The Apple Watch is out, and we’re seeing the first analyst estimates of some of its key costs. The variation in these early estimate is huge. It’s so big that, even if some estimates aren’t perfect, others are, to use a Tim Cook phrase, “in another universe”.

Here are two examples. These firms looked at the (1) hardware and (2) manufacturing costs of the Apple Watch.

• IHS’ estimate for the 38mm Sport version: $83.70.
• TechInsights’ estimate for the 42mm Sport version: $138.50.

Can both be correct? No. “But they looked at different-sized models” one might say. Nah; that’s almost irrelevant.

Apple Watch does not cost $84 … [it’s] Probably more than 2X that.

From my experience working with product and cost experts at a well-known mobile device company, I can tell you: Apple Watch does not cost $84 in hardware and manufacturing. It costs meaningfully more. Probably more than 2X that. And I’ll tell you why. Maybe I’ll even give you my estimate.

(By the way congrats to the TechInsights crew for having a reasonable estimate, in my view.)

---

First, it’s not for the reasons you see in the comments on the articles that re-publish these estimates. In those articles, you’ll typically see well-intentioned commenters say that one needs to account for research and development, sales and marketing, corporate income taxes, etc. None of that is accurate.

Let me put it in food terms. Let’s say we’re talking about your favorite pizza. Analyst firms are simply trying to estimate how much your pizza costs. Not the salary of the lab chef who formulated the new dough (R&D), not the funny TV commercial (sales and marketing) you chuckled at … just the costs directly related to building one typical pizza. Ingredients, labor, boxes, etc. The share for one pizza. Or in this case, one Apple Watch.

Just as new technology is a benefit multiplier for the consumer, new technology is a cost multiplier for the producer. (And if done right, it’s also a profit multiplier.)

(Those other expenses that people cite: they *are* important. If one knew them, they would provide holistic insight into what it really takes to make a product. But it’s very difficult for an analyst to estimate their contribution to a specific product. And, generally, fewer clients would pay for that work.)

Okay then, so what are the *real* reasons that some analyst cost estimates are off? Let me first tell you the most fundamental reasons:

A) They don’t build product. When you don’t build product, you don’t see the actual costs, the detailed costs, and you certainly don’t see them across the big picture –the entire product.

B) New products often include new technology. New technology often has lower yield rates, higher complexity, more custom materials and processes, higher logistics and licensing costs, and other variables. Most of this is completely invisible to an outsider. And this is why analysts usually omit some of these costs. But these factors impact component prices – which analysts do estimate – and the impact leads to significant error.

C) Estimating costs is hard, even for the companies that build the product. Building an integrated cost view and accurately accounting for dynamic part prices, tooling costs, and yield rates is difficult. It’s 10X harder for an outsider.

A supplier may have shown off a prototype, or a demo of item X, but if it hasn’t been manufactured to scale… it’s new

So let’s move on and discuss, more broadly, the types of costs that analyst firms sometimes underestimate, or purposefully omit (due to difficulty), or that people generally don’t consider.

And no, I don’t have inside information about Apple Watch. This is just me, with a beverage, typing.

1. COMPONENT COSTS ARE HIGHER THAN THEY APPEAR

Why? In a phrase: NEW TECHNOLOGY. Just as new technology is a benefit multiplier for the consumer, new technology is a cost multiplier for the producer. (And if done right, it’s also a profit multiplier.)

Prototypes:  Apple is a prolific prototyper. Not just in terms of industrial design. Think computer processors, mechanical systems (e.g., Digital Crown), accessory types, etc. Apple prototypes more, on more parts. While this happens during R&D work (which is excluded from product cost estimates), the material costs *are* mathematically peanut-buttered across the inventory of working, sellable watches. They’re part of the true cost of making one watch.

Very few people know [a part’s] cost, even at Apple. And if you’re outside Apple, you could easily be off by 2X.

