Does this propose these retailers are exclusive for 2-3 years. Who will make the calls on the partners. Both questions are valid whenever we proceed From: Steven Sinofsky Sent: Sunday, November 4, 2012 10:42 PM To: Kevin Turner; Steve Ballmer Subject: RE: Surface sales physics after a week Let's take a step back. The intent was not to surprise, but to act on where we are. I think we're all surprised. I'm surprised at where we are in volumes, and honestly I don't think anyone feels we have "plans" to broaden distribution in January. So let's just assume we all have a bit of surprise and see what we can do to manage what I think is a drastic situation we find ourselves. There's nothing intended other than to get us to a better spot and certainly this is not about the efforts going on everywhere to amplify what is going on in our own stores—we've all spent considerable energy personally and the efforts of those working in the stores are historic. We can disagree on how Surface will be labeled. Here's the data I am working with. Wall Street analysts have come up with estimates as low as 1 million units and on average in 3.5 million units for 2012. These are not made up. It is clear these are based on sourcing from component providers (likely nvidia but others as well). In other words, it will be pure "spin" to say we are hitting our own targets because in fact we have made about 1.2-1.5 million units this year (the 1.2 number is based on delivery before holiday, so add one more week of production). In addition, we are up against the confusion that got put in place over whether we are committed to Surface or it is a boutique product designed to "push OEMs". Since we will certainly not announce a number, it is likely everyone will conclude whatever we did (and whatever we say) are not meeting even our own expectations. We will be in a spiral very quickly. I've personally talked to several of the analysts who write the reports you send around and I am certain they will look at a lack of a number as a negative and if estimates start to surface (pun intended) that approach reality (250K) we will be in real trouble—that is 1/4th the lowest estimate and not the consensus estimate of 3M. The whole project will be labeled "Zune" and this will be a broad failure. The impact this has on sales in 2013 will be significant just as we saw with Zune—the partners won't carry it, the margins will erode, and pretty soon we'll be in RIM Playbook territory. We need to keep in mind how the Apple favored press will amplify these low numbers and in addition how the OEMs will use this as well. I really struggle with how we can make it through this. The story is out there written just waiting to be told. Of course we are just comparing our own views of how this can play out in the landscape. If we want to be optimistic we can be. But I don't know how 250K units will meet any views we currently have. But for the purposes of this we can assume the most optimistic scenario. That's because there are real financials and sales velocity we can rely on rather than just predictions about the press. There are no doubts about the velocity we are selling at—the sales per square foot. We need to have 4x the current sell-thru to be at a steady state. On current path we are going to be sitting on 1M Units of inventory. It is unclear under what plan we can ever burn through these. We can say we planned wrong, but we knew how many we could make and, frankly, we just went in optimistic over how much we could move online. We are currently moving about 28% of units as online. We need to be doing 70% online. But at the same time we more than double our online units we need to 5 EFTA_R1_01741988 EFTA02572392

increase our total units. It means our online units need to go up by about 8x. It does not seem like that has any chance of happening. The actions proposed are all excellent. I don't think one could disagree with them. But math says it won't be enough: 1. Take Surface NOW to other retailers in China. Gome, Mall Stores, etc. {Suning has a much larger footprint already so this won't double the runrate} 2. Extend our Holiday stores NOW for next calendar year. {The math on this will be positive, but this all is next calendar year and will not change the velocity we are seeing so this doesn't add to the unit runrate} 3. Start on a plan now to properly take Surface to mass retail in Jan/Feb with the Pro launch,. (We agree we have to plan now. I am suggesting that as we plan this we will find an appetite and capability to do more sooner} 4. Strategize how we tell our OEM's on our plan to take Surface to mass Retail, then we must properly tell the OEM's. {We have to do this no matter what if we follow your recommendation 3.} 5. Accelerate our physical Stores in the US and open physical and pop up stores in the Canada, UK, Australia and China in 2013. (This is a small marginal increase in volume} We're missing what we would do in Europe, Asia, Pacific. Online and experience centers are not working to move units. With the fulfillment challenges we have had it is not clear we can recover from that—there's a good chance we have a significant SAT issue that will remain for quite some time. Your issues raised are all real. Let me try to talk about them a bit in terms of the categories of issues: * OEMs. We are in a tough spot as we operationalize things internally in that when there are differing views one or more folks hold up the possibility of upsetting OEMs. But we're not consistent in what we do or how we do things, so this isn't really something we can model or operationalize. From the outset we've been in a tough spot. When we priced Surface at $499 we sided with "our own" interests. When we included Office we upset the x86 side of things. And so on. I think we can agree, that selling now or later will have the same impact on the dialog. One could argue that OEMs are stocked out for holidays and that there are no tablets or touch screens at volume so we're really not taking "money" as you put it. I'm just pointing out that this is not so cut and dry. • Retail execution. I've attached a very significant drill down on how this could work. The team that executing on Suning has put this together. It is essentially the deck we would go to Retailers with. It has margin details and an overall approach. We can doubt the ability of the team to execute, but they believe that they can deliver. We will never execute this as well as our stores. We have all done shopping this past week (I went to Northgate and Southcenter) so I'm assuming we all had the same experience of 8 out of 10 PCs "dead", no networking, and poorly trained retailers (I listened to a retailer completely fabricate a Windows 8 phone scenario). It looks really bad for Surface, but frankly unless our plan is store within a store I am not sure what we can ever hope to expect. Yes that makes me sad and is why we bet on our retail. * SKUS, Online inventory and merchandising. We made a call to price at $499. Given that we can't secure attach very well online we limited this to 10% of the runrate, which is driving the stock out. This was going to have the implications we expected. The color choice was a known tradeoff because we did not want to sell unbundled for margin, value prop, and competitive reasons. While we know enough now on color mix and attach, any change in this (like adding green) will make the complexity of this significant. We've seen the impact of stock out on the blue cover. 6 EFTA_R1_01741989 EFTA02572393

