| Episodes | 206 |
| Mentions | 314 |
| Cited here | 122 |
| First — last | #2 — #722 |
| Top guests | Jeff Keyzer, Chris Church, Mike Harrison |
| Related | digi key · bill of materials · component lead time · kickstarter · microcontroller |
Component sourcing is the identification, qualification, and procurement of the electronic parts from which a circuit is built, and it is one of the dominant costs of electronic design: finding, comparing, and qualifying parts can consume eighty to ninety percent of the engineering hours on a new design.[8][2] Because a single missing line item can stop an entire production line, availability is treated as a first-order design constraint rather than a procurement detail, captured in the maxim that engineering is “the art of building something with stuff you can get”.[646][366] The field spans catalogue distributors, manufacturer-direct sales, brokers and grey markets, and the stocked part libraries of contract manufacturers, each with distinct pricing, provenance, and lead-time behaviour.[367][600][216]
Role in electronic design
Component engineering—finding, comparing, and qualifying parts—can absorb the great majority of project hours, with designers reporting that the bulk of a project’s time is spent moving between distributor catalogues looking for suitable parts; this burden motivated EDA vendors to embed distributor stock and pricing lookups directly in schematic tools.[8][2] The work does not end when the schematic is drawn: availability churn drives schematic rework, because a converter available two months earlier may be gone while a previously unavailable alternative has returned, so revisiting a bill of materials means re-searching cost and availability across distributors.[128]
The practical knowledge of manufacturing and sourcing—data packs, standardised bills of materials, how to deal with contract manufacturers—is transmitted informally rather than taught, which constitutes a systematic gap for first-time hardware builders.[295] Sourcing must be designed in rather than delegated, since a turnkey assembler or fulfilment partner cannot repair a bill of materials that was never checked for availability.[197]
Sourcing channels
Catalogue distribution
Catalogue distributors are chosen on three axes: shipping speed, availability, and price.[38] Mouser, for example, ships from a distribution centre near Dallas, so orders placed with the cheapest ground service still arrive the next day across much of the United States.[38] For small builds, catalogue distributors are preferred over local suppliers because stock levels and backorder dates are visible and no negotiation or quotation cycle is needed, even though the unit price is higher.[178]
The sector has consolidated into a few groups: Farnell, Element 14, and Newark are regional brands of one distribution group serving the United Kingdom and much of Europe from a central hub, Avnet owns the Farnell and Element 14 brands, and Arrow operates several brands of its own, so apparently independent sources can share an owner.[266][367] Distributors have also bought into electronics media and search: Arrow acquired EE Times in 2016, and services such as GlobalSpec, FindChips, and Octopart sit in the component sourcing business.[504]
Many listed suppliers hold no stock at all; identical quantities appearing across several websites indicate that the same underlying inventory at a large distributor is being resold, with the order passed upstream on receipt.[367] A reseller showing stock but shipping several days later is buying the parts in from elsewhere, unless the seller is the manufacturer.[408] Listed quantities can also diverge from physical stock as parts are pulled from trays, so a listed quantity can be met with a cancellation notice when the bin holds fewer parts than the system believed.[598]
Distributor stock has a hard boundary: the unit beyond the quantity in distribution must be ordered from the manufacturer, bringing factory lead times and minimum order quantities into play, so a quotation should be obtained at the intended build volume early in the design.[367] Ten instances of a part on each of a thousand boards already exceeds what a catalogue distributor holds, at which point parts are bought as reels direct from the manufacturer.[645] Stock checked one day can be zero the next, replaced by a twelve-week lead time, so availability verified during design is not availability at the moment of ordering.[287]
Cut-tape and small-quantity handling is effectively a loss leader against the volume business, since the labour of cutting a strip off a reel and bagging it is not covered by the line price; cut-tape reels carry a per-reel handling charge and a leader tape, and pick-and-place setup can consume a large fraction of a small quantity, so ordering ten parts on a reel may not yield ten placements.[400][410] Freight and handling overheads dominate small-quantity sourcing: a hundred dollars of shipping on five modules doubles their effective cost, and a seven-dollar per-reel charge across fifty unique parts adds three hundred and fifty dollars to a build.[508]
