Amazon Plans Carefully Its Distribution Capacity Growth

It’s that time of year again, when shoppers hit the stores on Black Friday and go online on Cyber Monday, the biggest day of the year for Internet sales. Experts predict that Internet sales will increase 12 percent this year; Amazon’s sales this holiday season will grow much faster.

At the CSCMP Annual Conference in October, Mike Roth, Amazon’s VP of North American Fulfillment Services, gave a speech titled “Supply Chain Technology Powering Earth’s Biggest Selection.” Mike spoke about many items related to Amazon’s strategy, but what interested me the most was the careful planning the company needs to do to support its tremendous growth.

Because Amazon opens so many warehouses — 40 in North America — the company believes in rapid build cycles, which allows it to build capacity as the company needs it. Amazon is challenged by ever growing sales, year-over-year growth of 35 percent, and hockey stick seasonal demand, which is 3X greater than non-seasonal demand. Amazon’s CFO has made it clear that he does not want to pay for unused capacity. Mike showed a photograph of one site that had just broken ground, and another photograph of the completed warehouse five months later. And this was not a small facility — it was 1.2 million square feet with 40 foot high ceilings and four story pick modules complete with automation.

Since a highly-automated warehouse would take months longer to complete, Amazon uses warehouse automation sparingly to preserve scalability. The company  has opened four distribution centers this year with many more scheduled to go live in the coming months.

Amazon also improves its capacity utilization by carefully tracking DC capacity usage and by including forecasts of the capacity it will need in its S&OP process. Amazon uses population density maps overlaid with its DC locations. Mike explained that demand growth is largely located in high-density population areas. Finally, because Amazon is not building facilities to support peak demand, the company uses drop shipping and works with 3PLs to supplement its internal efforts.

Just as important as being able to scale with quick build cycles is the need for flexibility. Mike pointed to the exploding number of product categories Amazon covers today versus years ago. If the company had built distribution centers (DCs) just for books, it would have been in trouble as e-readers have eroded sales in that category. Meanwhile, highly automated systems designed for books would not support newer categories like pet supplies or appliances.

Mike was asked about the use of Kiva robots. He said they are still trying to figure out how to implement them in their warehouses. I’m not sure I fully believe that. It strikes me as unlikely that Amazon would spend $675 million to buy a company that builds flexible and scalable automation systems without having a pretty good idea of what that technology brings to the table.

But perhaps Amazon will conduct some experiments. There are a couple of ways the company can go. It can continue to build DCs with their existing layout – big DCs with four mezzanine levels — and use the Kiva bots on each mezzanine level. This approach would not affect the speed new DCs are built.

Or Amazon can build smaller warehouses, with ceilings as low as 10 feet high. Obviously, these warehouses would not have mezzanine levels. This option would greatly speed up the company’s ability to build new warehouses to meet new demand for distribution capacity. And it would make the job of purchasing suitable sites much easier too.

But small footprint warehouses would also complicate Amazon’s decision of where to hold inventory.  With its existing large format DCs, if a customer orders multiple stock keeping units (SKUs), there is a good chance they can be shipped together on the same truck.

With smaller warehouses, that would happen less often. DC efficiencies are already improved by analyzing the likelihood of one item being simultaneously ordered in conjunction with another item, and then placing those items near each other. This form of analysis becomes more important, with the difference being that co-ordered SKUs will need to be placed in the same warehouse, but would no longer need to be placed near each other inside that warehouse.

It will be interesting to see which way Amazon goes. I’m betting we will see a mix of small and large warehouses.


  1. Amazon has been, and continues to be, a very interesting animal to follow. Over the years we have all heard so much about Walmart’s excellence in supply chain, but they are nowhere near as creative and leading edge as Amazon in my opinion.

    The Kiva acquisition, coupled with the DC build-out portrayed in your piece are simply parts of the puzzle. Add to that where Amazon has gone with the Kindle, the IT server farms they have created and marketed, the 3rd parties who leverage Amazon’s IT and distribution capabilities, their developments in Amazon Lockers and same day delivery, and who knows what is next. Some of us believe that Amazon may well develop it’s own delivery model and move away from UPS, FedEx and On-Trac — A model unlike anything we see today, perhaps being total “green”.

    I agree with your comment about how they are not yet sure how they are going to use Kiva. I think they have taken a note from Apple’s playbook on how to keep plans secret, and how to come up with new visions outside of the box.