In partnership with

One view of the lifecycle of a business is that it raises money early, and then spends its mature years returning most of that money to shareholders. This is mostly true: Corporate America still invests, and some companies invest heavily, but in the aggregate last year the S&P 500 returned 86% of net income to shareholders in the form of buybacks and dividends. Put another way, their net retained earnings were about 0.6% of their market value.¹

Within those retained earnings, a large proportion of a typical company's capital expenditures have some predictable unit economics. A cloud company knows roughly what return the next datacenter will have, and a SaaS company has a pretty good range for when hiring a new cohort of salespeople will break even. But this is not the only thing companies spend money on. Amazon (disclosure: long) likes to invest in little side projects that can turn into big businesses, like Kindle and AWS (or, less pleasantly, the Fire Phone and various attempts to rethink brick-and-mortar shopping). Google has poured money into self-driving cars and drone delivery. Meta (disclosure: long) has its Metaverse. Microsoft (long!) has had a research group since 1991, where they've done work on speech recognition, security, their own strongly-typed functional programming language on the .NET framework, etc.

These sort of side projects are more colloquially known as corporate R&D labs, and they have existed for a long time; Carnegie Steel hired metallurgical researchers as early as the 1880s, and DuPont scaled up its chemical research in the 1920s and 30s because of a combination of bumper profits from selling gunpowder in the First World War and terrible PR from the same (there's nothing like your congressional testimony being referred to as the "Merchants of Death" hearing to motivate a pivot into a new business). Bell Labs did research as a sort of antitrust compromise—they were creating positive externalities from cheaply licensing some of their discoveries, which made their phone monopoly more tolerable. And Xerox PARC famously provided some free UX consulting to promising young founders in the 70s (and developed the laser printer).

R&D is not a normal capital allocation decision. It's a bit like buying flowers or chocolate on Valentine's Day: if you skip it long enough, you'll regret it. But that makes it hard to justify any particular budget. At some point, Microsoft has to say that they've brought on enough PhDs and can allocate their money to something else, but that point is hard to measure.

One way of looking at it is that product cycles in tech are not just short but uncertain. You don't know if a given category is going to be around for decades (word processing, email), or will be replaced by free alternatives (browsers, databases). The best way to spot this risk, per a PARC veteran, is to invent it.

But there are other reasons to invest in pure R&D without direct commercial applications. The skills involved in the pure research side often have some useful spillovers to the more commercial development that turns an idea into something with a price tag. So investing in R&D is a way to buy a sort of talent buffer: when business is good, researchers can focus on things that won't be practical for a decade, but when the business is threatened, there's a talent bench that can be redirected towards the projects that matter more in the next quarter or two.

This also helps with recruiting: even if a lot of the economics behind Google are driven by nudging something a pixel or two over and seeing the revenue tick up by a few basis points, or finding additional efficiencies in indexing and serving voluminous data, the company can point to wilder projects both to a) entice employees with the opportunity to do something cooler and b) to give them the sense that they're not just in the business of selling more ads to fund bigger buybacks.

And it's also a good way to keep senior executives engaged. It's usually bad for morale when the founder steps down, even if the founder has perfectly good reasons to do so. Bill Gates wasn't so much done with Microsoft as he was done with the DOJ, and he correctly believed that eliminating diseases that have afflicted humanity for millennia would be easier than figuring out what, exactly, Microsoft was still allowed to do.

Founder engagement is a difficult puzzle, because by the time it's a problem, the company has evolved far from what it was when the founder had more control and knew every employee personally, and because of the usual selection effect—whatever class of problem a company is exceptional at solving (technical, sales, interfaces, whatever) gets solved, and what's left is whatever the company is relatively bad at. So running a successful company is basically a process of growing your job into something you hate. Spending more on R&D is a way for the company to stay interesting to someone who ideally should be around the office and excited to be there when the core business runs into some kind of problem. Since the founder is already pretty rich, giving them more stock options usually doesn't move the needle much (and, in the case of Tesla at least, a needle-moving sum of money can also be so high that a court decides it's excessive and reverses it).

Recruiting new people and retaining the most senior people is not usually the main justification for corporate research projects, and in the very long run the only way a tech company stays a "tech" company is by inventing new technologies, which is an expensive and uncertain process. But when the decision to set next year's R&D budget gets made, it can't be in reference to some kind of direct P&L impact over the next year; basically nothing gets from the lab to customers that fast. Fortunately, those more talent-oriented social reasons always apply; they make temporary cuts too expensive in morale terms to be worth contemplating, and ultimately mean that the maintenance of vibes is a key input into the next generation of technological advances.

Put your money to work in a high-yield cash account with up to $2M in FDIC† insurance through program banks.

Get started today, with as little as $10.

Read More in The Diff

In The Diff, we’ve looked at corporate moonshot R&D from a few different angles:

• We’ve profiled IBM and AT&T ($) during their glory days of moonshot bets.

• We’ve examined how durable these businesses are and how much speculative research they can support ($).

• Some companies, like Antithesis, consist entirely of moonshot bets that gradually pay off.

• And here’s more on Google’s drone delivery.

1. This picture is a misleadingly bleak look at how much companies invest. A company big enough to be public is probably mature enough that its future results can be modeled with some degree of confidence. That matters to equity investors, but it matters even more to lenders—a bondholder looking at some company with a 50/50 shot of rising 10x or going bankrupt will focus on the bankruptcy odds, not the upside, because they won't be able to capture much of it. So as companies mature, they’ll reduce their reliance on equity financing and use bonds instead.

Join the discussion!