Eso es lo que menciono en el OP y lo que ellos mismos aclaran en la penúltima página del PDF. Que existen varias formas de analizar el precio de una consola y que ellos han optado por hacerlo así. Comentan que el margen de error estimado, por los factores mencionados, puede ser de un ~15%.
Cost modeling is tricky business. Multiple variables affect the actual production costs a manufacturer will experience, including development expenses,
unit volumes, supply-and-demand in component markets, die yield-curve maturity, OEM purchasing power, and even variations in accounting practices.
Different cost modeling methods employ different assumptions about how to handle these and other variables, but we can identify two basic
approaches: that which seeks to track short-term variations in the inputs to the production process, and that which strives to maintain comparability of
the output of the model across product families and over time.
Portelligent’s philosophy in cost modeling is to emphasize consistency across products and comparability over time, rather than to track short-term
fluctuations. During the past eight years, we have developed an estimation process that, while necessarily lacking an insider’s knowledge of the cost
factors that impact any one manufacturer, is reasonably accurate in its prediction of unit costs in high-volume production environments. We do not
claim that the model will produce the “right” answer for your firm’s environment. However, Portelligent does give customers a key analytical tool with a
complete set of data in our Bill of Materials (BOM). The BOM allows readers to 1) scrutinize the assumptions behind our cost model and 2) modify the
results based on substitution of their own component cost estimates where they have better information based on inside knowledge.
Our estimation process decomposes overall system cost into three major categories: Electronics, Mechanical, and Final Assembly. We begin by creating
a complete electronics bill-of-materials (BOM). Each component from the largest ASIC to the smallest discrete resistor is entered into a BOM table with
identifying attributes such as size, pitch, I/O count, package type, manufacturer, part number, estimated placement cost, and die size (if the component
is an IC). Integrated circuit costs are calculated from measured die area. Using assumptions for wafer size, process type, number of die per wafer,
defect density, and profit margin in combination with die area, an estimate of semiconductor cost is derived. Costs for discrete components and
interconnect are derived from assumption tables which relate BOM line items to specific cost estimates by component type and estimates for part
placement costs are included. For LCD display costs, we employ a model which tabulates expected cost from measurements of glass area, LCD type,
and total pixel resolution. When market costs are available from alternative sources, LCD panel costs are taken from and referenced to these sources.
Costs of mechanical enclosures and fasteners along with the cost estimates of final system assembly and integration are modeled using Boothroyd
Dewhurst design-for-assembly (DFx) tools (see next slide for further detail). Other system items such as optics, antennae, batteries, and so on are
costed from a set of assumption tables derived from a combination of industry data, average high volume costs, and external sources. In effect, we rebuild the torn-down product, tabulating final assembly costs as the process of reconstruction proceeds.
The three major categories for system cost contributors can be broken down into the subcategories of ICs, other electronics parts, displays (as
appropriate), printed circuit boards, electronics assembly, mechanical/housing elements, and final assembly. By adding the cost estimates for each of
these subcategories, an overall estimated cost is derived for the system under evaluation. Product packaging and accessories (CDs, cables, etc.) are
also documented and estimated for their contribution to total cost as appropriate.
We believe our cost estimates generally fall within ±15 percent of the “right answer,” which itself can vary depending on the market and OEM-specific
factors mentioned earlier. While the Portelligent cost model is imperfect, it yields important insights into technology and business dynamics along with
good first-order contributions to system cost by component type. Additionally, the consistency of approach and gradual modification to assumptions
(smoothing out frequently-shifting pricing factors) hopefully yields a credible, but user-modifiable, view of OEM high volume cost-to-produce.
Please feel free to contact us at info@portelligent.com with any comments, questions, or proposed corrections with respect to our cost estimates. We
welcome your input.