Manufacturer-direct sales and representatives
Before catalogue distribution reached worldwide one-off quantities, parts were bought through a manufacturer’s local representative network with account paperwork and bank payment; current-generation devices are now listed at reasonable single-piece prices from the day of release.[207] European distributors such as Farnell and RS were historically far more expensive in real terms and required a corporate credit account, which made the component hoards accumulated by individual engineers valuable.[246] Earlier still, sourcing a mechanical or electronic part meant working through printed manufacturer directories such as the Thomas Register and telephoning each listed maker to find who was still in business.[429]
Manufacturer web stores frequently accept an order and take payment without showing stock levels, only disclosing months later that the part is unavailable; distributor listings that expose live stock are safer for committing a design.[74] Manufacturer-direct stores price-match catalogue distributors rather than undercutting them, because both parties want to hold margin, and a five-thousand-piece reel bought from a semiconductor manufacturer’s own web store costs about the same as the same reel from a catalogue distributor because distribution agreements hold the prices together.[400]
Access to large semiconductor vendors is gated by volume. A large wireless semiconductor vendor refused to open an account or provide field applications engineering support without a commitment of around 250,000 units per year, and the very largest vendors will not engage below multi-million-unit volumes, so buyers below that threshold reach the silicon indirectly through module makers who do have the relationship.[295][474] Established buyers can obtain a part recommendation and ready-made CAD libraries from a vendor’s field engineers within days, a channel startups without volume commitments do not have, so their part selection and library work is done unaided.[295] Smaller semiconductor vendors will work more closely with small customers than large ones will, because they have more to gain from the relationship, which makes vendor size a selection criterion in its own right.[474] Some component suppliers will not sell to an individual at all and require a registered company before releasing parts, a barrier at the earliest prototype stage.[475]
Marketplaces, brokers, and the grey market
Alibaba functions as a Chinese component sourcing marketplace aggregating many suppliers rather than as an auction site, filling a role analogous to a Western catalogue distributor.[79] Most sellers listing components on Chinese marketplaces are trading companies or listing agents rather than the manufacturer, so the party a buyer negotiates with often has no control over the product; many companies presenting themselves as component or subassembly suppliers are middlemen, which limits how far a buyer can develop the technical partnership or pricing they expect.[420][350] Companies buying directly from Asian online dealers staff the function with a Mandarin speaker who negotiates in chat and voice rather than relying on English-language sales channels.[6] The Canton Fair draws roughly 210,000 buyers sourcing components and finished goods, illustrating how much industrial purchasing runs through Chinese trade events.[100]
Physical markets remain significant. Huaqiangbei in Shenzhen is the world’s largest wholesale component market, with thousands of stalls spread over six floors of some twenty buildings, organised for manufacturers but open to individual buyers.[125] In Indian cities, components are bought from a dedicated market district of small densely stocked shops, where staff will work through variants with a buyer in person; there is no catalogue or telephone equivalent.[618]
A broker sits between the buyer and an unknown upstream source, so the buyer cannot see where the parts came from; most brokered transactions deliver good parts and the occasional one does not.[157] Brokers price on their ability to find parts rather than on the parts themselves, quoting a fee per thousand pieces on the understanding that a high enough price will surface stock somewhere.[600] A broker with a legitimate-looking website that passes casual diligence and even delivers a first order can still be a fraud that disappears afterwards; the failure mode is not detectable from the web presence.[157]
Grey-market stock is often genuine rather than counterfeit: a contract manufacturer left with part of a reel after a build cannot use it once its date code has aged, and sells the remainder into the secondary market.[84] The grey market will break reels, selling a thousand pieces off a three-thousand-piece reel measured out by length, at prices such as eight cents apiece, but brokered parts carry no guarantee of provenance.[84] The risk carried by a grey-market or unbranded Chinese component scales with how hard the part is to verify: an LED can be checked by powering it, whereas confirming that a microcontroller is genuine and in-specification is far harder.[79]
