--- title: "Where does a chip travel before it reaches your board?" topic: "Chip journeys" answer: "Often to another continent and back. TI makes the INA240’s wafers in Richardson, Texas, 61 km from Mouser. They are packaged and tested in Kuala Lumpur, so the chip travels about 31,100 km to reach that shelf." url: "https://nearfieldnotes.com/notes/chip-journeys/" site: "Near Field Notes" author: "Michael Walker" note: 9 family: "Logistics" published: "2026-11-03" revised: "2026-11-03" revision: 1 video: "https://nearfieldnotes.com/videos/chip-journeys/" --- Note 9 · Logistics · Rev 1 · [Watch the video · 0:18](https://nearfieldnotes.com/videos/chip-journeys/) Chip journeys # Where does a chip travel before it reaches your board? **Short answer:** Often to another continent and back. TI makes the INA240’s wafers in Richardson, Texas, 61 km from Mouser. They are packaged and tested in Kuala Lumpur, so the chip travels about 31,100 km to reach that shelf. **Note 9’s opening figure: The INA240’s trip.** From the wafer fab to Mouser ![Note 9’s opening figure: The INA240’s trip](https://nearfieldnotes.com/notes/chip-journeys/fig-opening.png) *What the figure shows:* A map of the INA240's trip. TI makes its wafers in Richardson, Texas, 61 km from Mouser in Mansfield. They fly 15,550 km to Kuala Lumpur to be packaged and tested, then 15,570 km back to Mouser: about 31,100 km in all. **Key** - Factory - Distributor - Flight Great-circle distance, stop to stop. One documented route. TI also makes these wafers in South Portland, Maine. ## A chip’s two factories *TI’s INA240 current-sense amp* A chip is made in two factories. A wafer fab builds the circuit many times over on a silicon wafer, layer by layer, and probes each copy. An assembly and test site then cuts the wafer apart, puts each die in a package and tests it again. For the INA240 those two sites are 15,550 km apart. **Fig. 9: What each factory makes.** ![Fig. 9: What each factory makes](https://nearfieldnotes.com/notes/chip-journeys/fig-9.png) *What the figure shows:* Figure 9: what each factory makes. TI's wafer fab in Richardson, Texas builds many copies of the INA240 on one wafer. One die is flown 15,550 km to Kuala Lumpur, where it is wired into an eight-lead TSSOP package and tested. Not to scale. The package is drawn see-through to show the die inside. ## The trip, stop by stop *TI’s INA240* Getting from stop to stop takes 7 to 17 days. Chips are light and costly, so they fly: about 90% of Malaysia’s chip exports go by air. The two factories take 71 to 129 days. **Fig. 9b: The stops, to time.** ![Fig. 9b: The stops, to time](https://nearfieldnotes.com/notes/chip-journeys/fig-9b.png) *What the figure shows:* Figure 9b: the INA240's stops and how long each takes. Richardson, Texas, Wafer fab and probe: 57–87 days. Flight, 15,550 km: 3–7 days. Kuala Lumpur, Malaysia, Assembly and test: 14–42 days. Flight, 15,570 km: 4–10 days. Mansfield, Texas, Mouser, on the shelf. Days at each stop and on each leg. Dark is the low estimate, light runs to the high one. ## Six chips, six trips *One documented route each* The INA240 is not unusual. Here are five more chips, from the RAM on a Raspberry Pi 5 to an NVIDIA GPU. Each follows one documented route from wafer to shelf, and each travels more than 15,000 km. Pick one to follow it. **Note 9’s explorer: Chip journeys.** Micron LPDDR4X on the 8 GB Raspberry Pi 5 ![Note 9’s explorer: Chip journeys](https://nearfieldnotes.com/notes/chip-journeys/fig-explore.png) *What the figure shows:* A map of the chosen chip's route, its stops numbered as in the list beside it. **TI INA240 current-sense amp, stop by stop** 1. 1 **Richardson, Texas** TI’s RFAB Wafer fab and probe 57–87 days Estimate 2. Flight to Kuala Lumpur, 15,550 km 3–7 days Industry figure 3. 2 **Kuala Lumpur, Malaysia** TI Malaysia Assembly and test 14–42 days Estimate 4. Flight to Mansfield, 15,570 km 4–10 days Industry figure 5. 