Technology giants are reducing their dependence on China, but the evidence does not show a broad exit. As of August 16, 2026, Apple, Microsoft, Meta and semiconductor companies are adding factories, cloud capacity, data centers and chip operations in the United States, India and other countries while still relying on Chinese suppliers, customers, subsidiaries or manufacturing partners. The emerging model is “China-plus-many”: spread critical activity across several locations without abandoning China’s scale or market.
“Pivot away” covers several different decisions
A company can reduce China exposure in one layer while retaining or increasing it in another. Readers should distinguish:
As an Amazon Associate I earn from qualifying purchases.
- Final assembly: where phones, computers, servers and networking equipment are put together.
- Components: batteries, displays, glass, printed-circuit boards, connectors and camera modules.
- Semiconductors: wafer fabrication, memory, packaging and testing.
- AI infrastructure: data centers, cloud regions, GPUs, networking, cooling and electricity.
- Research and engineering: laboratories, software, hardware design and technical support.
- Commercial exposure: sales, cloud services, advertising, app distribution and Chinese enterprise customers.
- Capital and ownership: subsidiaries, joint ventures and local partnerships.
Moving assembly to India does not mean a product is free of Chinese inputs. A US-made chip can still depend on Asian wafers, chemicals, equipment or packaging. A company can reduce Chinese manufacturing while continuing to sell heavily to Chinese customers.
Why companies are spreading activity across more countries
Export controls and national-security rules
US restrictions cover advanced semiconductors, AI systems, chipmaking equipment and related software. The Bureau of Industry and Security’s Export Administration Regulations describe licensing rules affecting advanced-computing activity: BIS EAR §740. Nvidia’s filings show the business consequences: products may require region-specific designs or licenses, inventory can become unusable, and companies risk penalties if goods are diverted through third countries.
#1 Best Overall
Tariffs and policy uncertainty
Tariffs raise the cost of imported products and parts. Even when phones or other goods receive temporary exemptions, companies cannot treat those exemptions as permanent. Building another production route is an insurance policy against a sudden rule change.
Concentration lessons from disruption
Lockdowns, port closures, labor shortages and factory shutdowns demonstrated the cost of depending on one manufacturing region. Duplicating tooling, suppliers and logistics is expensive, but a second source can prevent a single disruption from stopping a global product launch.
Competition and customer pressure
Chinese companies are developing domestic chips, cloud systems, devices and manufacturing equipment. Governments, defense buyers, banks and healthcare organizations also increasingly ask where hardware, software and data are produced. Those demands encourage regional supply chains even when China remains the most efficient production base.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Apple: the clearest “China-plus-many” example
Apple is adding US production without replacing its Asian manufacturing network. Its American Manufacturing Program includes Broadcom chips in Colorado, Corning glass in Kentucky, GlobalFoundries capabilities in New York, Amkor packaging and testing in Arizona, TSMC-made chips from Arizona, and Mac mini production in Houston. Apple has also named Bosch, Cirrus Logic, TDK and Qnity Electronics as program partners: the 2025 program announcement and the 2026 partner update.
On July 8, 2026, Apple said its Broadcom commitment would exceed $30 billion and produce more than 15 billion US-made chips: Apple’s announcement. In February 2026, Apple said Mac mini production in Houston would begin later that year and that it expected to buy well over 100 million advanced chips from TSMC’s Arizona facility during 2026. Apple described Amkor’s Arizona project as a $7 billion facility and its first and largest customer: the Mac mini announcement.
Those are important additions, not proof of a completed withdrawal. Apple’s 2025 Form 10-K says a significant majority of hardware manufacturing remains with Asian outsourcing partners, including mainland China, India, Japan, South Korea, Taiwan and Vietnam, with final assembly of substantially all hardware products handled primarily by Asian partners: Apple’s 2025 filing.
The practical interpretation is selective reshoring, more Indian and Southeast Asian assembly, and continued Chinese production where the supplier network and economics remain difficult to duplicate.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Nvidia shows why semiconductors follow a different path
Nvidia is a chip designer rather than a conventional factory operator. Its China exposure runs through Chinese data-center demand, external foundries, advanced packaging, memory, server builders, export licenses and the risk that products are diverted through intermediaries.
In its fiscal 2026 first-quarter filing, Nvidia disclosed a $4.5 billion charge related to H20 inventory and purchase obligations after export restrictions weakened demand: Nvidia’s filing. In a later filing, the company said US authorities beginning in February 2026 granted licenses for small quantities of H200 products to specified Chinese customers, while China’s antitrust regulators had made a preliminary finding concerning Nvidia’s compliance with US controls and its Mellanox acquisition: Nvidia’s April 2026 filing.
Nvidia therefore illustrates forced decoupling at the leading edge, not a clean exit. Selling fewer advanced products in China may support US strategic goals, but it also sacrifices revenue and can give Chinese competitors an incentive to build substitutes. Product cycles are long; export rules can change faster than a chip can be designed, manufactured and qualified.
Microsoft and Meta are moving AI infrastructure, not simply factories
Microsoft’s India build-out
Microsoft announced a $17.5 billion investment in India from calendar years 2026 through 2029 for data centers, cloud and AI infrastructure, skilling and operations: Microsoft’s announcement. Reuters reported in August 2026 that Microsoft launched its largest India data-center hub in Hyderabad, with early customers including Adani Group and HDFC Bank: the report.