The conservative sourcing rule is to buy only through authorised distributors and avoid consumer marketplaces entirely, with the counter-argument being the price premium authorised distribution carries.[625] LCSC is an authorised distributor for the brands it lists and does not substitute parts, so a buyer receives the genuine device even when it is an unfamiliar Asian brand; consumer marketplaces such as AliExpress carry no such guarantee.[674] A purchasing department optimising purely on unit cost can buy grey-market parts that are counterfeit or out of specification—a connector quoted at a tenth of the branded price is the shape of that failure.[491] Where a device is only obtainable through informal sellers, incoming inspection becomes mandatory, with parts tested before being fitted to a customer’s board because provenance cannot be established any other way.[507]
Contract manufacturers and house part libraries
Contract manufacturers keep a house parts library of components they continuously restock and preload; designing to that library lowers cost because no new procurement is required for the build, and the customer names the library item rather than a specific manufacturer part.[216][125] Turnkey services from Chinese board houses are cheaper again when the design uses their suggested parts, because those reels are already loaded in the pick-and-place machines.[682] The turnkey route only suits designs whose components are generic; a precision instrument using 0.05 percent resistors and specific amplifiers has no substitution latitude and cannot be built from a house parts library.[682]
Component manufacturers will help an assembler obtain distributor discounts in exchange for that assembler stocking a hard-to-find part as a standing house part, because a stocked part gets designed in.[243] Low-cost assembly services only place parts they stock themselves, so a design that mixes stocked and non-stocked components becomes a partial machine build followed by hand assembly of the remainder.[552]
An overseas cost-down pass does not redesign a circuit; it replaces the specified devices with cheaper equivalents from the local component ecosystem, which then have to be validated against the original specification.[578] Substituting a locally standard Chinese component for an unusual one specified from a Western catalogue can reduce that line item by an order of magnitude, which is why contract manufacturers return marked-up part lists on a first build.[121] Conversely, a contract manufacturer that cannot source the exact specified part locally may import it instead, paying duties and returning an unexpectedly high quote, so a quote that looks too expensive is worth querying.[125] Offshore assemblers substitute components without notice unless supervised, so a customer moving production overseas must police the bill of materials continuously.[126] Component manufacture co-located with assembly in China allows a build to be scaled within days in a way that ordering from a Western catalogue distributor cannot match.[363]
Part selection practice
A workable part-selection procedure is to start from a vendor application note for the same function, take the part it specifies, then use a distributor’s parametric search for similar parts sorted by price.[141] If every part meeting a set of specifications costs about the same, the price is being set by the specification rather than by the choice of vendor, which is a signal to re-examine whether the specification is really required.[141] Selecting the highest-volume version of a commodity part lets a small buyer ride the pricing and availability created by large buyers, bringing capacitors down to fractions of a cent.[70]
Western distributor catalogues list every part but do not indicate which parts are commonly used and therefore available at manufacturing scale; that information lives in local markets and Chinese retail sites rather than in the catalogue.[121] A remote technique for discovering which parts are standard in the Chinese supply chain is to translate the generic component name into Chinese and search Taobao with it, using the relative prominence of listings as a proxy for how common a part is.[121] Buying the schematic and parts list of a high-volume consumer product—obtainable for a few dollars in the Shenzhen markets—is a deliberate technique for identifying components that are cheap and plentiful because a large manufacturer is already buying them.[109]
Industry practice favours established parts from large manufacturers, because a sole device from a smaller or restructuring vendor can be cancelled and leave a shipping product with no replacement.[191] A quick availability sanity check on a candidate part is to search the part number with the word “buy” and see whether several independent sources appear; a thin result set signals a sourcing problem before the part is designed in.[380] An unusually low distributor price can signal that stock is being cleared ahead of end-of-life rather than a bargain, so price alone is an unreliable basis for designing a part in; semiconductor manufacturers schedule production from demand statistics on each part number, so a device with declining order history drops out of the build plan and reappears as a long lead time for whoever is still using it.[210]
