OLPC XO Hardware Upgrade Cost Impact: $170 Laptop?

   
   
   
   
   

Now that we've explored the OLPC XO BTest-3 hardware upgrade power, memory, and performance, I am wondering what the financial impact will be.

We already know that the OLPC XO "$100 laptop" is now the $150 dollar laptop, and when required to buy in 1 million unit blocks, a $150 million dollar minimum investment. In addition, with the Merrill Lynch OLPC cost estimate, we have the following breakdown of the OLPC XO component costs:

olpc bill of materials cost

With a 50% price over-run already of $150 vs. the original $100 price point, let's explore how the four main BTest-3 hardware changes; doubling the RAM to 256K and the Flash memory to 1 Gigabyte, upgrading the chipset to a Geode LX700, and switching to LiFePo4 (Lithium Iron Phosphate) batteries, might increase the Children's Machine XO base price.

  • RAM Memory: If 128MB DDR266 prices out at $10, 256MB would plausibly be another $10, but to give OLPC a volume credit assuming 5 million orders, let's say its $8 for argument sake.
  • NAND Flash Memory: With flash memory prices dropping slowly yet demand soaring, doubling the memory will most likely double the cost, no matter the volume. Another $8 in cost increases.
  • Geode LX700 Chipset: I would expect AMD to hold prices flat for the higher processor to promote its use, even though it would price out higher for another project. We'll keep it flat at $28.
  • LiFePo4 BatteriesThis is the mystery factor. While Computerworld implies it will be more expensive than Nickel Metal Hydride (NiMH) batteries, the actual cost impact is unknown because the technology itself is still in development. $5 is the current guesstimate.
So in total, the OLPC XO BTest-3 estimated costs could now be around $171 dollars per OLPC laptop, when purchased in at least one million unit blocks from Quanta Computer. Or maybe not.

Could the One Laptop Per Child team and Quanta come up with hardware judo that not only double upgraded the OLPC hardware but also maintained or reduced its wholesale price? Hardware experts in the know, the Comments section is open for your estimation.

Related Entries

15 Comments

When was the original cost estimate published? If it's, say, six months or older, than today's real value may be quite different, lower, mostly for the memories.

Nick,

The cost estimates are from Merrill's report dated March 1 http://www.olpcnews.com/olpc-invetment-report.pdf which means prices of late February or so, or quite current.

Its worth noting a few things regarding components in the laptop.

Firstly the LiFePo4 batteries are a different chemical structure and supply a different voltage. The previous NiMH batteries produced 1.2V per cell therefore needing five cells to produce the needed 6 volts (minimum).
The LiFePo4 cells produce 3.2volts (up to 3.8volts charing) therefore only 2 cells are needed. Are 2 LiFePo4 cells cheaper than five NiMH cells and would they have the same capacity?

I didnt see a price for the camera component in the list. Would that maybe come under the 'others'?
I'd suggest that the device price for the camera has dropped.

Why a $10 price for the OS? I was under the impression that it would be a free operating system. Is there licensing involved? Does Red Hat get a cut from this? They do seem to be putting in a lot of work.

I'd still point out that unless someone has the bill of materials and prices that calculating cost (and a possible rise of a little over 10%) is pretty hit-and-miss.

The laptop may actually only cost $100 in components for Quanta to build. We just dont know...and they aint saying.

Robert,

Actually, multiple reports put the Bill of Materials for Quanta's configuration of hardware at $140 in cost, plus $10 in profit split between all the component manufactures, inclusive of all the BTest-2 components - camera, OS, etc http://www.olpcnews.com/prototypes/olpc/olpc_xo_100_dollar_laptop.html

My point is that hardware 'cost' is a relative term. Regardless as to the make up of the total cost of components, wouldnt the important issue be the total price of ownership. If a country buys one million laptops from Quanta for $150 million then the 'price per laptop' is $150.

Quanta could quite possibly have a component bill of $100. Any speculation about 'cost of parts' is irrelevant - unless it increases the 'price per laptop' and they become the $171 laptop as you mentioned.

Then, by that logic, these are at least $208 laptops based off the Libyan MOU - our best estimate of the total initial price of ownership when hardware & basic implementation costs are factored in. No word on maintenance, content development, Internet connectivity, etc.
http://www.olpcnews.com/countries/libya/one_laptop_per_libyan_child.html

If you look at DRAMexchange, http://www.digitimes.com/bits_chips/a20070403PR211.html The current spot price of dram (ddr2 eTT, which is of high quality but unbranded) is $11 for 256 MB and $10 for 1GB of flash, which roughly half the cost you and Merril are quoting.
Also with about 1/3 of the price being other, it's hard to consider this "estimate" accurate.

