Brick & mortar businesses are valued at 10x EBITDA. Tech companies in turn are valued at 10x their ARR. A traditional business with $100MM ARR and $10MM EBITDA can hope for $100MM price tag. A tech company with the same numbers would be worth $1B.
So here's my pitch for private equity owners, management teams and startup founders working with them: leverage AI aggressively to turn brick-and-mortar companies into tech companies. Do so and you will increase significantly their valuation. Be the first to do so in your industry and you might even build market leaders. Don't do it and risk going out-of-business.
Transformational AI Projects
Right now, Delos AI Inc. is working with 2 enterprise clients on large-scale AI implementation projects. I like to call those projects "semi-transformational".
The "transformational" part is because for both clients our work aims to apply AI to a large portion of their operations. Once completed, both companies will operate in a completely different way - with large portions of their business running automatically. Those projects encompass a mixture of AI, software engineering, and process definition. We go deep on a given industry and try to re-think what it would mean to run the given company with automation at its core. Both companies will see clear impact on their EBITDA over time.
The "semi" part is because I believe we could do more for those companies. I believe our implementation could be faster, more encompassing and drive even more value. In particular – if our work was taken to its most fruitful conclusion, such projects could lead to transforming those businesses into tech-first companies.
I believe there is a need in the market for fully transformational AI projects. Projects where a company would re-design itself from scratch with AI at the heart of its operations. Obviously such projects would go beyond just AI – they need to be all encompassing. In fact, they should involve:
- Strategy re-thinking
- Business model changes
- A RIF plan
- Changing P&L and capital allocations
- Ontology definition
- Story-telling and re-branding
- Process re-design
- Software engineering
- And eventually AI
Such Transformations Are Possible and Needed
It doesn't take a Ray Kurzweil to see that virtually every company in America could be run more efficiently with a combination of latest Generative AI models and some good old Machine Learning. AI can already automate documents, phone calls, and emails. That's virtually 99% of what office work is about. Machine learning can also do wonders at time-series prediction, demand forecasting, or logistics optimization. Applying AI to industries that have been traditionally tech laggards can drive significant upside. This is why so-called AI-powered roll-ups have gained a lot of attention – ie. acquisitions of non-tech businesses by tech startups that unify them under a common AI-powered OS.
So if you ask me why do it, the first argument is because we can. If the technology to make your business more efficient exists, there is no good reason why it shouldn't be made more efficient.
Second, I believe that AI or no AI, we are due for an "efficiency reset" among businesses. It does not take a Jack Welch to see that companies today suffer from significant managerial and people bloat. Designing efficient processes is hard, hiring more people is easy. And so, every bad process is fixed with more hires. More hires means you need managers, so you hire directors. Get enough directors and you become qualified to hire a VP. Get enough VPs and you unlock C-suite hires. After enough managerial hires you get into the rarified air where people can't do work because they are wasting time on managing or being managed. No one is accountable for anything, everyone is busy. Welcome to Corporate America 2025AD.
Far be it for me to espouse principles of communism, but I genuinely believe that if we had more individual contributors and less managers the world would be a better place. Far too many people today manage instead of doing actual work. If AI is nothing else but a forcing function to make corporate structures leaner, with less/no managers, then all the money spent on AI data centers will already will be worth it.
Finally, there are interesting opportunities that could happen if a significant portion of American companies reduced their OPEX. What if that was reinvested into more R&D and innovation? What if it could be used to fund new businesses? Or what if we just gave it to solve civilizational problems?
Why It's Hard
But let's pause the daydream here. Because doing such transformations is difficult. First of all, it requires a combination of a McKinsey, with an Accenture, a Palantir and more.
You can't reach that level of transformation with off-the-shelf AI. Or even with just AI for that matter. That's why AI roll-ups are taking off and why full-stack data engineers are suddenly everywhere. Because deep transformation requires custom AI built for one specific client - trained on their data, tailored to their workflows, aligned with their strategy.
Even then, more than AI is needed. You need:
- Process redesign
- Perhaps hardware innovation
- New communication models
- Storytelling that brings teams along for the change
- Often a redefinition of what the company is
The hardest part isn't the tech - it's the management. We've seen companies treat AI innovation like a feedback exercise: asking employees what tools they'd like and trying to build around that. But no one will suggest automating their own job. And many organizations have grown by adding layers of managers instead of fixing the system itself. AI exposes that structure - and that's what makes transformation painful.
The Reward
But the pain is worth it. Because economics makes it impossible to ignore.
For shareholders, this isn't just incremental. A company that's currently worth $100-200 million can become a $1 billion business simply by changing how it operates, without adding a single new customer.
That's an $800 million gain before revenue growth - and it comes from embedding intelligence into every layer of the business.
The logic is simple:
- Brick-and-mortar companies are valued on profit multiples because they scale through people.
- Tech companies are valued on revenue multiples because they scale through code.
AI gives traditional businesses the ability to scale like tech - and once that happens, markets start valuing them like tech.
But the upside isn't just financial. When AI becomes part of the operating fabric, the company starts compounding faster. Decision-making speeds up. Data flows more freely. Margins improve. OPEX can be reinvested in more growth or R&D. Those improvements compound, quarter after quarter, putting the company on a different level.
What's In It For Startups?
Now what if you're a startup? Should your offering be to help companies with such all encompassing transformations as opposed to whatever product you want to build? To make the case for a "yes" answer, let me run some numbers to show how much you could earn from such deals.
Take Palantir as a benchmark. Their top twenty clients generate around $64.6 million per year each, or roughly $325 million over 3–5 years.
Let's assume a startup works with a mid-market company to increase its valuation from $200MM to $1BN. Let's assume this takes 3 years. How much should your company get? Effective management teams, CEOs, search fund owners or PE funds will often receive 20-30% of the upside they manage to generate in flipping a business (be it via carried interest (as for PE funds), or some other equity stakes). So, a modest 15% of the upside means $120 million over 3-years for a mid-market engagement. Roughly $40 million per year or 2/3 of what Palantir makes from their top clients. Handle five such deep transformations in parallel, and that's $600 million in three years! Not bad for a company that effectively has only 5 clients.
On top of those numbers, there's a strategic payoff too. Each deep engagement gives the startup proprietary insight into how an industry works - its workflows, data, and decision patterns.
That knowledge compounds across clients. Over time, the startup that drives these transformations doesn't just deliver AI - it owns the operating playbook for the sector.
The Moneyball Effect
Early movers will reap the benefits; late movers will be forced to catch up. Then the field will even out.
I call it the Moneyball effect. The first baseball teams that adopted analytics gained an early advantage. But as others followed, analytics stopped being the edge - it became the norm.
The same will happen with AI. For first movers it will be an advantage. Then, it will become a baseline.
But, unlike in baseball, the late movers are not guaranteed a seat at the table. In fact, they're not guaranteed to even exist. Companies "going out of business due to AI-powered competitors" will be a trend that we will see most clearly in commoditized sectors - industries like energy, logistics, and transportation, where clients buy on price and availability – usually choosing the cheapest option.
Customer experience in those industries is limited to what I call the "don't f up" standard**: be on time, be professional, be responsive, handle paperwork correctly. Beyond that, there's no extra points you can score for a better brand. A client might leave FedEx if it's cheap but unreliable - but no one will pay DHL double because they like the logo or ads.
In those sectors, efficiency is the brand, and AI will decide who survives.
Other industries will still have moats that AI can't replace - brand, expertise, and human trust. A beverage company can still charge $10 for a soda if clients like the can design. Likewise, a hospital can still attract patients if it has the country's top oncologists.
But for everyone else, AI will become the baseline.