Custom Components:  Apple Watch contains a number of unique sub-systems — S1 chip (“the computer”), Force Touch display, Taptic Engine, Heart Rate Sensor, Digital Crown. There’s either no good cost comparison (e.g., Digital Crown) or the implementation Apple uses is so meaningfully different that the price Vendor X paid is *not* the price Apple is paying. Three examples:

1.  The S1 system-in-package (the “computer”): This is a custom system, using at least one major custom part, an Apple processor. Four of the firms that examined it noted:

You could say Apple took the same approach to electronics excellence as you’d expect from NASA. (TechInsights)

Apple and/or their suppliers have designed and manufactured a 26 mm x 28 mm package that is very unique. (Chipworks)

The precision … is incredible…which means production costs are high. … [And the encasing] saves space… but requires a whole new level of manufacturing expertise. (iFixit)

Judging by the complexity … one might think the Apple watch is a full-fledged cellular connected watch… (ABI Research)

Very few people know its cost, even at Apple. And if you’re outside Apple, you could easily be off by 2X.

2.  The Force Touch display:

The cost of the Apple Watch flexible … display is estimated to be several times higher than equivalently-sized … displays… (DisplaySearch, acquired by IHS)

3.  The bands: Some alloys in the watch bands are either custom in composition or use custom tools and metalworking. Read Greg Koenig’s excellent article “How Apple Makes the Watch”. Custom alloys X  custom process  X  custom machines = a part cost that’s hard to estimate and one that’s probably higher than any reference price might indicate. (Fun exercise: in Greg’s article, search for words like “custom”, “different”, “unique”, “innovative”. They appear ~20 times.)

Now picture this – somewhere in China there are probably huge bins of shiny Apple Watch displays, svelte S1 modules, fresh batteries, boxy Taptic Engines, and bins of Digital Crowns and seductive watch bands — *scrap bins* that is. Full of imperfect parts. Yield rates for new technology are low. Yield rates for new technology that meets Apple’s standards are probably even lower than that. This increases a component’s cost and the price that Apple pays for it.

Somewhere in China there are probably huge bins of shiny Apple Watch displays, svelte S1 modules, fresh batteries, boxy Taptic Engines, and bins of Digital Crowns and seductive watch bands — *scrap bins* that is. Full of imperfect parts.

By the way, before someone says “Technology X in the watch is not new – the vendor showed it at a tradeshow in 2012”, let me say this:  a supplier may have shown off a prototype, or a demo of item X, but if it hasn’t been manufactured to scale… it’s new. Or if existing components are combined in a novel way… it’s new.

Comparable part prices:  Even when technology is not new or custom, the comparison prices that analysts use can be grossly inaccurate. Sometimes they’ll get lucky, and suppliers or other sources will share exact part prices, or even complete price lists. Other times, they won’t get that information. This leads to inaccurate component cost estimates. If the product has many parts, like a smartphone, the variances sometimes cancel out, and the estimated bill of materials is reasonable. If the product has fewer parts, however, the variances might add up in one direction or another, skewing the results. Smartwatches have fewer parts.

Do you think Foxconn (or whoever assembles the Apple Watch) put their “B” or “C” teams on Apple Watch? 

2. MANUFACTURING NEW STUFF REQUIRES GREAT PEOPLE, AND OFTEN CUSTOM TOOLS AND TECHNIQUES. AND AT LAUNCH TIME, BUILD-TO-ORDER CAN MAGNIFY THE COST

Here are some factors that drive up costs in a way that analyst firms usually don’t take into account.

Labor:  Take a guess – do you think Foxconn (or whoever assembles the Apple Watch) put their “B” or “C” teams on Apple Watch? This product probably has an “A+” team, which means engineers, supervisors, and assembly staff that make a higher wage.

Custom Machines and Custom Processes:  I know, I mentioned these earlier, at the component level. But custom machines and processes are also sometimes used for final assembly, again increasing cost.