3 **Mansfield, Texas** Mouser On the shelf **Micron LPDDR4X on the 8 GB Raspberry Pi 5, stop by stop** 1. 1 **Taichung, Taiwan** Micron Wafer fab, probe, assembly and test 91–147 days Estimate 2. Flight to Pencoed, 9,980 km 3–7 days Estimate 3. 2 **Pencoed, Wales** Sony UK Technology Center Built onto Pi 5 boards Up to 3 days Estimate 4. Flight to Thief River Falls, 6,200 km 4–8 days Company figure 5. 3 **Thief River Falls, Minnesota** DigiKey On the shelf **NVIDIA B300 GPU, die path, stop by stop** 1. 1 **Phoenix, Arizona** TSMC Arizona Wafer fab and probe 87–126 days Estimate 2. Flight to Tainan, 11,600 km 3–7 days Industry figure 3. 2 **Tainan, Taiwan** TSMC AP2 Micro-bumps added 5–14 days Estimate 4. Truck to Zhunan, 190 km Up to 1 day Estimate 5. 3 **Zhunan, Taiwan** TSMC AP6 CoWoS-L packaging 28–56 days Estimate 6. Truck to Tongluo, 30 km Up to 1 day Estimate 7. 4 **Tongluo, Taiwan** KYEC Test and burn-in 7–21 days Estimate 8. Flight to Fort Worth, 12,460 km 2–6 days Industry figure 9. 5 **Fort Worth, Texas** Wistron Built onto GB300 boards 7–14 days Estimate 10. Truck to Memphis, 710 km 1–3 days Estimate 11. 6 **Memphis, Tennessee** xAI’s Colossus 2, as an example In a data center **ST L298N dual H-bridge, stop by stop** 1. 1 **Ang Mo Kio, Singapore** ST Wafer fab and probe 22–36 days Estimate 2. Flight to Bouskoura, 11,900 km 3–7 days Estimate 3. 2 **Bouskoura, Morocco** ST Assembly and test 14–42 days Estimate 4. Flight to Carson, 9,600 km 3–7 days Estimate 5. 3 **Carson, California** ST’s Los Angeles hub Passes through 6. Truck to Thief River Falls, 2,420 km 3–6 days Estimate 7. 4 **Thief River Falls, Minnesota** DigiKey On the shelf **Infineon IPT015N10N5 100 V MOSFET, stop by stop** 1. 1 **Villach, Austria** Infineon Wafer fab and probe 22–45 days Estimate 2. Flight to Melaka, 9,700 km 3–7 days Estimate 3. 2 **Melaka, Malaysia** Infineon Assembly and test 14–42 days Estimate 4. Truck to Singapore, 210 km 1–2 days Estimate 5. 3 **Singapore** Infineon’s hub Passes through 6. Flight to Mansfield, 15,650 km 3–8 days Company figure 7. 4 **Mansfield, Texas** Mouser On the shelf **Espressif ESP32 in a WROOM-32E module, stop by stop** 1. 1 **Hsinchu, Taiwan** TSMC, site not public Wafers, order to delivery 84–98 days Maker’s figure 2. Flight to Chengdu, 1,810 km 2–5 days Estimate 3. 2 **Chengdu, China** Unisem Packaging and test 28–42 days Maker’s figure 4. Truck to Danyang, 1,500 km 2–4 days Estimate 5. 3 **Danyang, China** Fulian Built into WROOM modules 21–28 days Maker’s figure 6. Flight to Mansfield, 11,860 km 5–10 days Estimate 7. 4 **Mansfield, Texas** Mouser On the shelf Numbers on the map match the list. Each time is labeled with where it comes from: the maker’s own figure, a company figure, an industry figure or our estimate. ## Where the time goes *Wafer start to shelf* Side by side, the six trips split the same way. A fab builds a wafer one layer at a time, at 1 to 1.5 days per layer. A finished wafer takes about 12 weeks, and up to 20 on advanced processes. A simple one like the L298N’s, with about 10 layers, takes about a month. Assembly and test can add 6 weeks more. **Fig. 9c: Six chips.** ![Fig. 9c: Six chips](https://nearfieldnotes.com/notes/chip-journeys/fig-9c.png) *What the figure shows:* Figure 9c: days each chip spends in factories and moving between them. TI INA240 current-sense amp: 71–129 days in factories, 7–17 days moving. Micron LPDDR4X on the 8 GB Raspberry Pi 5: 91–150 days in factories, 7–15 days moving. NVIDIA B300 GPU, die path: 134–231 days in factories, 6–18 days moving. ST L298N dual H-bridge: 36–78 days in factories, 9–20 days moving. Infineon IPT015N10N5 100 V MOSFET: 36–87 days in factories, 7–17 days moving. Espressif ESP32 in a WROOM-32E module: 133–168 days in factories, 9–19 days moving. Days from wafer start to the shelf on one documented route each. Dark is the low estimate, light runs to the high one. Six chips ### The flights take days. The factories take months. **Field marks.