Tim,

That DRAM info is very interesting. Since Quanta still says its $140 base for OLPC XO, what might you think each other part & assembly cost? Do you think the Geode LX700 Chipset or LiFePo4 batteries might cost more then?

Wayan,

DRAM and flash are unique in that they are comodities supplied by many manufacturers, and have a spot market. Actual quantity prices of most other components are negotiated and secret. Software lisencing cost is probably zero. Most of the $28 in the cost of the processor is probably amoratizing design and tooling, so AMD could be charging anything from $15 to $30, depending generous the want to be in their support for OLPC. I think that Negroponte said that the LCDs are going to be built on semi-dedicated,older generation LCD fab lines, so the costs would be very volume dependent. Besides I have no idea how LCD fab lines are valued in China. (I also have no idea what LiFePo4 batteries cost, but since they have a higher voltage than NiMH, you don't need as many.)

My point was that people outside of Quanta and OLPC can only make wild guesses at what the laptop will cost.

Guys - BOM analysis is not that difficult to do. People do it all the time to look at what is in a system and since MOST of the parts are reaily avaialble - and usually commodities for the most part - it is easy to see the price. What you are all missing is TWO things.

1. By increasing these hardware components, the COST of the OLPC XO WILL GO UP! No questions here...the debate is by how much and I personally think that Wayans estimates are pretty damn good.

2. Secondly - AND MORE IMPORTANTLY - is the philosophy that Nick setup for himself that HE IS BREAKING. Keep the hardware constant and the prices will come down and make the whole system cheaper over time...this WAS one of his cardinal rules. SO then, why is he breaking that rule now? He is already MORE THAT 50% over his promised COSTS and I would say he is even more than that with the new upgrades...he is increasing the costs, not decreasing them!

Ok - so three points - the third being the total cost of ownership being the real cost! There was a recent article saying the real cost of OLPCs were more than ~$970 (http://hardware.newsforge.com/article.pl?sid=06/12/01/1546208) and also linked here on OLPC News...and I would argue it is even HIGHER with NO Post Sales Support model and NO easy way to repair...and no integration within the classroom to bring this in as part of REAL education - not just tinkering and playing (Difference between a toy and a tool - but that is a topic for another article).

OLPC Skeptic, the original source for the NewsForge was OLPC News: http://www.olpcnews.com/sales_talk/price/the_real_cost_of_the.html

Wayan,

Sorry, my bad - innocent mistake.

my goodness, why such significant hardware changes so late in the game?

If the original specs( 500MHz CPU, 128MB RAM, 512MB FLASH ) were capable enough for Sugar to begin with, you should have been getting this out the door and talking about a hardware upgrade around 2 years out.

But if this added cost and delay is due to sucking up to Microsoft for supporting Windows, you guys are playing into their hands. Regardless of what Microsoft tells you, Linux must die in their minds and the OLPC project is a massive threat to growing Windows market share which can only happen outside of the developed world. Please do NOT fall for this ploy and therefore threaten the whole OLPC projects success.

I couldn't agree more with you Doug, they just want to crash every thing that could falter there monopoly, it can be seen so clearly with ODF, microsoft is doing everything to stop it from gaining ground so every people would be drawn to use it's office products.

OLPC would be completely best as open system and not as some closed proprietary system for couple simple facts.

1. When being open system bugs and problems get fixed quickly, when there is a lot of coders around the world helping in development.

2. Windows has so many security holes in it that if it would be used in OLPC there would be millions of completely identical systems and if there would be internet connection in them it could create a really, really big zombie network.

3. Adding to above you would need to have softwares just to protect your system and that would make lot more costs to the normal price.

4. Adding even more to above, people how have used windows pretty much know that you need to maintain your system regularly, so would you really want to give that kind of system to childes in developing countries and then teach them to maintain them?

OLPCSkeptic: the USAID project which I referenced and used some costing figured from in the "Real Cost" article included a lot of teacher training and curricula development specific to hardware features, so some of that's built in. As for repair costs, practically all of the inside is soldered together, so unless you're plugging on new "ears" or swapping the monitor chunk, you're replacing the whole laptop. I gave a margin of 5 computers lost or broken per 100 per year, which I think is reasonable (possibly low?)

Close