We know the demand for coverless 64GB as well. Our margin structure is based on 110% attach. Less than that is problematic. All of this was known in the move to $499 coverless. The retail model we have in place focuses on attach, but is a risky margin proposition and stock out of colors can be a real issue. • Enterprise. I'm excited to open this up. It is tough to see this as a material impact to the overall model. • Product quality. I think this is a red herring. We've seen nothing at all to indicate product quality issues. Even twitter is silent on quality issues. The tone of product reviews started off positive and is increasing. The 8% number doesn't match the data we have but even then is close to the modeled number at 6%. Regardless, I don't know what we would do other than merchandising and scripts to change this since the product is what it is. Supply chain. I think we've worked this out and know how to scale. • USA retail. I will defer. In many ways this is the least of our challenges given our footprint in the US. I have a hard time thinking WMT would stock us effectively given the low priced tablets they sell currently. I'm not trying to be flip and not pay due to these issues. What I want to say is that all of them are being worked on. I think we have three courses of action: • Plan starting now on how we expand for January. We can call this plan of record. I am not so sure it was. In any event let's do that now. It still involves folks getting rolling now. So we need to get them going. • With those dialogs happening we can see what we can execute in Europe, Aus now. This is where we have no "see and buy" presence and we can definitely get something going—at least folks think so. We should allow them to have the dialog. • Decide on a US plan. This seems a separate choice given the footprint we have here. There is a fourth action which is we essentially do nothing now. I don't think anyone is recommending that. Arguably there is an option where we just do way more marketing outreach to hope that is a substitute for physically touching the device for Holiday. What concerns me is that we're not having a dialog about the material issue. 1M units/$500M in inventory on 1/1 is a major problem and that grows by 250K units per week. There are only two ways to fix this. We sell more between now and 1/1 or we stop the manufacturing line. If we stop the line it is a crazy high cost since startup is high. This is just manufacturing management. Even slowing it down means letting workers go. If we don't sell more we're going to need a new building to hold all the inventory. This is the real issue. Everything else feels like tactical issues to address relative to either selling more or making less. I feel like under any circumstances we should unblock people from going to retailers and having the dialog about January. You indicate we need time to plan this and so we need to start. We can let them do this over the week and 7 EFTA_R1_01741990 EFTA02572394

based on what we hear we can decide to do more. Absent that, our only option is to reduce manufacturing. I don't see it as feasible to inventory 1M units. And then we'll see how we can manage Wall St expectations. Related to this is the notion of "Pro in January". I think we should temper expectations. This product is a niche product. It is $1000 and is not "the best tablet and the best laptop". It is a convertible with compromises. It runs x86 software but it does so at a size, weight, and battery life that will not please when compared to the world of 7" tablets. It is a small ultrabook (in 2 of 3 dimensions and weight). It is an ideal PC for a Windows 8 developer or for ruggedized industrial uses. It is not the ideal student PC (battery life) or knowledge worker PC (battery life, weight). It is incredibly nice. But it is not a blockbuster. Our chips are on Surface RT. I know everything about Surface has been thoughtful. This is just as thoughtful. I've attached the retail deck. This is real work that has been going on for months. The roadshow is what we present to retailers based on the Suning work. Slide 23 has the retailer margin. Slide 26 starts with the gives/gets. Slide 29 has the instore efforts on 12/1. Slide 39 is the RSP strategy. The GM looks like this: Open Max Suning 32gb Standalone 0% 0% 0% 32gb Bundle 1% 4% 4% 64gb Bundle* 8 EFTA_R1_01741991 EFTA02572395