An Inside perspective on Amazon's purchase of Kiva Systems (or Domo Arigato, Mr. Roboto)
I've been inundated in the last 24 hours with questions about Amazon and their seemingly irrational ~$775m acquisition of Kiva systems.
- Why would Amazon buy a robot company?
- Why would they pay so much?!
- What does Kiva do?
- Why have we never heard of this company?
- Will Amazon allow Kiva to continue licensing this technology to other retailers or keep it as it's own proprietary strategic advantage?
- Robots?! Robots?! Is this some weird Bezos space thing? How many robots do you get for $775m!? (Ok just kidding on that one)
These are all interesting questions and ones that I think are interesting. But before we jump into those, step with me into our blogging time machine back to 2002. The internet bubble hard burst, we were in a severe tech recession and Nelly was topping the charts with "It's hot in here".
I'm an alumni of NC State University (NCSU) (GO PACK!!!!) and actively involved a bunch of programs there. Around late 2002, I was approached by one of the professors I know there, Pete Wurman. He was starting a new business in the e-commerce space that would 'revolutionize e-commerce' and wanted to get my thoughts on the business.
I sat through a short pitch from Pete and his co-founder, Mick Mountz. It went something like this:
The Problem: Traditional e-commerce warehouses are technological dinosaurs - they rely on fixed shelving, bad placement algorithms and (lots of) error-prone humans. The result is an expensive operation with at best 95-98% operational excellence. Also they tend to scale linearly - the more you want to ship, the more you need.
The solution: What if built Warehouse 3.0 - a warehouse that took advantage of all the technological advances that have come around - advanced algorithms, robotics and sophisticated software to tie it all together. Mick+Pete had modeled out that they could offer a triple win to e-commerce retailers:
- Win 1 - Savings - Your average warehouse needs a good 20-40 people working a single shift to process a fair volume, and that doubles if you want to have high service rates because you add a shift and have to cover for sickness. Robots would cut that down by 60-80%. Robots don't need shifts. They don't get paid by the hour. They don't get sick. They don't need vacation time. They don't sleep, they won't stop ever! (sorry watched Terminator again recently). Imagine going from 40 people running your operation to 8.
- Win 2 - Increased throughput and capacity - Once you have robots running your warehouse, a magical thing happens. You have more room for product. In fact the way Kiva works, the robots self-organize and optimize the product shelves in a more optimal way than you can in human-based fixed warehouses. The result? More capacity for storage, and more throughput of orders shipped. (why did I count this as only one win?).
- Win 3 - Decreased errors - This is a huge one. The Kiva software would track every order and keep a very close eye, checking, double checking and even triple-checking error-prone humans throughout the process. Recently, Mick was walking me through a demo of their latest software and I tried a couple of ways to introduce an error and darned if the system didn't catch me every time. The electric shock made sure I didn't do it again (kidding).
At the time, we were a couple of years into ChannelAdvisor and I had mixed emotions. The idea and value prop seemed to be really amazing. For folks not in the e-commerce business, it would sound almost too good to be true. My concern with the business plan was that it seemed very hard to implement:
- Build the robot, ERP software and algorithms
- Sign up a big retailer
- Convert their existing operation to a new robotic operation
- Make sure it all worked and delivered on the three wins
- Go back to step 2
I also told them that if this worked, the biggest beneficiary would be Amazon and if they could get them as a client, well at $5m+/warehouse, that would be a huge win.
Conclusion: Great idea, hard to implement, but if you can.... it could be huge. I think Mick+Pete got a lot more push back from early investors and advisers on the whole 'futuristic robot' thing, but I had asked them a bunch of questions about that and they really had that part nailed, so I felt it was more of an execution challenge on the business side. The key question was: Would a retailer turn over their warehouse to a new, unproven technology?
Check this video:
(Hey, why do they always play classical music when robots are zipping around?)
Over the years I have kept up with Mick+Pete and it's been awesome to watch them executed against their plan. That first retailer I thought would be hard, they nailed down and had up within the first 18 months, then word got out quickly that this system was pretty much a silver bullet for any and all warehouse challenges because of the triple wins mentioned above. Over the years they added blue chip retailer after retailer including: Quidsi (Diapers.com and family), Zappos, Crate and Barrel, Staples, Gap, Saks, Dillards, TRU and more.
The company eventually got some great VCs on board (Bain, etc.) and really scaled up all parts of the operation. Another thing that was cool to see was that they both stuck to the original plan they pitched back in '02, but also they kept adding really smart enhancements and efficiencies. The original robots had to recharge pretty frequently, so they'd add more batteries. They improved the storage and routing algorithms. They constantly improved the entire system. Because a lot of it is software, they could deploy this to existing installations.