** Each chip spends 6 to 20 days moving between stops, and 36 to 231 days in factories. **Look-alike.** [Lead time](https://nearfieldnotes.com/glossary/#g-lead-time) is a different clock. The L298N takes about 45 to 98 days to make and ship, and its factory lead time is 52 weeks. Most of that is waiting for a slot, not making. **Range.** Only Espressif publishes its own times: about 3 months for wafers, 4 to 6 weeks to package and test, and 3 to 4 weeks to build modules. The other times are industry figures, company figures or our estimates. **Habitat.** One documented route per chip, wafer start to shelf. The lead time is DigiKey’s and Mouser’s on 26 Oct 2026. ## One package, four countries *NVIDIA B300* The B300 spends the longest in factories, and it stops at the most of them: five, with three in Taiwan. Its memory and substrate fly in to meet the die there. **Fig. 9d: The B300 comes together.** ![Fig. 9d: The B300 comes together](https://nearfieldnotes.com/notes/chip-journeys/fig-9d.png) *What the figure shows:* Figure 9d: where a B300 package comes together. Wafers made in Phoenix, Arizona fly to Tainan, Taiwan for micro-bumps, then go by truck to Zhunan for CoWoS-L packaging. HBM3E stacks fly in from SK hynix in Icheon, Korea, and the substrate from Ibiden in Ono, Japan. KYEC in Tongluo tests it, and it flies to Wistron in Fort Worth, Texas. One documented route, for a die made in Arizona. TSMC also makes these wafers in Taiwan. The B300’s parts ### One B300 package holds 106 dies **Field marks.** Two GPU dies and eight HBM3E stacks. Each stack is 12 DRAM dies on a base die, so 2 + 8 × 13 = 106. On this route four countries take part: the die comes from the US, the HBM from Korea and the substrate from Japan, and Taiwan puts them together. **Look-alike.** A wafer made in Arizona still flies to Taiwan to be packaged, then back to Texas to go on a board. In all it travels about 25,000 km. **Range.** The HBM is the slow part. It takes over six months from wafer start to a finished stack, against three to four for the GPU die. **Habitat.** GB300 NVL72 racks. Wistron builds the boards in Fort Worth. SK hynix and Micron both supply the HBM. ## Where was yours made? *Infineon’s IPT015N10N5* Every chip on this page follows one documented route. Most parts can come from more than one fab and more than one back end, so the reel on your bench may have come another way. Infineon has qualified six routes for this one MOSFET. **Fig. 9e: One part number, six routes.** ![Fig. 9e: One part number, six routes](https://nearfieldnotes.com/notes/chip-journeys/fig-9e.png) *What the figure shows:* Figure 9e: Infineon's IPT015N10N5 can come from three wafer lines, Villach 200 mm with lot code VE, Villach 300 mm (VF) and Dresden 300 mm (ZF), and two assembly and test sites, Melaka in Malaysia and Tongfu in China: six routes for one part number. The wafer lines and back ends Infineon has qualified for this 100 V MOSFET. The wafer lot number on the reel’s label starts with its line’s code. Key to chip journeys ### Where was my chip made? 1. 1a The lot code on the reel or the marking on the part The fab that made yours, or its country 2. 1b The maker’s change notices for the part number Every site it can come from 3. 