Build vs Buy: Can Internal R&D Still Beat M&A?

The CODEW | Build vs Buy Series, Part 8

While Microsoft was closing its $16 billion acquisition of Mistral and OpenAI was racking up a seventeenth acquisition since 2023, Apple was doing something that looked, by 2026 standards, almost reckless: nothing of the sort. Apple deliberately limited its AI-specific capital expenditures, choosing instead to sit on more than $130 billion in cash while it bet on internal development and a licensing partnership with Google's Gemini to power a rebuilt Siri. Some analysts have openly criticized the approach as too slow. But the calculation behind it is the cleanest test case in this entire series for the opposite question every other company profiled so far has answered by buying: does internal R&D still beat M&A, ever?

Photo by Егор Ахматьяров from Pexels

The Case Apple Is Making, Whether It Says So or Not

Apple's leadership reportedly holds an internal view that large language models are heading toward commoditization — that the underlying model layer won't stay a durable source of competitive advantage for anyone, including the companies spending tens of billions to build one. If that's true, acquiring a frontier AI lab at a premium valuation buys a capability that's depreciating in strategic value even as the price tag goes up. Renting best-in-class intelligence through a partnership, while investing internally in the layer Apple actually controls — on-device integration, hardware, the interface — is a bet that the moat was never the model to begin with.

Apple isn't avoiding deals entirely. It has made smaller, targeted acquisitions — Xnor.ai for on-device efficiency in 2020, an Israeli audio-understanding startup in 2026, an AI knowledge-retrieval company days into the same year — while reportedly still exploring a possible move for Mistral. That's a hybrid, not a pure build strategy: buy small, specific capabilities; build and license the expensive, fast-depreciating stuff instead of owning it outright.

Where Build Actually Wins: The Moat Test

The clearest sign that internal R&D still beats acquisition is when the thing you'd be building is itself the durable competitive advantage — not just a feature. NVIDIA's CUDA software ecosystem, built over nearly two decades, is why NVIDIA doesn't face the same acquisition pressure as its AI-hardware rivals: nobody can simply buy their way past an ecosystem that deep and that entrenched. It has to be built, and building it first is the advantage.

The broader pattern: internal R&D tends to win when a company's spending is genuinely disciplined rather than defensive. Big Tech overall has ramped R&D spending aggressively over the past decade, but Apple and Microsoft together spend a notably smaller share of gross profit on R&D than most other large tech peers — a gap partly explained by Apple's preference for acquiring specific technologies rather than building everything from scratch, and partly by both companies' scale letting a smaller R&D percentage still fund enormous absolute budgets.

Where Build Loses: The Speed Test

The same restraint that looks disciplined can just as easily look like falling behind. Apple's AI chief retired abruptly at the end of 2025 amid internal concerns about unclear product direction, Siri's overhaul has been delayed repeatedly, and one prominent Wall Street analyst has warned Apple may eventually be forced into a large-scale AI acquisition if internal progress doesn't show up soon. Every other company profiled in this series concluded that in an environment where the underlying technology is advancing every few months, waiting to build internally means losing the market to whoever bought their way to the front of the line first. Global M&A activity is on pace to exceed $3.8 trillion in 2026, much of it justified by exactly that logic: the years-long R&D cycle that used to be normal is now, for many companies, simply too slow.

A Simple Test for Which One You're Actually Facing

Strip away the specifics of any individual company's decision and three questions do most of the work:

Is the capability itself the moat, or just a feature? If it's the moat — CUDA, a data advantage, a distribution network — building it yourself keeps it defensible. If it's a feature bolted onto an existing moat, buying is usually faster and cheaper.
Is the category still advancing faster than your internal roadmap? If yes, buying compresses years into months. If the category has started to stabilize, internal development has time to catch up without losing the race.
Can you afford to be wrong twice? Apple's $130 billion cash position lets it wait and still acquire later if its bet on Gemini and internal development doesn't pay off. Most companies don't have that luxury — for them, a wrong build decision is much harder to walk back than a wrong acquisition.

The Takeaway

Internal R&D still beats M&A when the thing being built is the actual competitive moat and the company has the balance sheet to be patient. It loses when the market is moving faster than any internal roadmap and a competitor is willing to pay a premium to skip the line. Apple's bet is the biggest live experiment on this question in tech right now, and it just moved from theory to test: the Gemini-powered Siri overhaul opened to the public beta of iOS 27 on July 14, its first release beyond developers and the first real-world read on whether patience paid off. Whether "build" wins depends entirely on whether large language models really do commoditize the way Apple's leadership is betting they will. If they don't, the $130 billion war chest becomes the down payment on the acquisition Apple was trying to avoid.

Previously in the series: Why Startups Choose Acquisition Over IPO. Next: The Economics of Acqui-Hires Acquisitions.

Advertisement

Building instead of buying? Start with a company that actually exists.

Every "build vs. buy" decision starts with a more basic one: is your company legally formed yet? Firstbase handles US incorporation end-to-end — LLC or C-Corp setup, EIN registration, a business bank account, and ongoing compliance — all from one dashboard, no lawyer required.

Start Your Business with Firstbase →

Erwin Castro

Founder & Editor • The CODEW

Erwin Castro is the founder and editor of The CODEW, an independent technology publication covering artificial intelligence, enterprise software, SaaS, cybersecurity, startups, venture capital, technology M&A, developer tools, and IT infrastructure. With more than a decade of technology journalism experience, he has contributed to Sportskeeda, International Business Times (IBTimes), University Herald, Blasting News US, and Seeking Alpha. His work focuses on explaining the business strategy, market trends, and competitive forces shaping the global technology industry.

Tech M&A  |  Cloud Computing Watch  |  Startup Funding Watch  |  Enterprise Software Watch  |  Cybersecurity Watch