A paper catalogue can surface a substitution that a parametric search will not, because unrelated part families appear on the same page; a mobile phone SIM connector costing tens of pence has been used as a sprung surface-mount contact in place of a spring probe.[294] Once a component is purchasable from a catalogue distributor it carries little supply-chain differentiation, since any competitor can obtain the same parts, so the differentiation of a product built entirely from catalogue parts has to lie elsewhere.[259] In precision instruments a substantial part of the value lies in sourcing rather than circuit design: obtaining higher-specification parts than the open market offers, or binning ordinary parts by test, is what produces the published performance.[554]
Part selection can be risk-tiered by duty: unbranded low-cost silicon is acceptable in an unstressed battery-powered design, while components operating near their limits are restricted to manufacturers with a known record.[671] Passive component quality cannot be verified by the buyer, since establishing a capacitor’s endurance would require thousands of hours of stress testing per batch, so selection rests on manufacturer reputation and trust in the seller.[671]
Second sourcing and substitution
High-volume design requires selecting parts that have a second and third source; a pin-compatible alternative from another vendor is what keeps a product shippable when the primary supplier stops delivering.[5] A bill of materials is only de-risked when each critical line has a second source and a stated fallback, including mechanical and specialty items whose single supplier could vanish or reprice.[173] A second source need not be a different manufacturer: a higher-grade or higher-priced variant from the same vendor, in the same package, can serve as the substitute that keeps a build moving.[104] A single-sourced part with a three-month lead time can force a board respin purely for supply reasons, with no change in the electrical requirement.[104]
Vendors increasingly release parts in unique footprints, which makes a competing device a board respin rather than a drop-in and so hardens the single-source position; many modern devices are effectively single-sourced by construction.[136][173] A part chosen for its feature set can turn out to have no pin- or package-compatible relatives within its own family, leaving the design locked to one orderable device with no substitution path.[269] A design that deliberately uses the most commoditised footprints keeps several manufacturers’ parts droppable into the same land pattern, achievable for standard logic but not for switching regulators or microcontrollers.[453] Designing around interchangeable functional blocks specified only by their interface lowers supply risk but sharply raises the design and qualification effort, because every candidate part must be accommodated.[273]
Part equivalence is project-specific rather than absolute: a substitution acceptable on one design may be barred on another by tolerance, temperature coefficient, durability, or a requirement that the part be sourced from a particular country.[542] Swapping in an electrically identical part from a different manufacturer should still be recorded against the build, because lot-to-lot differences must remain traceable to the units they went into.[178] A practical substitution search during a shortage means opening every device in a distributor’s category listing and checking data sheets for a matching footprint, rather than relying on parametric equivalence, because a functional equivalent frequently arrives in a different package and the substitution becomes a board respin rather than a bill of materials edit.[637][577] Through-hole packages of a device sometimes remain available when the surface-mount versions do not, making the older package a legitimate shortage workaround.[613]
Pricing and volume
Component pricing depends on the buying relationship as well as the quantity on a single line: a customer taking a thousand different reels from one supplier is priced differently from one taking a single high-volume reel.[96] A designer phoning a vendor for volume pricing without a purchase history will not receive a real quote; obtaining genuine volume prices is the work of a full-time purchasing person with established vendor contacts.[96] Pricing also tracks how badly the buyer needs the part, so telling a vendor that a design is committed to their device removes the buyer’s leverage, and disclosing a full production bill of materials carries this cost.[136]
