The OECD’s Warning Is Already on the Factory Floor

The OECD’s March 2025 interim outlook did more than trim its growth forecast. It explicitly warned that proliferating trade barriers and broader fragmentation could weaken business investment more than expected, especially in tightly interlinked supply chains (OECD Economic Outlook, Interim Report March 2025). That matters because semiconductor, electronics, and industrial technology supply chains are now governed less by textbook comparative advantage than by an increasingly expensive mix of tariff risk, subsidy competition, export controls, and political timing.

The usual trade narrative asks how many goods will cross borders. That is now too narrow. The more consequential question is where boards approve the next fabrication plant, advanced packaging line, equipment cluster, or electronics assembly campus. The answer is increasingly shaped by uncertainty itself. The WTO warned in April 2025 that world merchandise trade volume could contract by 0.2% in 2025 under then-current conditions, and by 1.5% if tariff tensions worsened and policy uncertainty spilled over more broadly (WTO Global Trade Outlook and Statistics, April 16, 2025). But for technology supply chains, uncertainty does not merely suppress trade. It changes the geography of capital expenditure.

That is why the OECD downgrade should be read as a factory-siting signal. In semiconductors, the lead times are long, the assets are immobile, and the economics are deeply dependent on policy credibility. A fab can take years to build and billions to finance; once committed, it cannot be redeployed like inventories. Tariff uncertainty therefore acts less like a temporary tax than like a distortion in long-horizon investment calculus, pushing companies to overbuild in subsidized locations, delay projects in contested ones, and duplicate capacity where governments offer protection against the next policy shock.

Tariff Uncertainty Has Become a Capex Variable, Not Just a Trade Cost

The strongest evidence that policy volatility is now affecting investment decisions comes not from speeches but from central-bank fieldwork and trade data. The Federal Reserve’s April 2025 Beige Book said trade policy was a pervasive theme across districts and reported that economic uncertainty, particularly surrounding tariffs, had risen considerably (Federal Reserve Beige Book, April 23, 2025). By June, the Beige Book cited a Dallas Fed survey showing that 44% of respondents expected lower capital spending, while 29% expected a drop in production, with tariff uncertainty playing a central role (Federal Reserve Beige Book, June 4, 2025).

The front-loading data tell the same story from another angle. In March 2025, the U.S. trade deficit hit a record $140.5 billion as businesses and consumers rushed to import ahead of expected tariff changes; imports rose to nearly $419 billion while exports totaled about $278.5 billion (AP News). This was not ordinary demand strength. It was precautionary behavior induced by unstable trade rules. For electronics and chip-adjacent products, that kind of stockpiling is a short-term hedge. The long-term hedge is relocation.

This distinction matters. Inventories can cushion a quarter; fabs and assembly plants are meant to cushion a decade. McKinsey noted in 2025 that semiconductor companies and downstream manufacturers were assessing tariffs not only on chips themselves but across the many stages of the supply chain, from manufacturing equipment to devices containing semiconductors (McKinsey). Once firms begin modeling multiple tariff points across an ecosystem, the investment question shifts from “What is today’s duty rate?” to “Which jurisdiction offers the most durable operating assumptions?”

That is why tariff uncertainty can be more powerful than tariffs already in force. A known tariff can be priced, passed through, or absorbed. An unknown tariff forces firms to add contingency layers: extra capacity in a second country, more packaging options, more local content, more warehousing, more legal structuring, and more lobbying. All of that raises the fixed cost of staying global in the old way.

Case One: TSMC’s Arizona Buildout Shows How Policy Risk Pulls Advanced Capacity

No case better illustrates the new geography of semiconductor investment than TSMC’s U.S. expansion. On March 4, 2025, TSMC announced it would expand its U.S. investment by an additional $100 billion, bringing its total planned investment in the United States to $165 billion (TSMC press release). The package includes three additional fabrication plants, two advanced packaging facilities, and an R&D center in Arizona (AP News).

This is not merely an industrial-policy success story. It is also a risk-premium story. Tariff threats on semiconductors and the wider uncertainty surrounding U.S. trade policy changed the strategic value of producing on American soil. The Peterson Institute for International Economics argued that tariffs alone would not economically justify a buildout of this magnitude, implying that subsidies and political pressure are doing much of the work (PIIE). That is precisely the point: when trade policy becomes unpredictable, companies invest not only for efficiency but for immunity.

