Software Eats, Then Software Frees: The Three Acts of Every Industry

A thesis in plain language

In 2011, Marc Andreessen wrote that software is eating the world. He was right, but he stopped one chapter short. Software does not just eat the world. Software eats the world, then it consolidates the world, then it gives the world away for free. That is the arc. Three acts. Every industry runs the same play, just on a different clock.

I want to make the case for that arc, walk through a few parallels, and then show what it looks like when you point it at multifamily real estate—an industry that, by the way, is sitting right at the start of Act One and does not fully realize it yet.

Act One: Software eats the workflow

The first thing software does to an industry is not glamorous. It does not invent a new product. It just digitizes a workflow that used to live on paper, on phones, in filing cabinets, in someone's head. Word processing replaced typewriters. Spreadsheets replaced ledger pads. Email replaced inter-office memos. Listing sites replaced classified ads.

Nothing about the underlying job changed in Act One. People still wrote letters. People still did accounting. The difference was that the work moved from atoms to bits, and once it lived in bits, it could be copied, searched, shared, version-controlled, and improved a thousand times faster than the analog version it replaced.

This is the act everyone notices, because it is the act that makes headlines and creates the first wave of obvious winners. It is also the act that puts the word "disruption" on a TED stage and a tote bag.

Act Two: Software consolidates the industry

Here is the part that gets underappreciated. Once a workflow is digitized, the economics of the industry change in a very specific way. Marginal cost collapses. Distribution becomes free. Data compounds. And the companies that move first build moats out of scale that nobody competing the old way can ever climb.

So you do not just get software in the industry. You get one or two or three giants who own the industry. Not because they are evil, and not because regulators were asleep, but because the underlying mathematics of digital businesses rewards consolidation in a way that physical businesses never did. A factory has a maximum size. A database does not. One needs a parking lot. The other needs a power bill.

Look at retail. Millions of catalog and storefront operations became Amazon. Look at search. A whole industry of directories and yellow pages became Google. Look at rides. A century of taxi medallions became Uber and Lyft. Look at short-term rentals. A fragmented bed-and-breakfast cottage industry became Airbnb.

In each case, the same pattern. Digitize the workflow. Aggregate the supply. Aggregate the demand. Own the rails. Set the prices. Take the rake. Become, for a window of time, a kind of dominant player the pre-digital version of that industry could not have produced.

This is the phase that scares people, and it should. The interim winners of Act Two get bigger than anything the analog version of the industry ever produced. They look permanent. They feel permanent. They are not permanent. Neither was Blockbuster, and Blockbuster had a logo you could see from space.

Act Three: Software becomes free, and the work becomes robotic

Two things happen at the same time in Act Three, and they are deeply connected.

The first is that the digital execution layer—the software itself—drifts toward zero cost. This is not a prediction. This is just what digital things do over a long enough timeline. Word processors used to cost hundreds of dollars and came in a box. Now you can write a novel in Google Docs for free. Encyclopedias used to cost a thousand dollars a set. Now Wikipedia is free. Operating systems used to be a moat. Now Linux runs most of the internet, and you can download it for nothing. Photo editing, video editing, 3D modeling, code editors, databases, email clients, messaging—every category that started as a paid product eventually grew a free, open-source, or commodity equivalent that was good enough for the majority of users. The dominant player of Act Two does not always die, but the pricing power does.

The second thing is that once a workflow is fully digital, it stops needing a human to push the buttons. Software that manages a process can be wired to robotics that execute the process. The digital system stops describing the world and starts manipulating the world. A warehouse management system becomes a warehouse full of robots. A trading algorithm becomes the entire trading floor. A CAD file becomes a CNC machine making the part with no machinist. A self-driving stack becomes the driver. The line between "software" and "operations" dissolves, because the software is the operations.

Put those two together and you get the end state of every industry that goes through this cycle. The execution layer is robotic. The software layer is open source or near-free. The dominant player of Act Two is still around, often still profitable, but no longer extracting the rents it used to. The cost to the end consumer falls toward the cost of the underlying physical inputs and energy, and not much more.

That is the arc. Eat. Consolidate. Free.

Now point it at multifamily real estate

Multifamily is interesting because it is one of the largest asset classes on earth, it touches almost every household, and it is still being run, in 2026, more or less the way it was being run in 1986. There are spreadsheets now. There are property management systems. There are listing sites. But the actual work of owning, financing, acquiring, managing, and operating apartment buildings is still overwhelmingly a human-mediated, relationship-driven, paper-shuffling business. Multifamily is sitting at the very front edge of Act One. The industry's most advanced technology is, in many shops, a very confident man with a very large spreadsheet.

Let me walk through what each layer looks like as the three acts play out.

Acquisition. Today, deals are sourced through brokers, relationships, pocket listings, and a lot of phone calls. In Act One, acquisition becomes a data problem. Every parcel, every owner, every loan maturity, every rent roll, every permit, every code violation, every demographic shift becomes queryable. The good operators stop hunting deals and start subscribing to deals. In Act Two, one or two platforms aggregate that data so completely that originating a deal outside the platform becomes irrational. In Act Three, the matching of capital to asset becomes automatic, the underwriting model is open source, and the spread that today's acquisition shops earn for being smart and well-connected compresses toward the cost of the compute that runs the model.

