alone AI services. That means that monthly prices are headed up. Cloud services will see price increases as well, especially from smaller cloud companies unable to secure an order a year in advance. This will hit consumers as well, who will find their buying power again shrunk as the economics of scarcity drive up prices of technology like smartphones, TVs, tablets, and home computers/laptops.
Enterprises will need to re-evaluate their goals for the rest of the year, considering cost increases and possible delivery delays. Enterprise buyers and planners need to have a heart-to-heart with their VAR or vendor partners and get an idea as to how much this new scarcity will affect them. This year’s upgrade projects may need to be re-evaluated and possibly pushed back.
This is where the errors of short-term gain over long-term gain, groupthink, and errors in believing that AI would be like cloud or the internet itself, an unstoppable economic juggernaut. Whether the push for AI has been too fast, if AI can achieve the fantastic promises made about it, is a moot point. The AI boom, which is fueling the data center boom, is having unintended consequences. That is and will continue to drive prices up for enterprises and customers.