Software Is Leaving Silicon Valley; AI Is Turning It Into A Local Trade
Kognitos Founder & CEO Binny Gill previously served as CTO at Nutanix and holds nearly 100 patents in computer science.
In the early days of electrification, power companies did everything. They generated the electricity, ran the lines and sent their own workers into buildings to wire them up. That arrangement did not last. As the grid matured, the utilities retreated to what they did best, which was supplying power, and a new trade emerged. Electricians became independent professionals, hired by every household, every shop and every factory to wire, customize and maintain the systems that ran on that power.
Something very similar, if not identical, is happening to AI-driven software right now, and the story is bigger than jobs. It is about where the capability to build software lives.
For the last two decades, the ability to build software has resided within the SaaS providers because that is where the tools, the infrastructure and the talent are. Frontier AI models are breaking up their monopoly on capability as world-class software-building ability now lives inside the AI models, and the models are available to everyone.
At the most basic level, your local barbershop now has access to the same software building capabilities as the largest technology companies. The “scarce IT resource” is no longer programmers or programming ability; it is knowing what problems need to be solved for the business with AI-driven software.
Look at the fastest-growing job title in technology right now. Job postings for forward-deployed engineers, the people technology companies are embedding inside customer organizations to make AI work in production, have grown more than 1,100% year over year, according to an analysis of jobs data by Bloomberry.
Why is this particular role exploding? Because businesses outside the technology industry are rapidly shifting to building software instead of buying it off the shelf and configuring it, and they cannot recruit the engineers fast enough to build it. So technology companies are lending out their own. It is a transitional arrangement, the same way early power companies once dispatched their own wiremen. The demand exists, but the permanent structure to serve it does not yet.
The pressure behind that demand is real. MIT researchers found that 95% of enterprise generative AI pilots showed no measurable business impact. A lot of it failed because the know-how of how to apply AI to provide deterministic outcomes that provide ROI to businesses was still lacking in-house. We know that the bottleneck in enterprise AI is no longer model capability, but getting software to work in the messy, specific reality of an individual business.
Access to the models alone is not enough, just as access to the grid does not wire a building. An electrician does not need to be a physicist to make electricity useful and safe, but you still need the electrician.
Knowing how to direct AI to build software that is secure, reliable and maintained over time is a skilled trade in its own right, and a new class of practitioner is already forming around it. Picture how they will work, because it looks nothing like a Silicon Valley engineering team.
Someone walks into a barbershop, or a dental practice, or a freight brokerage, and asks the questions a good tradesperson asks. How does this business run? What do you want the screen to show? Where does the business make losses and why?
Then they build it, in a day or a week, depending on how custom the owner wants it, and they move on to the job next door. When the software breaks, they come back the way an electrician does, because when your systems are down, your business is in trouble. These practitioners will work for everyone else, wherever there is an internet connection.
Businesses never actually wanted identical software. They accepted it because bespoke software was too expensive to build, so thousands of companies bought the same systems and then spent heavily trying to customize them at the edges.
More affordable custom software does more than save money. It lets a business run differently from its competitors, and the difference is the whole point. No two hotels are laid out the same and no two restaurants serve the same menu, yet nearly every business runs on the same handful of systems, because until now, sameness was the price of affordability.
There is a second force pushing in the same direction. The cloud's promise of safety rested on concentration: The biggest providers employed the best security engineers, so your systems were safer in their hands. That assumption weakens when the ability to attack democratizes along with the ability to build.
As AI lowers the bar for offense as well as defense, more businesses and governments will want critical software built close to home and kept within their physical control. Sovereignty and proximity are becoming features.
The last 20 years concentrated software creation inside a handful of technology companies because that was where the expertise lived. That era has ended. The capability to build software now exists anywhere there is an internet connection and a business need. Silicon Valley will no longer be the center of gravity of software because understanding business needs is the remaining challenge which is best solved locally where the businesses are globally. The engineers will be forced to move out (at least virtually if not physically) of tech hubs and into hospitals, retailers, manufacturers, law firms, accounting practices and small businesses.
There are an estimated 47 million developers worldwide, according to analyst firm SlashData, and their work is not disappearing. It is dispersing, the way industrialization once pulled work off farms and into cities; the dislocation was real, and so was the new work waiting on the other side.
Software engineers are not vanishing. They are becoming the local tradespeople of the AI economy and getting a new name: FDE.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?