1c The distributor’s listing The countries only Micron prints the wafer fab’s and the assembly site’s countries in its markings. The 3IC77 on a 2023 Pi 5’s RAM reads Taiwan for both. #### Conditions - **Routes:** One documented route per part, from the makers’ change notices, filings and distributor listings - **Distance:** Great circle, stop to stop, on a 6,371 km sphere. Real flights and trucks go farther. - **Times:** Days at each stop and on each leg, as a low and a high estimate - **Evidence:** Maker’s figure: the maker gives it for this part. Company figure: the company gives it for this step. Industry figure: typical published times. Estimate: our arithmetic from those. - **Checked:** 26 Oct 2026 - **Map:** Natural Earth land, 1:110m and 1:50m Sourced, not tracked. Every site comes from a public source, and most times are estimates. #### Limits - Most parts have more than one qualified fab and back end. Your reel may have come another way. - Only Espressif publishes its own times. Every other time here is an industry figure, a company figure or our estimate. - The B300’s route ends at xAI’s Colossus 2 as an example of a GB300 site. No source ties these boards to it. #### Files - data/routes.py Every stop, leg and time range, with its evidence - research/The sources behind each chip, stop by stop #### Sources - [TI PCN 20211220001.1, RFAB added for the INA240](https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/4884/PCN20211220001.1.pdf) - [Mouser, INA240A1PWR](https://www.mouser.com/ProductDetail/595-INA240A1PWR) - [Raspberry Pi, Introducing Raspberry Pi 5](https://www.raspberrypi.com/news/introducing-raspberry-pi-5/) - [ST PCN IPD-IPC/12/7320, the L298’s wafers in Ang Mo Kio](https://www.anglia.com/registration/pcn_ptn/docs/pcn/PCN%20IPD-IPC_12_7320.pdf) - [ST PCN APM-IPC/07/2731, Multiwatt assembly in Bouskoura](https://www.anglia.com/registration/pcn_ptn/docs/pcn/2731.pdf) - [ST PCN CRP/19/11361, the Los Angeles hub](https://www.anglia.com/registration/pcn_ptn/docs/pcn/st_PCN%20CRP_19_11361.pdf) - [Infineon PCN 2021-037-A, wafer lines and lot codes](https://media.futureelectronics.com/PCN/86116_SPCN.PDF) - [Infineon PCN 2023-001-A, Melaka and Tongfu](https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/5019/2023-001-A.pdf) - [Espressif’s 2025 reply to the Shanghai exchange, its own times](https://file.finance.qq.com/finance/hs/pdf/2025/07/11/1224144510.PDF) - [NVIDIA, Blackwell in Arizona and boards in Texas](https://blogs.nvidia.com/blog/nvidia-and-partners-build-in-america-for-america/) - [TechInsights, the HBM in an HGX B300](https://www.techinsights.com/blog/high-bandwidth-memory-package-NVIDIA-HGX-B300) - [TrendForce, HBM takes over two quarters](https://www.trendforce.com/presscenter/news/20240318-12081.html) - [TrendForce, SK hynix’s 12-high HBM3E for the GB300](https://www.trendforce.com/news/2025/03/18/news-sk-hynix-reportedly-locks-in-exclusive-12h-hbm3e-deal-for-nvidias-gb300/) - [NVIDIA, substrate deals with IBIDEN, Samsung Electro-Mechanics and Unimicron](https://nvidianews.nvidia.com/news/nvidia-and-supply-chain-partners-expand-capacity-to-accelerate-ai-infrastructure-buildout) - [BCG and SIA, fab cycle times](https://www.semiconductors.org/wp-content/uploads/2021/05/BCG-x-SIA-Strengthening-the-Global-Semiconductor-Value-Chain-April-2021_1.pdf) - [SIA, why making a chip takes time](https://www.semiconductors.org/chipmakers-are-ramping-up-production-to-address-semiconductor-shortage-heres-why-that-takes-time/) - [Semiconductor Engineering, days per layer in a fab](https://semiengineering.com/battling-fab-cycle-times/) - [MIDA, 90% of Malaysia’s chip exports go by air](https://www.mida.gov.my/mida-news/govt-to-boost-chips-export-logistics/) - Micron CSN-11, reading the marking on a Micron FBGA - [Natural Earth](https://www.naturalearthdata.com/) What should the next note be? [Ask for one](https://nearfieldnotes.com/#ask-for-a-note) --- Source: https://nearfieldnotes.com/notes/chip-journeys/ Cite as: Michael Walker, “Chip journeys”, Near Field Notes, Note 9, Rev 1, 3 Nov 2026. https://nearfieldnotes.com/notes/chip-journeys/ Figures: free to reuse under CC BY 4.0 (https://nearfieldnotes.com/license/). Each one's image is linked under its name. Credit Michael Walker, Near Field Notes, and link the figure: this address plus #fig- and its number, like https://nearfieldnotes.com/notes/chip-journeys/#fig-9.