Amazon Doesn't Believe in Research and Development Spending

Maybe its approach to tech and content investment is one that other tech companies should think about copying.
Amazon Doesn't Believe in Research and Development Spending
The Amazon logo sits on an Inc. pickup and collect locker. (Photographer: Chris Ratcliffe/Bloomberg)

(Bloomberg View) -- In a series of letters that it sent to Inc. executives during the fall and winter and released to the public last week,  the Securities and Exchange Commission urged the Seattle-based e-commerce and cloud computing giant to disclose its research and development spending as other companies do. It seemed like a reasonable enough request -- especially given that, as I discussed in a recent column, Amazon now appears to lead the global list of corporate R&D spenders by a multi-billion-dollar margin:

Amazon Doesn't Believe in Research and Development Spending

To arrive at that $22.6 billion in 2017 R&D spending by Amazon, I had to assume (as pretty much everyone tracking corporate R&D spending does) that the line item for "technology and content" on Amazon's earnings statement is equivalent to R&D. Is it? Here's how Amazon described technology and content in its most recent annual report:

Technology and content costs include payroll and related expenses for employees involved in the research and development of new and existing products and services, development, design, and maintenance of our websites, curation and display of products and services made available on our websites, and infrastructure costs. Infrastructure costs include servers, networking equipment, and data center related depreciation, rent, utilities, and other expenses necessary to support AWS [Amazon Web Services], as well as these and other efforts. Collectively, these costs reflect the investments we make in order to offer a wide variety of products and services to our customers.

So it's ... mostly R&D? Maybe? The Statement of Financial Accounting Standards No. 2, adopted way back in 1974 by the Financial Accounting Standards Board (the chief arbiter of the generally accepted accounting principles that corporations are supposed to adhere to in their annual and quarterly reports) and now incorporated into Accounting Standards Codification 730, defines R&D as activities aimed at developing a new product, service, process or technique, or making a "significant improvement" to an existing one, then adds:

It does not include routine or periodic alterations to existing products, production lines, manufacturing processes, and other on-going operations even though those alterations may represent improvements and it does not include market research or market testing activities.

Amazon's "technology and content" clearly mixes the significant and the routine. So in a letter dated Sept. 27 that included several questions about Amazon's 2016 annual report, William H. Thompson, the accounting branch chief of the SEC's Office of Consumer Products, made this request:

Please tell us and disclose in future filings the total amount of research and development costs charged to expense for each year presented in the consolidated statements of operations as required by ASC 730-10-50.

It's a little surprising that it took the SEC so long to ask for this, given that Amazon has been folding its R&D spending into "technology and content" since its 1999 annual report (before that, it had an earnings-statement line item for "product development"). Then again, it is only in the past few years that Amazon's spending has ranked it among the potential global leaders in R&D. The company's technology and content spending has doubled since mid-2015, quintupled since 2012 and is up tenfold since 2011.

Amazon Doesn't Believe in Research and Development Spending

How did Amazon respond to the SEC's polite request? In an Oct. 26 letter, vice president and worldwide controller Shelley L. Reynolds did offer up some edits to flesh out the company's description of technology and content, which are reflected in the above annual report excerpt. But she also made the case for why Amazon did not break out R&D spending:

Our business model encourages the simultaneous research, design, development, and maintenance of both new and existing products and services. For example, our teams are constantly working to build new Alexa skills and simultaneously maintain current skills, and these activities are within a continuum of those described in ASC 730-10-55-1 and 2 and are not easily distinguishable operationally.

Accounting Standards Codification 730-10-55-1 lists activities that "typically would be considered" R&D, while ASC 730-10-55-2 lists those that typically wouldn't be. Amazon's argument was that its people tend to work simultaneously on things on both lists, so it couldn't separate the two. This did not initially convince the SEC's Thompson, who responded on Nov. 24:

As previously requested, please disclose the total amount of research and development costs charged to expense for each year presented in the consolidated statements of operations as required by ASC 730-10-50-1.

To which Amazon's Reynolds replied on Dec. 21 (and yes, I know it's a big chunk of text, but it is quite revealing):

Because of our relentless focus on innovation and customer obsession, we do not manage our business by separating activities of the type that under ASC 730-10-55-1 are “typically … considered” research and development from our other activities that are directed at ongoing innovation and enhancements to our innovations. Instead, we manage the total investment in our employees and infrastructure across all our product and service offerings, rather than viewing it as related to a particular product or service; we view and manage these costs collectively as investments being made on behalf of our customers in order to improve the customer experience. We believe this approach to managing our business is different from the concept of planned and focused projects with specific objectives that was contemplated when the accounting standards for research and development were developed under FAS 2. Given the significant breadth of projects and improvements that we have underway, our employees routinely work concurrently on multiple projects, including projects that could be defined as research and development in nature and also more routine, ongoing activities to refine, enrich, or otherwise improve or adapt our existing products and services. Similarly, our activities may focus on developing new products and services, but these activities often result, in whole or in part, in enhancements to existing products and services.