Arizona is becoming more than a fab site. It is being designed as a cluster. TSMC said its first Arizona fab has already entered production, and the enlarged footprint is intended to support a broader U.S.-based advanced manufacturing ecosystem (TSMC press release). The addition of advanced packaging is especially important. Packaging is the stage that integrates chips into usable systems, and it has become strategically significant for AI and high-performance computing. By bundling fabs, packaging, and R&D in one jurisdiction, TSMC is not just relocating production; it is shortening exposure to cross-border disruption.

Yet this also shows the cost of the new order. Capital that might once have been allocated according to pure production economics is now being steered by the need to hedge against future tariff escalation. That does not necessarily make the supply chain safer everywhere. It may simply make it more duplicated, more subsidized, and more expensive.

Case Two: Malaysia Shows How Uncertainty Can Freeze the “China+1” Dividend

If Arizona represents the pull of protected onshoring, Malaysia represents the friction facing countries that hoped to benefit from “China+1” diversification. Malaysia launched its National Semiconductor Strategy in 2024 with at least RM25 billion in fiscal support to deepen its role in the chip supply chain (MIDA). By July 2025, Prime Minister Anwar Ibrahim said the strategy had already secured more than RM63 billion in investments as of March 2025, including RM58 billion from foreign sources (Malay Mail).

Those are large numbers, and they confirm that diversification away from a single-country model remains real. But they also obscure a more uncomfortable truth: tariff uncertainty can interrupt even the beneficiaries of relocation. A Reuters-reported industry survey cited by regional coverage found that 65% of Malaysian semiconductor firms expected negative impacts from U.S. tariffs, while 74% worried about declining investment attractiveness (Caixin Global). In other words, the same country positioned as an alternative manufacturing base still struggles to convert geopolitical opportunity into stable long-term capex.

This is the underrated consequence of tariff uncertainty. It does not simply redirect investment from one country to another; it can also force “wait-and-see” behavior across the second-tier destinations that were supposed to gain. Malaysia hosts major facilities tied to Intel, GlobalFoundries, and Infineon, and remains central to assembly, testing, and packaging. But if firms fear that today’s tariff exemption becomes tomorrow’s sector-specific penalty, they may postpone expansions even where the current business case looks favorable.

Bank Negara Malaysia has acknowledged this wider vulnerability. During discussions around the country’s 2025 outlook, the central bank warned of downside risks from higher trade restrictions even while arguing Malaysia’s diversified export base offers some cushion (The Star, citing Bank Negara Malaysia). The lesson is that relocation works best under credible rules. When rules are provisional, relocation becomes partial, hedged, and slower than policy slogans suggest.

Case Three: Vietnam’s Electronics Surge Reveals Relocation Without Full Security

Vietnam is often cited as the clearest winner from supply-chain relocation, but the 2025 data show both the scale of that shift and its limits. In 2024, Vietnam’s electronics industry recorded export turnover of $126.5 billion, accounting for about one-third of the country’s total export value, according to official statistics cited by Vietnam News from the General Statistics Office (Vietnam News). In 2025, preliminary customs data indicated exports of computers, electronic products, and components could reach $107.75 billion, up 48.4% from 2024 and equivalent to about 23% of total export turnover (Vietnam.vn).

These figures confirm that supply chain relocation is real, measurable, and concentrated in electronics. But Vietnam’s exposure also shows why export growth is not the same as investment security. The Ministry of Industry and Trade said in 2025 that it had not yet revised its export growth target even after new U.S. tariff announcements, while continuing talks around the treatment of Vietnamese goods including electronics (VOV World). That is a revealing stance: production has moved, but tariff terms remain unsettled.

The vulnerability lies in the structure of the model. Vietnam has become deeply integrated into electronics assembly and component trade, but much of that integration depends on externally owned production networks and imported inputs. When trade rules change abruptly, assembly hubs face a double risk: pressure on final exports and uncertainty over intermediate sourcing. The OECD’s warning about tightly interlinked supply chains applies here directly (OECD Economic Outlook, Interim Report March 2025).

