Buying speed: IP acquisitions in the age of AI code generation

Public and private market investors demand double-digit growth from enterprise and business-to-business (B2B) software companies, including software-as-a-service (SaaS) providers. Organic growth rarely meets those targets, so strategic acquirers use their balance sheets to buy growth via M&A.

Most of those deals buy revenue, products, customers, and free cash flow. But a recent strategic shift will impact acquisitions – machine-developed code.

Strategic companies will still leverage intellectual property (IP), just not in the same manner they have for the past 50 years. They will leverage IP for unprecedented speed.

Speed is the strategy

Established companies have effectively optimized what they do. They design, build, sell, and support products. The dependability and reliability of those products successfully built the global technology industry, but it came at the expense of rapid creation and iteration. Historically, the case for an acquisition was to gain new technology, products, customers, or labor. IP was almost always an afterthought.

Why?

Because many enterprise buyers wanted new capabilities but would not accept the associated startup vendor risks. Those included uncertain funding, thin support, and immature security, among many others. When a mature acquirer applied its own quality and security standards to acquired technology, those enterprise buyers felt more comfortable adopting it.

The acquiring company liked this arrangement too. An acquisition in a critical field delivers a core technology far sooner than an in-house team can build it. If the acquirer had the talent and bandwidth to build it, it would have already done so. Buying the capability moves the cost from research and development (R&D) on the P&L to the balance sheet and delivers it in a fraction of the time.

It also removed risk. Internal development starts with an untested version 0 product. An acquired company brings working technology and customer traction. The acquirer buys proof, not a hypothesis.

AI moves IP from a sleepy asset to game changer 

For decades, software and SaaS companies spent most of their R&D budgets writing code. Software development produced more than code. As engineers built, they invented novel methods, algorithms, and architectures that became patents, alongside source code protected by copyright. The act of building was where a company’s IP was born.

Artificial intelligence (AI) has changed that. AI coding tools turn a well-defined specification into working code in hours instead of months, and companies now pay for tokens instead of engineering hours. Code generation is no longer the hard part.

The hard part is holding the right to build. Struggling and failing software companies often own valuable patents, designs, and architectures they lack the capital to commercialize. Buying that IP gives an acquirer the legal right to the underlying innovation – and AI turns those rights into new features or products in a fraction of the time an old build used to take.

The economics are compelling. A distressed seller prices its IP well below the original R&D cost, and the acquirer skips the workforce, contracts, and liabilities it does not want. Precise design inputs also keep token spend in check, since vague instructions force expensive rewrites. Buying the IP rather than the company is the fastest, cheapest path from idea to product.

Where IP lives now

Invention assignment agreements once secured a company’s ownership of every method and line of code its engineers created. AI breaks that link. Models from Anthropic, OpenAI, and Google, among others, now write much of the code under subscription agreements. Given identical prompts, models from different vendors will not produce identical output, but the differences seldom amount to a protectable invention. The machine builds without inventing on the company’s behalf.

The law reinforces the point. In January 2025, the US Copyright Office concluded in Part 2 of its AI report that purely AI-generated material is not copyrightable and that prompts alone do not make a user the author. In March 2025, the D.C. Circuit held in Thaler v. Perlmutter that copyright requires a human author.

As more code is machine-written, fewer new patents and copyrights emerge from the build process. IP created in the era of human engineering becomes scarcer and more valuable – and much of it sits inside companies that cannot commercialize it. Acquirers that buy those portfolios secure rights competitors can no longer generate by building.

From defense to offense

Large technology companies have purchased patent libraries to keep portfolios out of hostile hands and inoculate themselves against patent trolls.

Strategic companies will now make IP purchases to speed new features and kinds of products to market. Buying a distressed company’s patents, copyrighted code, and designs, then building on them with AI, gives an acquirer freedom to operate and a head start no internal team can match.

Acquirers that move early will build pipelines of distressed IP targets and lock up portfolios before the market catches up. When code is cheap, the right to build is the scarce asset. Buying speed now means buying that right first.