Understanding Amazon’s “HQ2” announcement

Late last week, Amazon announced that it intends to set up a second headquarters campus in another city in North America, where it would employ up to 40,000 people over time.  Since then, local politicians and pundits have sounded off on what this means for Seattle.

I used to work in business and operations for a division of Microsoft; I was involved in the strategy, selection and setup of several new sites, and I worked with people across the company who did similar work. Based on that knowledge, here’s my take on what Amazon is doing, why they’re doing it, and the most likely candidate cities for their new campus.

Amazon’s business challenge is, in one sense, very simple: they are trying to sustain a 20-25% annual increase in sales. That is a huge growth rate, and they need to make a large number of investments to try to grow that much. To wit, here’s a quote from Jeff Bezos in Amazon’s last quarterly earnings report:

“Our teams remain heads-down and focused on customers,” said Jeff Bezos, Amazon founder and CEO. “In the last few months, we launched Echo Show (our newest Echo device with a video screen), introduced calling and messaging via Alexa on all Echo devices, debuted Inside Edge on Prime Video (the first of 18 Indian Original Series), introduced Amazon Channels in both the U.K. and Germany, launched four new Fire tablets, expanded Amazon Fresh to Germany, launched Prime Now in Singapore, launched our 25^th airplane with Prime Air, hired more than 30,000 new employees, opened three new Amazon Books stores, launched more than 400 significant AWS features and services, migrated more than 7,000 databases using AWS Database Migration Service, and held our third annual Prime Day — signing up more Prime members than ever before. It’s energizing to invent on behalf of customers, and we continue to see many high-quality opportunities to invest.”

Since then, they have closed the acquisition of Whole Foods.

The compounding sales numbers mean that sustaining the growth rate just gets harder and harder to achieve. In 2013, Amazon’s 22% growth meant increasing sales by $13.3 billion. This year, they will need to increase sales by $30 billion in order to have that same 22% growth. Rite Aid’s annual sales are about $30 billion — that’s how much Amazon needs to grow. The Whole Foods acquisition will cover about half of that; they will need to grow another Whole Foods over the course of the year.

Amazon’s success in meeting their growth target and on all of these opportunities — and the ones that they have yet to announce — is entirely dependent upon their ability to hire people to work on them. Hiring is their single biggest business challenge, just like every other technology company, and the competition for talent is fierce. So long as their growth is based on new opportunities, their headquarters hiring will continue to grow in proportion to their sales growth. And it’s exponential; they’re speeding up their growth.

The company’s announcement that it is building a second “headquarters” campus in another city is an admission that it can’t hire as fast as it needs to here in Seattle. Part of that is that the company is hitting the inevitable physical limits of how many people it can add to its workforce here. That’s a reflection of the city’s limits: housing, transportation, and other services. And it’s a reflection of the local pool of workers that it can draw from. But it’s equally because a company can only “scale up” locally for so long. Amazon employs about 40,000 people in Seattle; if it has an annual attrition rate of 10% — which might be a low estimate — then the company needs to hire 4,000 people every year just to sustain their existing workforce, before they even get to hire people to grow new businesses.

But there’s a second, equally important admission: there are people they really want to hire who will never agree to move to Seattle. Seattle is dark, cold and rainy for a big chunk of the year. It’s not a very racially diverse city. And perhaps most important: many people want to live and work close to family members who live elsewhere in the country. Improving Seattle’s housing, transportation and infrastructure can help a little bit with racial diversity, but can’t address the other two issues.

It’s actually quite amazing that Amazon has kept so much of its headquarters and R&D staff in Seattle for this long. But it was inevitable that the company would hit the point where the only way it can meet its growth target is to “scale out” instead of “scale up.” Amazon has no choice but to spread its headquarters and R&D staff to another site. Yes, Seattle should feel a little bad about the ways in which it constrains Amazon’s growth, but that’s just a small part of what’s happening here, and it’s probably not the most significant part. Amazon’s internal constraints are equally to blame. Besides, Amazon will continue to grow here, because it needs to hire like crazy at both sites in order to meet its exponential growth target. If it can’t meet its target just by hiring in Seattle, it certainly can’t meet it just by hiring somewhere else.

Where will Amazon choose to put its new campus? Officially, they are putting out an RFP, encouraging local municipalities to put together packages to entice them and to compete against each other to offer the best deal for the company. Privately, they almost definitely have a short list of places they would like to be. Here’s what’s important to them:

