The chip industry has a Suzuki problem.
The chip industry spent roughly $4 trillion on capex in the decade after 2013 and still ran out of microcontrollers for cars in 2021. A convenience-store chain in Japan solved the same structural problem, volatile demand, complex distribution, perishable inventory, in 1978, with a cash register and a delivery truck. Toshifumi Suzuki built Seven-Eleven Japan on one conviction: the enemy of supply-chain efficiency is not volume, it is information latency.
By the mid-1990s a 7-Eleven Japan store took 3-5 deliveries a day and turned inventory 55-65 times a year, four to five times the Western benchmark. The proprietary POS system, which senior executives opposed at the time, fed every SKU's demand back to every supplier in near real time, so the bull-whip effect was structurally suppressed: the signal never got stale. The semiconductor crisis of 2020-2022 was a textbook bull-whip, automakers cancelled, fabs reallocated, demand snapped back, and certain auto MCU lead times hit 52 weeks. Seven-Eleven solved this in rice balls. The chip industry is still working on it in wafers.
7-Eleven Japan POS the 1980s register that captured every SKU sale and fed it back upstream within the day.
Suppliers & makers respond to that signal with small, frequent deliveries instead of big infrequent ones.
TSMC's Joint Business Plan the chip-world analogue: tier-1 customers inside a 12-18 month frozen-capacity loop.
Fabless & OEM buyers either inside that information loop (allocation) or on a 90-day forecast (leftovers).
Automakers the cautionary tale, who treated $0.40 MCUs as interchangeable commodities in 2020.
1. A sale is rung up; the POS captures demand at the SKU level in real time.
2. That signal reaches every supplier the same day, before it can go stale.
3. Suppliers ship smaller, more frequent loads, so inventory turns 55-65x a year.
4. Fresh signals suppress the bull-whip: small swings never amplify into big upstream ones.
5. In chips, the same loop is a foundry JBP; latency, not volume, decides who gets allocation.
Neutral process view. A methodology comparison, not an endorsement of any company or stock.
Pay for information instead of inventory.
The 2020-2022 shortage was a bull-whip event, precise to the textbook. Automakers cancelled chip orders in Q2 2020 when plants shut; fab planners read the cancellations as real demand and moved capacity to consumer electronics; when auto demand returned in Q3, the chips were gone. By 2021, some automotive MCUs ran 52-week lead times, and the industry wrote down $10B+ of excess inventory in 2023 when the phantom demand corrected. The 90-day buffer didn't hedge the shortage. It created the bust.
Suzuki's lesson isn't hold less inventory, it's pay for information instead of inventory. Higher delivery frequency raised per-unit logistics cost but slashed working capital and erased write-down risk. The chip-world version exists: TSMC's Joint Business Plan loops tier-1 customers into 12-18 month capacity planning, and those customers hold an estimated 30-40% less safety stock on leading-edge parts. That gap is the financial value of the information partnership. Apple never ran out of M1 chips; Ford halted F-150 lines for six weeks. The split was information proximity, not capex.
Companies that bought information proximity: the JBP insiders (Apple, NVIDIA and peers) who held less stock and never starved. Foundries and OSATs that monetize the loop with priority allocation. And any buyer that already reports demand-signal latency as a managed metric.
Transactional buyers on 90-day forecasts and contractual minimums, structurally last in line. Automakers that treated $0.40 MCUs as ambient goods. And balance sheets still carrying shortage-era buffers that quietly became the next write-down.
The startup opening. The opening is in the latency layer: demand-signal-latency analytics that turn a fuzzy idea into a board KPI, perishability-based component classification tooling, and JBP-style information brokerage for the mid-market buyers too small to sit at the foundry table alone. The method is 40 years old and still mostly uncopied, which is exactly the opportunity.
Bull-whip effect: how small swings in end demand amplify into huge inventory swings upstream. Suzuki's POS suppressed it; the 2021 chip crisis was a textbook case of it.
POS: point-of-sale, the register data 7-Eleven used to capture demand at the SKU level in real time and feed it straight back to suppliers.
JBP: Joint Business Plan, TSMC's framework that loops tier-1 customers into 12-18 month capacity planning. The chip-world version of the 7-Eleven information loop.
Inventory turns: how many times a year a company sells through and replaces its stock. 7-Eleven Japan ran 55-65; Western peers, 12-18.
Safety stock: buffer inventory held against demand surprises. Information proximity lets JBP insiders hold 30-40% less of it on leading-edge parts.
Demand-signal latency: how many days old the demand data is that your capacity planning runs on. Above 30 days, no amount of safety stock fixes it.
Corrections & coffee: [email protected]