The traditional quotation flow for a build is to email the bill of materials as a spreadsheet to several distributors and compare the returned prices line by line; line items are split across distributors only when the saving justifies the extra handling, and where the difference is around a hundred dollars, buying the whole bill of materials from one supplier is the better trade.[411] Shipping costs often erase the savings from splitting a bill of materials to chase the lowest unit price, since a few cents saved on a line item can be swamped by an eight-dollar or fifty-dollar carriage charge.[243] Loading a bill of materials into a single distributor’s BOM tool converts reordering into a single transaction with correct part numbers and quantities; the price premium buys back the time and the certainty.[104]
Semiconductor vendors publish reel price breaks directly, with a 500-piece reel of a microcontroller at eighty cents a unit against seventy cents on a 3,400-piece reel.[574] Precision parts are single-sourced and carry order-of-magnitude price breaks, with a resistor costing around four dollars in reel quantity against roughly twelve dollars as a single piece, so low-volume production absorbs the difference directly out of margin.[554] Minimum order quantities dictate step changes in cost: parts not stocked in one-off quantities must be bought by the reel, so an extra thousand units of product can require an extra thousand-piece reel at four dollars a part.[177] Small buyers face a structural penalty at the boundary between catalogue and reel quantities: buying single reels from a catalogue distributor costs more than the local supplier price available to someone taking twenty reels.[160] House parts and market parts cross over as volume rises: an assembler’s stocked house part can be cheaper in small quantities while a market part saves a few cents a board once volumes reach the low thousands, so the comparison has to be made at the intended build quantity.[243]
Component manufacturers set minimum engagement quantities that exclude small builders: a valve supplier would not discuss an order below two thousand pieces against production runs of one to two hundred, forcing bulk purchase and inventory, and camera module vendors treat a million units as a small order, so an order of ten thousand attracts both a high unit price and a lead time of around eighteen weeks.[618][517]
Volume arithmetic has to be done before part selection: fifty LEDs on each of fifty thousand boards is two and a half million parts, a quantity available from few sources and carrying a lead time of at least twenty weeks.[502] Once parts are ordered at production volume the design is frozen against them, so a late discovery that a device is thermally or electrically unsuitable strands the committed inventory.[502] Component choice can be dictated purely by which part a distributor holds in sufficient quantity, with electrical selection reduced to whatever device there is enough stock of.[412]
Lead times, allocation, and shortages
Semiconductor lead times of thirty to forty weeks followed the fab shutdowns of the 2008–2009 downturn, because restarting mothballed capacity is slow; forty-week lead times were normal for semiconductors in the early 1990s, and shortages since have pushed typical lead times back towards twenty-eight weeks and beyond.[5][175] Lead times published on distributor websites are a starting figure rather than a prediction, and parts frequently arrive substantially sooner than the quoted date.[472] A production shock propagates into component availability on a delay: interruptions at a passive component factory do not appear as distributor shortages for weeks, which makes the leading indicators easy to miss.[482]
Semiconductor allocation follows revenue: engineering support and supply are directed at the largest customers, so a design consuming a few thousand parts is served after those consuming millions.[227] Very large buyers commit cash to multi-year purchases of chip and fab output, removing that capacity from the market for everyone else, and a single very large customer buying an entire fab’s output can remove a part from the market—a supply risk a small company cannot mitigate by ordering earlier.[240][437] Scarcity also propagates from the bottom of the market: a single small manufacturer can absorb the entire distributor stock of a part, and a single project can consume the world stock of a specific part number—roughly thirty thousand connectors bought out across several distributors leaves any concurrent user of that part with nothing.[527][294]
A single missing line item stops an entire production line, so a bill of materials has to be complete rather than mostly available.[646] Every unique reel on a bill of materials is a separate procurement risk, and a shortfall discovered at the assembly house produces substitution requests mid-build, such as whether a 1K resistor may be fitted in place of a 1.02K.[500] A lead time quoted by a distributor is often matched by the manufacturer, so escalating to the factory does not necessarily shorten it; a two-month wait on a single critical resistor can hold up an entire product reorder.[210] Buying critical parts up front on behalf of an assembler preempts the delay between contract signature and their purchasing department placing the order, during which stock can disappear.[197] An assembler’s quoted component price should be checked against aggregator listings before a design is changed: a part quoted at three dollars fifty each was available at around a dollar fifty elsewhere, which is cheaper to resolve than a redesign.[546]