So Vietnam offers a crucial correction to simplistic relocation narratives. Tariff uncertainty does not just move production from China to Southeast Asia. It can also turn Southeast Asia into a provisional platform unless governments deepen domestic supplier bases, upgrade logistics, and negotiate more durable market access. A relocated assembly line is not the same thing as a secure industrial ecosystem.

Case Four: Delays in Europe and the U.S. Show That Subsidies Cannot Eliminate Strategic Hesitation

The reordering of technology supply chains is not only about new announcements; it is also about delayed or reconsidered projects. Intel’s Ohio project offers one example. Reuters-reported coverage in March 2025 said Intel’s first new fab there, once expected to start production in 2025, was now expected to begin in 2030, reflecting weaker demand and a tougher capital environment (Al Arabiya, citing Reuters). The number matters: a five-year shift in first production is not a scheduling footnote. It is evidence that even with strategic urgency and public support, companies still stagger spending when the demand-policy equation is unclear.

Europe provides a second example. In April 2025, the European Court of Auditors concluded that the EU’s microchip strategy had made reasonable progress but that the Chips Act was “very unlikely” to be sufficient to achieve its overly ambitious objectives (Official Journal notice referencing ECA Special Report 12/2025). That conclusion landed just as Europe was trying to present itself as a more resilient semiconductor location. The problem is not only the scale of subsidies; it is that the global investment contest is now being conducted amid competing tariff threats, national-security screening, and changing industrial priorities.

South Korea’s response also underscores the same point. In April 2025, Seoul expanded its support package for the semiconductor industry to 33 trillion won, about $23 billion, explicitly citing the uncertainties created by U.S. tariff hikes (AP News). That is a striking figure. It shows that governments are now using industrial policy defensively, not just developmentally. Subsidies are no longer aimed only at technological upgrading; they are being deployed as insurance against somebody else’s trade policy.

Taken together, these cases show that tariff uncertainty is not generating a clean map of winners and losers. It is generating a more fragmented map in which some projects accelerate, some pause, and many become dependent on public money. The result is not deglobalization in a simple sense. It is selective regionalization underwritten by subsidies and shadowed by policy risk.

The New Supply Chain Logic: Build for Optionality, Pay for Duplication

The old efficiency model rewarded concentration. A company could centralize wafers in one location, packaging in another, and final assembly in a third, trusting low tariffs and predictable logistics to tie the system together. The new model rewards optionality. Firms want a second qualified site, a domestic packaging backup, a politically favored jurisdiction, and enough regional redundancy to survive the next rule change.

That logic is rational at the firm level but costly at the system level. It raises capital intensity, duplicates assets, and can leave governments bidding against one another for projects that may have been built somewhere anyway. PIIE has warned that the CHIPS Act already shifted investment toward the United States and that policy instability could poison that well by making future investors question whether promised support will remain credible (PIIE). The same warning applies beyond the United States: if tariffs are discretionary and exemptions reversible, capital will demand higher returns before committing to new capacity.

Investors should pay attention to where optionality is being built, not just where factories are announced. Advanced packaging, specialty chemicals, chip equipment servicing, and power infrastructure are the less visible layers that determine whether relocation becomes a functioning ecosystem. Countries that attract only final assembly may see temporary export growth without locking in durable technological depth.

By 2028, the most competitive semiconductor and electronics hubs are likely to be those that combine three things: credible trade access, targeted subsidies that extend beyond headline fabs, and domestic capabilities in packaging, engineering talent, and industrial utilities. On that basis, governments should shift from one-off megaproject chasing to ecosystem discipline. The U.S. government should pair semiconductor incentives with a more predictable tariff framework and clearer long-term treatment of derivative electronics, while Southeast Asian governments should use the current relocation wave to build supplier depth rather than relying on tariff arbitrage. The OECD’s warning will remain relevant until trade policy stops behaving like an intermittent shock and starts acting like infrastructure. Until then, the next phase of globalization will be built not where trade is cheapest, but where policy is least likely to change mid-construction.

Trending Topics

Browse by Category

The OECD’s Warning Is Already on the Factory Floor: How Tariff Uncertainty Is Rewriting Semiconductor Investment Maps