Imagine you are warehouse manager and you can process 1000 packages a day. Kiva updates your software, maybe adds some robots and tweaks and now you can process 1200. Historically you would have to either shut down your existing operation for months to change a process or what most retailers do is build a 2.0 in a new facility, keep the old one running and then shut down the old one eventually (or keep both if they need the capacity).
The last article I read on Kiva suggested that they had revenues over $100m and were growing at 80-100% y/y - none of that surprises me and I'd guess they would be at a ~$200m run-rate by the end of 2012 given the demand I've been hearing about for their systems.
The bottom line: Having a technology advantage in your warehouse gives you an advantage over your competitors both on the cost and error side, but also on the deliver/SLA side. As we head to a world with free one day shipping, you're going to need an army of robots to be able to do that efficiently.
Kiva+Amazon = ?
One question I've been getting is "why did Amazon buy Kiva?" Remember, via Quidsi and Zappos, Amazon is already a Kiva customer.
I don't have any insight into why Amazon ended up buying Kiva and why today, but I day dream it playing out like this skit:
Amazon super-star DC operation manager: I heard you guys had a pretty efficient warehouse - we have been building and operating warehouses for 10yrs and we think we've got about every bit of juice squeezed out. Let me see your numbers.
Zappos super-star DC guy: Do you guys use robots? We do... Here are our numbers.
Amazon super DC guy: (long pause)....... This can't be right, you must have a different way of measuring everything. These numbers are more than double ours.
Quidsi super-star DC guy: weird, we have the same numbers as the zappos guys, but we are on version 4.2 of Kiva so ours are a bit better.
Amazon super DC guy: Ok, ok, but your labor has got to be twice ours or more, you guys running four shifts?
Zappos and Quidsi guys: Well, if you look at our cost/order it's X and our number of employees are actually 20% of yours.
Amazon super DC guy: (sheepishly) ummmm so tell me more about this Kiva robotic system again...
<10 days later>
Amazon super DC guy: (on phone with Kiva) Yes, how much would it cost to deploy this system in 50 domestic DCs and say 20-30 internationally? Ok, $5m/warehouse, ok.
Amazon super-DC guy: Mr. Bezos. You know every year we've been able to get 10% improvement on our DC metrics. Well, I figured out how we can double the productivity of our warehouses and significantly reduce our costs, but it's going to cost us $600m. I know that's a big number sir, but what if we don't have to build 20 more warehouses this year because of it? My calculations have the payback on this as less than 18 months.
Bezos: (after picking apart the numbers, touring Zappos/Quidsi and falling in love with some orange 'bots) Instead of licensing this, we should just buy the whole dang company, do you realize what a huge strategic advantage this would give us over everyone? Plus we can make our customers happier with fewer error rates and even deliver products faster than we do today. Think of it - one day delivery around the country powered by robots at a cost that is less than what our competitors pay for 3 day delivery!
(Insert Bezos laugh)
Now for those questions:
Q: Why would Amazon buy a robot company?
A: Because it's smart. For the price that's say double what it would cost to roll out, they own the entire thing and have given themselves a significant edge over the competition.
Q: Why would they pay so much?
A: Well, I imagine this actually is cheaper than the alternative of building capacity through more warehouses. Would you rather them spend $775m on the current warehouse 1.0 model or spend $775m on Kiva, roll it out to their warehouses, cut costs, double productivity and increase service levels? Can you say one day Prime?
Q: Why have we never heard of this company?
A: Tablets are sexy, robots are sexy, square orange robots toiling away all night in 200,000 sq-ft warehouses while the AC is turned off is not sexy. If you have 50 DCs and can find a way to cust costs, remove errors and improve throughput - that's AWESOME!
Q: Will Amazon allow Kiva to continue to license the technology to other retailers.
A: Sure. Amazon is the most open platform company you will find. Look at FBA, AWS, etc. Perhaps those conversations will include some discussions around selling on Amazon. Will they license the technology to GSI/eBay or Walmart? Hmmmm, I seriously doubt it.
Q: Why now?
A: Your guess is as good as mine. Maybe Amazon finally worked through the math on this. Maybe they did a test install and it went well. Maybe someone else was going to buy Kiva. Maybe Kiva was on the verge of going public.
Hopefully you've enjoyed this semi-insider's view of the transaction and Kiva. I've been meaning to blog about Kiva for sometime as the "hottest company in ecomm you never heard of", and this transaction gave me a great opportunity to do it.
I wanted to give a truly heartfelt shout out to Mick, Pete and the team at Kiva. They have built a great business using what can only be described as a science-fiction sounding concept. Ideas are easy and execution is really really hard and to their credit these guys executed on this concept flawlessly and deserve all of the rewards from this great exit for Kiva. You guys have really lived the entrepreneurial dream with this company!!
Great job guys - I hope you are doing the Robo-boogie today!
P.S. Do robots get stock options?
SeekingAlpha Disclosure - I am long Amazon and Google. eBay is an investor in ChannelAdvisor where I am CEO.