The 2020–2021 shortage
The 2020–2021 shortage was amplified by demand-side behaviour: customers cancelled backlogs at the start of the pandemic, then reinstated them within months, and semiconductor capacity had already been reallocated to buyers who never cancelled.[530] Distributor stock of zero paired with a thirty-two-week lead time was a routine state, including for simple eight-pin through-hole controller chips, and the shortage reached basic through-hole components, with plain T1-3/4 LEDs taking three to four months to arrive—distinguishing it from earlier years when one backordered part would clear in a couple of weeks.[541][570] Nearly all the active devices on a typical hundred-part board became unobtainable while the roughly eighty percent of the bill of materials that is passive stayed available, so the redesign burden concentrated on the ICs and FETs; repeated substitutions drove board revisions on their own, with one small product reaching its fourth revision purely because components kept becoming unavailable.[601][568] Board-to-board connectors for compute modules became unobtainable in 2021 and 2022 and were found only by canvassing industry contacts and forums outside normal distribution channels; a design house can supply scarce components to a client’s manufacturer as free-issue material so that a build can continue.[645]
During an allocation the working rule shifts from designing a part in and buying it later to buying stock at the moment it appears, in the quantity the known build requires.[530] One working practice was to buy ahead on the assumption that any order might take a full quarter to arrive, keeping enough stock on hand to absorb the delay.[472] Small hardware firms bought a year of chips ahead of production runs that had not been scheduled, tying up working capital in inventory for products that might not be built.[552] Panic buying propagates a shortage: a designer who switches platform and then buys three reels of the replacement removes that stock from whoever had planned on it, moving the problem one step down the chain.[573] Personal or shelf stock becomes a real supply channel during a shortage, with builds completed from parts held outside distribution when nothing can be bought.[542] Buying from the grey market at inflated prices is a rational choice for a company that must ship to make payroll, which is why broker pricing holds during an allocation.[601]
The shortage shifted risk tolerance in part selection, making designers willing to qualify unfamiliar Asian silicon and to search Chinese distributors first, where previously they would have defaulted to a known family that was in stock; Chinese-designed commodity interface chips undercut Western equivalents substantially, with a USB-to-serial device listed at 48 cents in one-off quantity against more than 122,000 pieces in stock.[595][587]
Shortage reports stated as a blanket claim that a distributor is out of a whole component class are usually false in general but true for specific case sizes and voltage ratings, alongside higher prices, so the response is to re-specify rather than to give up.[391] In the capacitor shortage of the late 2010s, a reel of surface-mount one-microfarad capacitors could not be obtained at any price, showing that during a severe allocation money does not substitute for availability; the shortage bit hardest on high-voltage and high-capacitance parts, while commodity values such as 100-nanofarad 0603 remained obtainable at roughly double their previous unit price.[596][412] Popularity itself creates scarcity: the lowest-end and most widely used microcontroller variants became the hardest to obtain precisely because everybody had designed them in.[412]
Small Chinese manufacturers of commodity goods build from whatever parts the local market holds at the time, so successive orders of the same model number can differ in both hardware and firmware.[539]
Obsolescence and product lifecycle
Products intended for long service lives require written commitments from component suppliers that a part will still be supported ten to fifteen years out; obtaining such letters from semiconductor vendors is difficult but possible.[1] An aftermarket industry exists specifically to keep obsolete semiconductors available for long-life programmes, with Rochester Electronics as an example of a company built on that need.[1] Replacing an obsolete discrete semiconductor typically means working down a chain of increasingly marginal sources: catalogue distributors first, then specialist obsolete-part manufacturers, then brokers.[290]