Capital raising and equity deployment. Today, this is the most relationship-bound part of the entire industry. You fly to meet LPs. You build decks. You manage a CRM by hand. You wait six to nine months for a commitment. In Act One, capital formation digitizes—data rooms, investor portals, automated reporting, programmatic distributions. In Act Two, two or three platforms become the default place capital meets sponsors, and the platforms set the terms. In Act Three, the syndication itself is just a smart contract, the diligence is performed by an agent reading every document instantly, and the friction cost of moving a dollar from an investor to a building approaches zero. The 2-and-20 economics that the sponsor class earns today exist because raising and managing capital is hard. When it stops being hard, those economics compress. Two and twenty becomes zero and a thank-you note.

Asset management. Today, an asset manager is a person with a spreadsheet, a few dashboards, and a calendar full of calls with property managers. In Act One, the spreadsheet becomes a system of record that pulls live data from every property. In Act Two, a handful of platforms become the operating system of the asset class, the way Bloomberg became the operating system of fixed income. In Act Three, the asset manager is an agent. It watches every variable on every property continuously, reforecasts in real time, flags decisions that require a human, and executes the rest itself. The role does not disappear, but the leverage per human goes up by one or two orders of magnitude, and the management fee follows the leverage down.

Property management and operations. This is where the second half of Act Three—the robotic half—matters most, because property management is the one part of multifamily that is unavoidably physical. Toilets clog. HVAC fails. Leaves fall. Trash accumulates. Tenants lock themselves out. For a long time, this looked like the moat that would protect the industry from full digitization. It is not a moat. It is a delay. Toilets are not a business model. Leasing is already moving to self-tour and digital application. Maintenance triage is already moving to AI vision and predictive sensors. Turn work, landscaping, cleaning, and inspection are all on the path to robotics, and the path is shorter than most operators think. The end state is a building where the digital system does the dispatching, the diagnosis, the scheduling, the vendor management, the accounting, and a growing share of the physical work itself, and a human shows up only for the genuinely novel problems.

Ownership. This is the strangest one to think about, and the most important. Today, ownership of an apartment building is a legal and financial construct held by an LLC, with a bank, a sponsor, LPs, and a stack of documents. In Act One, that construct becomes digital. In Act Two, ownership records and the secondary market for those records consolidate onto a small number of platforms. In Act Three, ownership is liquid, fractional, and programmatically transferable, and the cost of holding, transferring, and operating that ownership trends toward the cost of the underlying physical asset plus a thin layer of compute. The "alpha" that today's owners earn for being good at the game compresses, because the game itself is now a public utility.

The interim phase is going to be wild, and brief

I want to be very clear about something, because I think people in real estate either dismiss this entire arc or skip straight to the end of it.

The interim phase—Act Two—is going to produce dominant players in multifamily larger than anything the industry has ever seen. Not the largest REIT today times two. Something categorically different. A single operator, or a very small handful, that owns or controls a meaningful percentage of the institutional rental housing stock in the United States, runs it on a unified digital and robotic stack, and underwrites, finances, and trades it on a unified platform. That is what the math points to, and the math has been right in every other industry that has run this play.

That phase is coming, and it is going to feel permanent while it lasts. It will not be permanent. It will be the bridge between an analog industry and a near-zero-cost one. The dominant player of Act Two in multifamily will, in the fullness of time, find itself in the same position as the dominant word processor of 1995 or the dominant encyclopedia of 1990. Still around. Still useful. No longer extracting the rents that defined the era.

What this means for the end consumer

The reason any of this matters is the renter. Right now, in most American cities, rent is the single largest line item in a household budget, and it has been growing faster than wages for a generation. The reason is not greed and it is not zoning alone. It is that the cost stack of producing and operating rental housing is full of analog friction—acquisition friction, capital friction, management friction, maintenance friction, vacancy friction—and every layer of friction is a margin that someone has to earn, which means a dollar the renter has to pay.

Take the friction out and the margins compress. Compress the margins all the way and the cost of housing the marginal renter falls toward the cost of the land, the materials, the energy, and a thin layer of compute. That is not a utopia. Land is still scarce. Materials still cost money. Energy still costs money. But the soft costs, which today are most of the stack, can go to almost nothing, and when they do, the price of shelter for an enormous number of people goes down in a way that no housing policy in the last fifty years has managed to deliver.

That is the prize at the end of Act Three. Not a better landlord. Not a smarter REIT. A structurally cheaper category.

Closing

Software ate the world. Then software consolidated the world. Then software is going to give large parts of the world back, at a price that approaches the cost of the atoms involved. Multifamily is going to run that play. The interim winners are going to be enormous. The end state is going to be cheap. The only real question for anyone working in the industry today is which act they want to be building for.

I am building for the third one. The first two are already crowded, and the third one has better margins on the way in and a better world on the way out.