It’s been a tough time for the overall economy. Revenue growth has been limited, and limited revenue growth means little or no profit growth. The stock market rewards companies that show profits and punishes companies that aren’t showing more profit. So management does the only thing they can: profit equals revenue minus cost, so if revenue isn’t rising enough, cutting costs is the only answer. In most cases, that means cutting jobs, and the patterns of job losses (and gains) can tell us a lot about the future.
The 294 companies I’ve talked to, including vendors, service providers, and other verticals, say that customer support is the area losing the most jobs. Service providers had the largest job loss reported, and were the most interested in sustaining workforce reductions this year. OK, that seems to make sense given that AI is supposed to be replacing workers with chatbots in that area, but the truth is more complicated. Less than 20% of the jobs lost in this space have been lost to AI, and only a little over a third of expected job losses are directly attributable to AI. That’s our first interesting pattern.
AI does play in, though, as a kind of cover story. Many companies are interested in cutting costs, cutting headcount, but don’t want to generate questions about their health. So an excuse is in order. Companies aren’t going to say it in public, but the broad expectation that AI chatbots are going to take over support roles lets them shift away from voice support to chats, and from chats to chatbots. Most of these aren’t really based on AI (yet), but because everyone thinks AI is taking over, they view the job losses as a sign of progress. That’s our second pattern.
If AI expectations are justifying the elimination of basic customer support jobs, those same expectations are producing skilled, even expert, jobs. Every category of company says that they are hiring AI experts as fast as they can, paying anywhere from a thirty to sixty-percent premium for the skills. This, despite the fact that almost half of all companies in non-IT verticals admit they think they’re getting people who don’t have the requisite experience because they can’t recognize those who do.
What’s driving those enterprise AI-expert hires is the growing interest in self-hosted AI, meaning the use of enterprise data centers to deploy GPU systems in racks rather than to use public AI tools, or cloud hosting. While there have been enough stories on generative AI applications in email writing, creating documentation and press releases, and even writing job descriptions to fill a library, enterprises need a much better business case than that to justify any real AI investment. The only place to find it is churning through company data to make better decisions, the classic analytics applications. The problem is that these involve data so secret that companies will never let it outside their own facilities, so they need to run AI where the data is secure. Only about ten percent of companies say they have the staff skills needed for that, and they’re on a hiring binge even though they don’t know how many AI specialists they’ll need. There’s already a shortage, but companies admit that it’s being created by the expectation of AI’s future impact, not by any current impact itself. That’s our third pattern.
If things are happy in AI-expert-land, then maybe not quite so much in the cloud space. It’s not yet to the point of layoffs for most (only nine companies said they were laying off cloud software engineers now), but 186 companies said they were “slowing” their hiring of cloud-certified software engineers, and 175 of that group said they expected that would continue through this year and next. The companies who didn’t report the slowing were all either cloud providers, software/IT vendors, network operators, or network vendors.
Two factors are contributing to the decline in cloud emphasis. First, almost every company who saw cloud hiring growth declining said they’d over-hired in an excess of caution, just as they’re now doing in AI. The cloud was hot, so the candidate experts would likely be snapped up. (Get yours while you can.) Second, two-thirds of the companies said that the pace of their cloud development was slowing, even starting to decline. Yes, a part of that was due to having moved stuff to the cloud that didn’t belong there, but most was simply a matter of having used up the cloud-suitable projects.
The one unvarnished happy area in the cloud job market in what you could call the “cloud project architect” space. Enterprises in all verticals, cloud providers, and software vendors are all looking for people who can make the connection between cloud technology and business operations. The low apples of cloud opportunity have been picked, so getting higher on the tree is now essential. That’s pattern four.
There are also other areas in trouble. Outside the security space, both IT and network jobs are more at risk than typical. Vendors (like Cisco, most recently) cite softness in a market sector (service provider spending, for example) or a buildup of inventory as drivers of layoffs, but all this is just dressing up the challenge of a lack of new services or applications to drive more spending. Tech seems to be suffering from the slows. So, Pattern five.
Patterns are useful if we can draw conclusions based on them, so let’s give that a shot. Here’s what we observed:
So what can we learn from these patterns?
First, companies aren’t ready to call the end to economic uncertainty, because the signs of recovery aren’t as clear as they thought they would. At the end of last year they were worried they might miss the upswing, and many created alternative business plans, one for bad times and one for good. That good-times plan is on hold. Perhaps worst of all, that there’s a growing risk that uncertainty regarding economic recovery will in itself suppress that recovery. If everyone looks over their shoulder instead of forward, a lot of them will hit a wall.
Ironically, the patterns also suggest that too much optimism is just as bad. Companies know now that they over-hired for the cloud. They’re pretty sure they’re over-hiring for AI too, and just as the cloud boom has slowed, the AI boom will eventually slow too. Companies are overconfident about each tech wave, because they’ve been believing hype a bit too much. That may be setting them up for constant disappointment.
Finally, the patterns show that the computer, Internet, and now AI revolutions are becoming “de-popularizing.” The industrial revolution let average people build complex stuff, and created jobs overall, so it had populist outcomes. We’re seeing tech justified increasingly by destroying jobs that the average person can do, creating jobs for a narrowing group of elites. Most say this is because of AI, but AI isn’t a new threat, it’s simply a new flavor of an older one, the threat of shifting productivity goals from producing more with the same resources, to maintaining the same productivity with fewer resources. Can consumption continue to drive profit, if a lot of the consumers are marginalized or out of work? We might find out.