Distributors increasingly forward end-of-life and obsolescence notices to customers who have previously bought the affected part, which gives a design owner an early warning channel.[382] A shipping bill of materials should be re-run against distributor availability every few months, since re-quoting is what surfaces a part that has quietly become unobtainable.[382] Designing in an obscure part carries two compounding penalties: it is more likely to be discontinued as its customer count falls, and during an allocation its users are at the back of the queue with no distribution stock to buy through.[384]
Poor component selection early in a programme can strand it: on a military project the prototype hardware could not be rebuilt because the specified parts were no longer obtainable, and the design could not be revisited within the schedule.[116] An obsolescence can arrive with no last-time-buy window at all, leaving a manufacturer to purchase the remaining world stock; where the part defines a platform that other modules plug into, substitution breaks compatibility with everything already shipped, and one remedy is to commission a manufacturer to copy the original design and produce a compatible device.[178]
The historical arrangement was that a chip customised for a large customer was afterwards released to general distribution, so the part eventually became purchasable by anyone; that route has narrowed, leaving devices that never reach open distribution.[507] When a device is unavailable in any distribution channel, harvesting it from a new retail assembly that contains it can be the only supply route; charging cases have been bought purely to remove one power management chip.[507] Rebranded and house-marked microcontrollers make a product unrepairable in practice, because the replacement device cannot be identified or sourced.[554]
Procurement organisation and tooling
In-house buyers act as a design gate, rejecting parts that cannot be obtained in the required annual volume regardless of how well the design works.[157] Large organisations force new components through an internal approval and part-numbering process before they can be used, because at production volume the part must be purchasable, payable, and schedulable across multiple approved vendors, not merely electrically correct.[435] Internal part numbering decouples a design from its vendors: one company part number for a logic device can map to five approved manufacturers, and purchasing buys against whichever is available without engineering involvement.[445] High-end CAD tools carry approved alternative parts against each bill of materials line, which lets purchasing pick whichever qualified device is available in a given week without returning to engineering.[178] Some companies impose a blanket engineering rule forbidding the use of a particular manufacturer’s parts, enforced by purchasing rather than by design review, when that manufacturer has repeatedly failed to deliver production volumes.[44]
Supply chain information has traditionally been opaque to design engineers because it is held by a separate purchasing team and consulted weeks or months after the part selection was made.[577] A procurement team cannot substitute an unavailable part on its own authority, so every shortage becomes a context switch back to the engineer to approve, test, or re-select—a structural inefficiency between design and manufacturing.[577] The distributors familiar to engineers are not the ones used for volume production; suppliers digitising their catalogues and exposing pricing and lead time through APIs is what lets a designer see real availability at the point of part selection.[577]
Component inventory software tracks what parts are physically held, associates them with builds as they are consumed, and drives purchasing from the resulting shortfall.[542] Parts held at a contract manufacturer can go missing without a tracking system, so an inventory system is used to record where each part physically is as well as how many there are.[722] Enterprise resource planning for hardware ties design data to purchasing and receiving, so components are ordered against forecast builds and their arrival dates are tracked; the established systems are priced for large companies, with customisation of a major package running to about a million dollars.[722]
Supplier payment terms function as working capital: net thirty on components let early Apple buy parts, build computers, and sell them before the invoice fell due, financing production without capital.[417]
Logistics and geographic constraints
International courier shipments of components can sit in customs clearance for weeks without notification, so transit time on an imported reel is not the same as the courier’s quoted service level.[231] Shipping components into countries outside the main distribution networks carries disproportionate cost and delay: three dollars of cable delivered to Chile cost two hundred and fifty dollars, sat three days in customs, and attracted import tax, with a week as the fastest achievable transit.[248] Component availability is geographically determined: an engineer in Yugoslavia had to travel a thousand kilometres to Munich to buy parts that are now purchasable locally.[247]