• A place where a well-educated workforce wants to live. Good schools, good food, arts and cultural activities, good parks. Single people can find partners; married people can raise a family there. Well-educated spouses can also find jobs (don’t underestimate the importance of solving “two-body problems” in hiring talent). Amazon noted in its RFP that it is looking for a metropolitan area of at least 1 million people.
• Cheap housing nearby and good transportation infrastructure.
• A hub airport. This is far more important than it seems. Amazon is a tech company, a retail company, and a logistics company. It has sites and partners all over the country and the world. Its headquarters and R&D staff travel a lot, and they need to be able to travel on short notice.  Air travel capacity, frequency and convenience is critically important to managing their business. There is no way Amazon will site its new campus in a city that doesn’t have direct flights to most major North American cities.
• A pool of talent to hire locally. Amazon’s headquarters staff primarily consists of technical and business (marketing, sales, logistics, finance, legal) staff. They benefitted in Seattle from having Microsoft and other tech companies whose tech employees they could hire away, and many other large businesses (Starbucks, Nordstrom, etc.) that they could steal from for their business ranks. Equally important are the local universities, particularly UW with its top-ten computer science program and its strong business schools. UW gives them a steady stream of entry-level hires, as well as interns, consultants, and spinouts.
• Office space. Amazon specified in its RFP that it initially wants about 500,000 square feet of office space with the ability to expand up to 8 million square feet over time. That could take several different forms: it could be a new, undeveloped site outside of an urban area, or an existing office building in an urban core with adjacent space that could be built on or redeveloped (as the company did in South Lake Union). The latter is more expensive, but the former would provide better housing and transportation options.
• Economic incentives. That could be a sweet deal on the campus, or tax incentives. Given the short timetable for Amazon’s RFP, it will be challenging to arrange tax incentives — which in many cases are controlled by state governments rather than local ones. While everyone likes to focus on this as “corporate welfare,” it’s probably pretty low on Amazon’s priority list.
• A business-friendly environment. Probably the most important aspect of this is quick turnaround on approvals for construction, and a desire to invest in infrastructure. Amazon’s headquarters is almost entirely people in offices; it’s not about sales taxes, unions, industrial pollution, or moving freight. Amazon has those issues at their fulfillment centers and other sites, but not at headquarters. They care about building new office space quickly when they need it, and getting people to and from the campus. Sure, they care about B&O and head taxes, impact fees, and all that, but those would have to be extreme outliers to affect Amazon’s decision.

So which cities are in the best position? Here’s my list:

1. Atlanta. It checks nearly all the boxes. They don’t have a lot of other tech companies locally, but there is a strong core of businesses downtown as well as lots of property prime for redevelopment. It has a good mass transportation system and a huge hub airport with flights to everywhere (and plenty between there and Seattle). Georgia Tech is a top computer science school, and Emory University has a strong business schools. And Atlanta’s housing is still cheap. Atlanta is also a much more racially diverse city than Seattle, which would help the company deal with its serious diversity issues.
2. Boston. In the 1970’s and 1980’s, Boston was a tech hub on par with Silicon Valley; today it’s a distant third behind the Valley and Seattle, but with strong aspirations. There continues to be a large pool of tech talent there. Also, MIT is one of the “big four” computer science schools (along with Stanford, Berkeley and Carnegie-Mellon), and both MIT and Harvard have cream-of-the-crop business schools. Boston has one of the best transportation systems in the country, and a decent airport. But it’s not particularly business-friendly, and housing is extremely expensive.
3. Austin, Texas. Like Boston, Austin is a mix of good and bad. It’s a great city with a strong commercial core and fantastic arts and cultural scene. Dell, IBM, and other tech companies have or had major presences there for years. UT Austin has a top CS program. It’s also the corporate headquarters for Whole Foods, so Amazon already has an office presence there. But its transportation infrastructure is terrible, and its airport isn’t up to Amazon’s needs.
4. Toronto. It’s the commercial capital of Canada, and with the country’s more agreeable immigration policies it would allow Amazon to hire people to the site without all the hassles of the U.S. government’s visa programs (and the open hostility of the Trump administration). University of Toronto is a strong school nearby (and University of Waterloo is about an hour away). Decent transportation infrastructure, and a major airport. Lousy weather, and not much of a tech talent pool to hire from. But splitting its two headquarters campuses across national borders could end up being a huge and expensive impediment to getting work done.

There’s one more factor that Amazon hasn’t addressed, which will have a significant impact on the plan for the second campus: how it plans to split its businesses across the two sites. It says that it wants two “co-equal” headquarters sites — though I can’t think of a single company that has ever made that arrangement work. But that aside, would each business split management, business and R&D staff across the two sites, or would each of the businesses choose one campus as its primary site?

There is a large body of research that shows that geographically dispersed teams tend to be much less productive and efficient than single-sited teams. Spreading teams across time zones is one impediment since they have fewer hours in the day where they can communicate instantly to resolve issues. But some of those studies have shown that even splitting a team between two adjacent buildings can have a large effect.

If Amazon decides to split staff for each business across the two sites, it would do well to pick a site in a time zone closer to Seattle. But if each business (or unit) keeps it staff together on one of the two sites, then the distance between campuses is less of an issue. It’s unclear, though, whether Amazon’s internal culture would allow enough autonomy from the executive offices to successfully run parts of the business from afar.

What should Seattleites take away from this? Actually, not very much. It’s a positive sign for Amazon that they are bullish on their own growth for the next several years; that speaks highly for the company’s prospects. This move was inevitable with rate of the company’s growth, and it’s only a small reflection of Seattle’s growing pains. Amazon will continue to grow its local headquarters workforce even as it scales out to another location, though depending on how it organizes itself across the two sites there may be some shuffling of people. To that end, it’s still critically important that we continue to invest in housing, transportation and infrastructure in Seattle.