For the past two decades, investors have favored “bit” businesses such as software, apps, and online platforms. The logic was simple. Software can scale cheaply, be distributed instantly, and generate huge profits once built. In contrast, “atom” businesses, which involve physical assets such as factories, logistics, and energy, were considered time-consuming, capital-intensive, and unprofitable.
That balance is starting to change.
More investors and founders are claiming that the rise of AI has reversed the scenario, making atom-based businesses relatively more attractive than bit-based ones.
Travis Kalanick’s return underscores that point. The Uber co-founder has restarted his efforts focused on manufacturing, food logistics, mining, and robotics under a new umbrella called Atoms.
“Software has automated tasks in language and mathematics, but full automation, or autonomy, of the physical world remains a largely untapped area,” he writes.
supply dynamics
This shift is happening because AI is upending the supply dynamics across technology and other industries.
AI is dramatically increasing the supply of digital goods such as code, text, images, and even video. Digital tasks that once required skilled labor can now be performed faster and cheaper with the help of AI machines. While this is great for productivity, it also means that software and other digital products become more abundant and less valuable according to the laws of supply and demand.
Joe Fath, a partner at Eclipse and an investor in the physics industry, describes this as a structural shift. He said software businesses have historically had a big advantage because they require far less capital to reach “escape velocity” to survive. In contrast, asset-heavy companies require more capital, face uncertain schedules, and are often riskier.
“That’s changing rapidly,” Fass said.
In addition to automating digital tasks, AI is beginning to make the physical world more programmable. Advances in robotics and systems that combine vision, language, and movement are enabling companies to bring software-like efficiencies to industries such as manufacturing, logistics, and energy.
“For the first time, AI has made the physical industry meaningfully programmable,” Fass said. “Opening the door to faster expansion with less capital and labor.”
Commoditization of software
At the same time, AI is putting pressure on traditional software business models. When anyone can generate good code or build apps with the help of AI, it becomes difficult to stay competitive. This dynamic is already manifest in public markets, putting pressure on software valuations.
“If AI commoditizes software, what is actually safe?” Quiet Capital partner Michael Bloch recently asked about X. His answer included:
- A regulated and accountable business (someone has to be in charge).
- Anything related to the physical world (hardware, manufacturing, energy).
- Businesses with high operational demands (“bad” businesses become the best ones).
Hard assets = moat
As venture capitalist Marc Andreessen recently pointed out, even companies building AI models face uncertainty from open source alternatives and competition from China. When core technologies become commoditized, value must be moved elsewhere.
One place that is changing is the infrastructure, the physical backbone of AI.
Big technology companies like Microsoft, Google, Amazon, and Meta are pouring hundreds of billions of dollars into data centers, chips, networking equipment, and energy. These are very physical, asset-intensive investments. In other words, even the largest “bit” companies are becoming “atom” businesses.
Googlers famously wrote in a 2023 internal memo, “We don’t have an outer moat, and neither does OpenAI.”
This reflects a broader perception. If AI models become nearly equivalent, the winner may be the one who can deliver them the fastest, cheapest, and most reliably. It depends more on the physical infrastructure than just the software.
Fusion of real and digital worlds
The same trend is happening across all industries. Companies like Tesla, SpaceX, and Amazon have been blending software and real-world operations for years. Now, new players like Wave, Anduril, and Redwood Materials are doing similar work, using AI to improve everything from defense systems to battery production. Importantly, AI will complement these businesses, not replace them.
“Physical industries still need to build and operate physical objects,” Fass said. “AI cannot fully replace this in terms of taking over call centers or acting as coding agents.”
He expects there will be many more examples from companies like Redwood and Anduril. “Capital that once flowed primarily into software and consumer internet will increasingly shift in this direction over the next few years,” Fass added.
Impact on the job market
This change is also reflected in the job market. Roles related to digital work, such as programming, customer service, and data entry, are increasingly exposed to AI-powered automation. Meanwhile, demand for skilled physical occupations such as construction, electrical work, and maintenance is increasing in part due to the creation of AI infrastructure itself.
“AI is actually still digital,” Elon Musk said on a recent podcast. “Ultimately, AI can improve the productivity of humans who do things with their hands, like welding, electrical work, plumbing. Anything that physically moves atoms, like food preparation or farming.” “These jobs are going to be around for a lot longer. But the digital stuff, where you have someone sitting at a computer doing something, is going to be taken over by AI like lightning.”
scarcity = value
Still, the move to atoms is not without risks. Physical businesses remain difficult to scale, something Musk calls “production hell.” It requires significant capital, operational expertise and strong execution.
Firth acknowledges the challenge. “You don’t just have to solve a technical problem, you have to scale it. And when you’re physically building something, it’s incredibly difficult to scale it properly.”
Still, the direction seems clear. As AI lowers the cost and value of digital works, scarcity, and therefore value, is returning to the physical world.
“The future is physical,” former OpenAI researcher Rohan Pandey recently wrote about X. “When AI reduces the cost of computer work to up to zero, the bottleneck (and thus capital) returns to the physical world.”
In that sense, the future of technology may look more like a fusion of bits and atoms than pure software. The latter will play an increasingly central role.
Sign up for BI’s Tech Memo newsletter here. Please contact us by email. abarr@businessinsider.com.