Chinese factories and component suppliers shut for roughly two weeks around Chinese New Year; orders placed just before the break are accepted but not shipped until production resumes.[79] A turnkey build of a simple fifteen-component board through an overseas assembler took about two months end to end including quotation cycles and shipping, with the Chinese New Year shutdown adding further delay.[232]
Export controls constrain sourcing directly: sanctions blocked Russia and Belarus from buying Taiwanese processors above a clock threshold, and comparable rules govern international distributor purchases; export regulation can also force a manufacturer to cripple a product for foreign markets, with thermal cameras above roughly eight or nine frames per second treated as controlled and units sold outside the United States software-limited.[594] Some manufacturers deliberately restrict their bill of materials to domestically made components and will change a part specifically to keep the sourcing local.[580] Second-hand and surplus mechanical components carry a quality cost that must be tested for: about a quarter of used solenoid valves bought from surplus dealers were unusable and had to be scrapped after fitting.[618]
Prototyping and small-batch production
Sample requests are often routed through a local distributor who requires the project name, intended volumes, and contact details before releasing parts, and follows up with sales calls.[16] Designing a board entirely from free samples is a known trap: the parts obtained as samples may be obscure devices that cannot be purchased in production quantities, leaving a working prototype that cannot be built, and a vendor can offer an excellent free-sample service while carrying forty-week production lead times.[16][44]
The marginal cost of the first component on a new board is dominated by fixed costs: once the distributor order, board fabrication, and engineering time are counted, putting a single resistor onto a first article runs into thousands of dollars, and turnkey assembly of a first five boards can cost several hundred.[434] Assembly houses require excess components beyond the board count because parts are lost when a reel is loaded and during placement; supplying exactly one reel of one hundred resistors for one hundred boards will not produce one hundred boards.[24] One-time-programmable devices must be programmed before assembly, so they cannot be ordered by the assembler in the normal flow and have to be supplied to the build already programmed.[243] Precision components are held in stock by almost nobody, so a design using them requires the developer to buy up most of the available world supply in advance and carry thousands of dollars of inventory before any revenue arrives.[176] Handing a bill of materials to a local assembler for a box build transfers the sourcing labour as well as the placement work, which for a large bill of materials on a small prototype run is usually cheaper than doing it in house, and a prototype-sized order from a catalogue distributor can be easier than pulling parts off production reels already held in stock.[445][628]
Crowdfunded hardware concentrates these constraints. Long-lead parts must be bought before a campaign opens rather than after it closes; a single reel purchased in advance for about four thousand two hundred dollars is the price of being able to fulfil on the promised schedule.[175] Crowdfunding platforms differ in when cash reaches the creator: some release funds as the target is met, while others hold the money for about two weeks after the campaign ends, which delays part purchasing and is a common cause of late hardware delivery.[184] Setting a crowdfunding price before the board is built and before any component pricing has been negotiated with vendors is a structural way to underprice a product, because the bill of materials is not yet known.[254] A distributor endorsement badge on a crowdfunding campaign reduces to a check that the listed parts can be obtained, and does not cover design, mechanical tooling, or delivery risk.[345] Supply readiness for a hardware launch is assessed by whether suppliers are lined up, the supply chain is in place, cash flow is secure, and payment terms have been agreed, none of which is normally visible to backers.[393]
In cost-dominated designs the sourcing decision is the design: in a robot bill of materials, motors and motor controllers alone account for roughly eighty percent of cost, so switching those two line items to Chinese suppliers produced a two-thousand-dollar reduction even with tariffs applied.[712] Designing to availability inverts the usual order entirely: browse a low-cost distributor sorted by price, assemble a parts budget from what is cheap and stocked, and derive the product from that.[625]