Microsoft’s own corporate information still lists China, India and Vietnam among subsidiary locations: Microsoft’s company information. India’s expansion reflects market growth, talent, local data needs and infrastructure demand as well as diversification. It does not establish a China exit.
Rank #4
Meta’s 168-megawatt Indian data center
Meta and Reliance announced an AI-enabled data center in Jamnagar, Gujarat, with initial capacity of 168 megawatts. Reliance will build the facility and lease it to Meta; the project can scale and is planned to use renewable energy and desalinated seawater cooling: Meta’s data-center description. Meta also announced nearly 1 gigawatt of renewable-energy agreements in India: the partnership announcement.
This is primarily a compute-location decision: put capacity near users, secure power and cooling, and serve a large non-Chinese market. Data centers cannot be relocated as easily as a factory order because they depend on electricity, connectivity, land, permits and data-residency rules.
Where the new capacity is going
| Location | Likely role in diversification | Important qualification |
|---|---|---|
| United States | Chip fabrication, advanced packaging, specialized components, AI data centers and high-value manufacturing | Higher costs and thinner supplier density make a complete replacement for China unrealistic. |
| India | Smartphone assembly, cloud and AI infrastructure, engineering, and semiconductor assembly and testing | Micron reported commercial shipments from its Gujarat facility and a 2026 production ramp: Micron’s filing. |
| Vietnam | Consumer electronics and component diversification | It is part of a Southeast Asian network, not a stand-alone substitute for China. |
| Japan and South Korea | Advanced components, memory and specialized manufacturing | Capacity remains internationally interconnected. |
| Taiwan | Leading-edge fabrication and critical chip production | Taiwan is a separate manufacturing and geopolitical node; a Taiwan crisis would create a distinct shock. |
| Mexico | North American manufacturing and final assembly | It can shorten regional logistics but still relies on imported components. |
| China | Large-scale assembly, batteries, displays, materials, tooling, logistics and domestic demand | China remains embedded in many supposedly relocated products. |
Why China remains difficult to replace
China combines dense component networks, experienced electronics labor, rapid tooling and prototyping, mature ports, established supplier relationships, engineers and a large domestic market. Moving a simple assembly step can be faster than relocating the specialized parts, testing, tooling and upstream materials that make the product economical.
That is why a “China-plus-one” strategy often changes risk rather than eliminating it. India, Vietnam or Mexico can become new concentration points, while a relocated factory may continue buying critical Chinese inputs. “Made in” labels generally describe final assembly, not the origin of every component.
How to tell whether a company is genuinely pivoting
- Check capital allocation: Is money committed to a facility, supplier or data center, rather than only discussed?
- Check operating status: Is the project announced, under construction, in pilot production or shipping commercial quantities?
- Check criticality: Does the new site make a strategic product or a marginal line?
- Check measured dependence: Has China’s share of production, sourcing, revenue or engineering actually declined?
- Check irreversibility: Are local suppliers, trained workers and long-lived assets being established?
Announcements such as Apple’s $600 billion US commitment are not the same as operational capacity. The status and volume of each project matter.
The trade-offs for businesses and consumers
- Resilience versus cost: Duplicate tooling, qualification, logistics and compliance raise expenses.
- Security versus market access: Controls can protect sensitive technology while shrinking Chinese sales and encouraging substitution.
- Speed versus quality: Mature products move more easily than advanced products dependent on tightly integrated suppliers.
- Localization versus fragmentation: Regional rules can produce different product versions, inventories and compliance systems.
- Short-term prices: Diversification may improve continuity but can increase manufacturing and compliance costs before efficiencies emerge.
Businesses evaluating suppliers should ask whether a proposed alternative has local component depth, skilled labor, reliable power and water, transport capacity, export-control safeguards and a plan for tariff or political changes. Consumers should expect more complicated origin labels and, in some categories, region-specific products rather than an immediate end to Chinese content.
Google, Amazon and the limits of broad claims
Google and Amazon participate in the wider infrastructure-diversification trend, but the public evidence is less uniform than Apple’s explicit manufacturing announcements or Microsoft’s and Meta’s India projects. BIS rules identify Google, Amazon, Microsoft, Apple, Meta, Nvidia and others in connection with advanced-computing authorizations, showing that hyperscalers are affected by the export regime; that source does not prove any one company is withdrawing from China: BIS EAR §740.
Recommended Free Tools
An IMD analysis describes Amazon, Google and Microsoft as developing more internally controlled technology stacks and expanding infrastructure in places including the United States, India and Vietnam while retaining operations elsewhere. It is useful context, not a company-specific exit announcement: IMD’s analysis.
What this means for the technology economy
The result is a more distributed but still interdependent system. Strategic chips, AI compute and selected assembly are spreading across North America, India and Asia. China retains major advantages in components, manufacturing scale, logistics and domestic demand. Companies are accepting redundancy costs to reduce the chance that one government, lockdown, tariff or transport failure can disrupt the entire chain.
The most accurate description is therefore not “technology giants are leaving China.” It is that they are building a China-plus-many model: less concentration, more regional capacity and continued commercial and technical links to China.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