The SEC's Jan. 22 response, this time from Mara L. Ransom, assistant director of the Office of Consumer Products, was:

If you are unable to identify or estimate research and development costs, please explain in detail the reasons for your inability.

Amazon's Reynolds, writing on Feb. 16, reiterated the arguments in her previous letter, then added:

In reassessing this conclusion in response to your comment, we discussed with our investor relations department whether quantifying traditional research and development costs within the scope of ASC 730-10-55-1 would be meaningful or useful for investors, and we also reviewed our earnings call transcripts for the past three years to see whether investors ask questions about research and development costs within the scope of ASC 730-10-55-1. These discussions and review reaffirmed that such costs are not material for two reasons. First, as discussed above, because of the range of innovation activities we undertake to support the hundreds of millions of different products and services that we offer, aggregate research and development cost data would not indicate any particular area of activity and would not reveal trends with regard to development efforts that are material to an understanding of our business. Second, we believe that distinguishing between costs attributable to activities of the type described in ASC 730-10-55-1 and those attributable to activities described in ASC 730-10-55-2 would be confusing and misleading to investors, as the resulting disclosures would not fairly present the investments we make in order to offer a wide variety of products and services to our customers.

With that, the SEC appears to have given up. From Thompson, on March 9:

We have completed our review of your filing. We remind you that the company and its management are responsible for the accuracy and adequacy of their disclosures, notwithstanding any review, comments, action or absence of action by the staff.

Now, as somebody who likes to compile charts of corporate R&D spending, I am of course disappointed by this. Amazon's gigantic technology and content spending numbers will continue to be not quite comparable to other companies' smaller R&D numbers. I also have some sympathy, though, for Amazon's insistence that it doesn't want to have to think of R&D as something separate from its everyday activities. The determination has no direct impact on the company's bottom line, but it may have an impact on the company's culture.

There's a famous-within-certain-circles 2011 rant by Steve Yegge, an engineer at Google who had previously worked at Amazon. After declaring near the beginning of it that "Amazon does everything wrong, and Google does everything right," he proceeded to explain why Amazon -- despite its "dirt-smeared cube farms," poor hiring processes, less-than-great pay and benefits, messed-up operations, and lack of engineering standards -- was better than Google at coming up with new things that customers might value. His explanation for this involved a lot of details about software architecture that I don't fully understand, but the gist seemed to be that by the time Yegge left in 2005, "Amazon had transformed culturally into a company that thinks about everything in a services-first fashion." Maybe it's a service for internal use, maybe it will end up being sold to outsiders, maybe it will create a new product category, maybe it will incrementally improve an existing offering. The approach is the same in any case. As Yegge rephrased it near the end of his essay, the idea is to "Start with a Platform, and Then Use it for Everything." Which Google, in his view, did not do.

It is testimony to what a great place Google must be that not only was Yegge not fired after his critique went public, but he also stuck around for almost seven more years. It is testimony to the challenges faced by Google and its parent, Alphabet Inc., that when Yegge departed in January for a job at Singapore-based ride-hailing company Grab, he announced:

The main reason I left Google is that they can no longer innovate. They’ve pretty much lost that ability.

Amazon still seems able to innovate. Unlike Google, which, after developing a remarkable profit engine out of selling targeted ads against Web searches, has been unable to come up with anything else even remotely comparable, Amazon started out as a low-margin online bookseller and built several more lucrative businesses on top of that (although none is anywhere near as profitable yet as Google's ad business). Its approach to technology and content investment may make more sense for a company in today's software-is-eating-the-world environment than one that treats research and development as functions separate from delivering products and services to customers. It pains me to say it, but other tech companies may want to consider following its example.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Justin Fox is a Bloomberg View columnist. He was the editorial director of Harvard Business Review and wrote for Time, Fortune and American Banker. He is the author of “The Myth of the Rational Market.”

  1. On the SEC's Edgar website it's not at all clear when the letters were made public, but the Bloomberg terminal's company-filings function lists them as all having been uploaded on April My Bloomberg Gadfly colleague Shira Ovide alerted me to their existence.

  2. The FASB's statements of financial accounting standards are public, but in order to pay the bills the board keeps the codification of how they all fit together  (ASC et al.) behind a registration wall and, if you want more-than-bare-bones functionality, a annual paywall.

To contact the author of this story: Justin Fox at

To contact the editor responsible for this story: Brooke Sample at

For more columns from Bloomberg View, visit

©2018 Bloomberg L.P.

Get Regular Updates