Build to Order:  When all this is on a build-to-order basis — as it likely is during the first few months of Apple Watch – the costs of manufacturing (and logistics, below) are even higher.

3. PARTS AND PRODUCTS GET TRANSPORTED IN MYRIAD WAYS

Expediting:  This happens when a product is important, when it’s shipping later than hoped for, and when demand is greater than supply companies will expedite shipments. For instance, Apple might fly some watches on a 747 rather than use a cargo ship. Similar things can happen at the part level: Factory A might FedEx parts to Factory B, before final assembly. I don’t know if this is actually happening, but as Japan’s Nidec ramps up production of the Taptic Engine (to compensate for issues at other suppliers), it’s quite possible that it’s expediting parts to part-starved factories in China.

Complex Supply Networks:  Some Chinese OEMs leverage the Chinese Shenzhen ecosystem to its maximum: parts that go into their smartphones are built and transported within a fairly compact area in Shenzhen. Apple is the opposite. At Apple’s scale, it uses geographically-dispersed suppliers to a) simply keep up with demand b) to source what it considers to be the best parts; c) to have back-up options and negotiation leverage. What does this mean? More trucks, trains, ships, and planes than many other OEMs. And probably some related taxes too. (And yes, to be fair, probably some related tax breaks.)

Yes…things do cost less when you ignore the time when they cost more …that’s what makes it look easy.

4. ROYALTIES – A BLACK BOX

Royalties are the costs you pay to use someone else’s intellectual property. The totally royalties paid on a product are meaningful amounts (e.g., perhaps 3% to 18% of the retail price), and they vary by device manufacturer, supplier, product, product price, and where the product is sold. Unless you’re the VP of licensing or a VP in supply chain, you probably won’t know the figure for the entire product.

5. WARRANTY & RETURNS ARE FACTS OF LIFE

Yes, these are a part of the “cost of goods sold”. The returns, exchanges, and repairs are part of the cost to sell X number of watches. It’s hard to (accurately) guess this number for a product line. And when it’s for a product that people don’t yet really know if they want or need, and when there isn’t much data on how many units will actually fail and require warranty coverage, it’s even harder.

If you thought estimating your own s___ was hard, estimating someone else’s is doubly so.

6. PACKAGING AND IN-BOX ITEMS COSTS VARY MEANINGFULLY BETWEEN COMPANIES

Apple often introduces new packaging systems, higher quality materials, new levels of recycled content, and highly-engineered accessories (e.g., Apple’s Lightning cable is more complex than a USB cable. The Apple Watch wireless charger is even more so).

---

 

That’s it – these are the factors that I keep in mind when I read these analyst estimates. And as I mentioned, analyst firms don’t try to include all of these into their estimates. But these are the real-life conditions that drive device cost, and often, device price. Now some might say “Well, at steady state, the costs are different; they are lower”. … Yes…things do cost less when you ignore the time when they cost more …that’s what makes it look easy.

So how much do I think Apple Watch actually costs to make, if you include the cost I outlined above. 1) My estimate will be wrong, given the uncertainties I described; 2) My estimate is: $225 for the 42mm Sport, in the first few months. Perhaps $185 later, after build-to-order and component quality issues are resolved. I have more confidence in the $185 figure. For the curious, that would be 35% and 47% gross margin, respectively. If you recall, Apple CFO Luca Maestri said Apple Watch, as a product line, will make <40% gross margin in the first quarter of life.

To conclude: If you thought estimating your own stuff was hard, estimating someone else’s is doubly so. That’s why Tim Cook said:

Generally, [the] cost breakdowns that come out around our products … are much different than the reality. I’ve never seen one that is anywhere close to being accurate. And so if that’s the basis for your comment, I’d really dig on the data if I were you.

So, enjoy your Apple Watch, your Moto 360, your Pebble, or if you have an exercise band, your Jawbone or your Microsoft Band. It might not have been cheap to buy, but it wasn’t cheap to make, either. At least not as cheap as some estimates you’ll see on the web.