Nikkei Asia reported on 28 October 2026 that leading PC makers, including HP of the United States and Asus and Acer of Taiwan, are combating soaring component costs by shifting some production from Thailand and Vietnam back to China and using a new generation of contract manufacturers. The report’s headline ties the move to AI-driven cost hikes.
The same week, SK hynix said it expects demand for memory to exceed supply through 2030, and Nikkei Asia reported that Fujifilm will raise its magnetic-tape production capacity fivefold to meet AI-driven demand. As we read it, the strain from AI demand has now been reported at four points of the hardware chain: memory, storage, circuit boards and final assembly.
Short explainer · AI supply squeeze
A vertical news Short narrated from the published headlines and opening lines of this piece and four related FourWeekMBA pieces on AI-driven supply shortages.
Business Pill · The memory wall
An evergreen Business Engineer explainer on why memory, not only compute, limits AI performance. It explains the general idea only and does not discuss the PC makers or the other companies in this piece.
The key insight: As we read it, the move reverses part of the China+1 shift: Nikkei Asia says HP and Asus built facilities in Vietnam and Thailand to diversify away from China, and are now returning some production there to combat soaring component costs.
Nikkei Asia said HP and Asus are among PC makers returning some production to China after building facilities in Vietnam and Thailand as part of a “China+1” diversification strategy. China+1 is the practice of adding factories outside China to reduce dependence on it.
The rest of the Nikkei Asia article is behind a paywall. We have not read which plants, product lines or volumes are moving, or which contract manufacturers are involved, and HP, Asus and Acer have not commented in anything we have read.
Nikkei Asia reported on 4 August 2026 that the same three PC makers, HP, Asus and Acer, had started to use small amounts of chips from China’s ChangXin Memory Technologies amid an unprecedented memory shortage fueled by demand for AI infrastructure, while being careful to avoid angering their top suppliers Micron, Samsung and SK Hynix.
How far ahead supply is short or spoken for, as each company states it: SK hynix expects demand to exceed supply through 2030 and has long-term agreements covering more than 50% of its 2027-2028 capacity (results, 28 Oct 2026); Seagate says the vast majority of its Data Center nearline supply is allocated through calendar 2028, with some cloud supply agreements through calendar 2030 (results, 27 Oct 2026). ‘Now’ is the date each company published its results.
Memory: Short Through 2030
SK hynix said on 28 October 2026 that supply is not keeping pace with market growth: “Given technical difficulty of leading-edge processes and the long lead time required to build new production facilities, it expects the structure in which demand exceeds supply to persist through 2030.”
The company said it has signed long-term agreements with 14 major customers, including leading global cloud service providers, and that those agreements cover more than 50% of its 2027 to 2028 production capacity. It said the contract terms are typically five years.
The Register reported that on the earnings call, chief financial officer Kim Woo-hyun said “shipment volumes for PCs and mobile devices are expected to decrease due to cost burdens from rising memory prices and supply shortages.”
Storage: Disks Allocated, Tape Rebuilt
Fujifilm plans to raise its magnetic-tape production capacity fivefold to meet AI-driven demand, its largest investment in the medium in roughly 20 years, Nikkei Asia reported on 28 October 2026 (Nikkei has learned; no Fujifilm announcement found). Today = 1 is our framing of ‘fivefold’.
Seagate said on 27 October 2026 that the vast majority of its Data Center nearline supply is allocated through calendar 2028, with some cloud supply agreements extending through calendar 2030. Nearline drives are the high-capacity hard disks that cloud data centers use for bulk storage.
Nikkei Asia reported that Fujifilm will raise production capacity for magnetic tape fivefold to meet AI-driven demand, marking the company’s largest investment in the medium in roughly 20 years. The report is based on what Nikkei says it has learned; we have not seen a Fujifilm announcement. Nikkei Asia notes that tape is widely used by data center operators, financial institutions and research organizations for data backup.
Circuit Boards: A Commodity Becomes a 2027 Worry
In a separate report on 27 October 2026, Nikkei Asia said that as executives at PC makers, including HP and Dell, map out their supply needs for 2027, “a mundane component has emerged as an unlikely concern.” The report’s summary says printed circuit boards, once commoditized, have now become essential AI building blocks.
Canon put a figure on the same constraint. In its third-quarter results on 27 October 2026, it said that since August, shortages of printed circuit boards and the electronic components used in them have emerged as a result of growing AI-related demand. Canon raised the estimated full-year sales impact of production delays in its medical business to approximately 90 billion yen, from approximately 45 billion yen, and listed the board shortages among the causes alongside a system update and a fire at a supplier.
Connecting the Coverage
This piece extends the AI squeeze that hit Canon as SK hynix booked over half of 2027-28, published on 28 October, which covered Canon’s board shortages and SK hynix’s long-term agreements. What’s new here is the assembly step: PC makers moving some production back to China, plus the Fujifilm tape plan.
As we read it, this week’s reports place the strain at four points of the hardware chain: memory chips, storage, the circuit boards inside devices and the factories where PCs are assembled.
SK hynix said long-term agreements with 14 major customers, including leading global cloud service providers, cover more than 50% of its 2027 to 2028 production capacity, and The Register reported that its CFO expects PC and mobile shipment volumes to decrease.
Canon said board shortages caused by growing AI-related demand have hit production since August, and Nikkei Asia said printed circuit boards have emerged as a concern for PC makers including HP and Dell as they plan for 2027.
Kim Woo-hyun, SK hynix CFO, on the earnings call, as reported by The Register
“shipment volumes for PCs and mobile devices are expected to decrease due to cost burdens from rising memory prices and supply shortages”
Three Implications
FOR PC MAKERS Nikkei Asia reports that HP, Asus and Acer are shifting some production from Thailand and Vietnam back to China and using a new generation of contract manufacturers to combat soaring component costs.
FOR MEMORY BUYERS SK hynix says it expects demand to exceed supply through 2030 and that its long-term agreements with customers typically run five years.
FOR STORAGE Seagate says the vast majority of its Data Center nearline supply is allocated through calendar 2028, and Nikkei Asia reports that Fujifilm will raise tape capacity fivefold.
The framework reads the AI economy as a layered industrial stack and notes that “at any given moment the binding constraint sits at a different layer than it did six months ago.”
This week’s reports place AI-driven shortages in memory, storage, circuit boards and PC assembly. The framework was published in June 2026 and does not discuss these companies.
What Is Not Established
We read only the opening paragraphs and summaries of the four Nikkei Asia reports cited here, including the August report on CXMT chips; the rest is paywalled. We do not know how much PC production is moving to China, when, or at what cost, and we did not find a statement from HP, Asus, Acer or Dell.
The Fujifilm plan is reported by Nikkei Asia, not announced by Fujifilm in anything we found; we do not know its cost or timing. SK hynix’s 2030 outlook and Seagate’s allocation windows are the companies’ own expectations.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Nikkei Asia reports that HP, Asus and Acer are shifting some PC production from Thailand and Vietnam back to China to combat soaring component costs, in the same week SK hynix said it expects memory demand to exceed supply through 2030 and Nikkei Asia reported Fujifilm’s plan for fivefold tape capacity.
A note on sourcing. The PC production shift, the circuit-board report, the Fujifilm plan and the August CXMT report come from Nikkei Asia reports we could read only to their opening paragraphs and summaries. The memory outlook comes from SK hynix’s results release and The Register’s report of its earnings call; storage allocation from Seagate’s results slides; the board shortage figure from Canon’s results presentation. The chart shows each company’s stated windows, not a forecast of ours.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
