What is the Section 232 AI chip tariff? The Section 232 AI chip tariff is a 25% duty imposed under the Trade Expansion Act of 1962 on imports of advanced semiconductors critical to artificial intelligence infrastructure, including chips like the NVIDIA H200 and AMD Instinct MI325X. On April 14, 2026, the U.S. Trade Representative and the Secretary of Commerce must deliver the Phase 2 negotiations report to President Trump, a pivotal deadline that could trigger broader tariffs on semiconductor imports from Taiwan, South Korea, and Japan. For CRE data center investors, this report could reshape the cost structure of every AI facility in the pipeline. For a broader look at technology reshaping commercial real estate, see our guide on AI tools for commercial real estate investors.
Key Takeaways
- The Section 232 Phase 2 report due April 14 could expand AI chip tariffs beyond the current 25% rate on advanced semiconductors from key producing nations.
- NVIDIA H200 and AMD MI325X chips are already subject to 25% duties, adding $25,000 per $100,000 in customs value for covered imports.
- Current data center exemptions protect most domestic GPU procurement, but Phase 2 could narrow or restructure these carve-outs.
- A proposed tariff offset program would reward companies investing in U.S. semiconductor manufacturing with preferential duty rates.
- CRE investors with data center exposure should model GPU cost increases of 10% to 25% into forward-looking underwriting scenarios.
Understanding the Section 232 AI Chip Tariff Timeline
President Trump signed Proclamation 11002 on January 14, 2026, imposing the initial 25% ad valorem tariff on a narrow category of advanced computing chips. The administration cited national security concerns, arguing that concentrated foreign production of AI semiconductors, particularly by TSMC in Taiwan, creates supply chain vulnerability across all 16 critical infrastructure sectors identified by the Department of Homeland Security.
The tariff structure follows a deliberate two-phase approach. Phase 1, now in effect, targets only the highest-performance chips meeting specific Total Processing Performance thresholds above 14,000 and defined DRAM bandwidth parameters. Phase 2, contingent on the April 14 report, could impose what the proclamation describes as "broader tariffs on semiconductors, at a rate of duty that is significant." This language signals that the scope and rates could increase substantially.
Three critical deadlines define the regulatory timeline. April 14 brings the Phase 2 negotiations report. July 1 requires a Commerce Department update on the semiconductor market for U.S. data centers, potentially triggering further tariff modifications. And the ongoing trade negotiations with Japan, South Korea, and Taiwan will determine country-specific rates going forward.
How Phase 2 Could Impact Data Center Economics
The current exemption structure is more favorable than many investors realize. Chips imported for U.S. data center operations, domestic R&D, repairs, startups, and public sector applications are all exempt from the 25% duty. In practice, this means the Phase 1 tariff primarily targets advanced semiconductors destined for re-export, particularly to China.
Phase 2 changes that calculus. If broader tariffs are imposed, the exemption framework could narrow. Even if data center exemptions survive, the tariff offset program introduces a new variable: companies investing in domestic semiconductor production would receive preferential rates, creating a two-tier cost structure that advantages vertically integrated operators over pure-play data center landlords.
For context, the AI data center market is already navigating extraordinary cost pressures. According to Morgan Stanley, AI data center debt issuance exceeded $200 billion in 2025, with projections for even higher volumes in 2026. GPU procurement represents one of the largest capital expenditure line items for hyperscale and colocation operators. A 25% tariff applied broadly to GPU imports would add approximately $7,500 to $10,000 per high-end chip, with cascading effects on rack-level economics and ultimately on lease rates for tenants. For more on how these financing structures are evolving, see our analysis of AI data center GPU debt financing.
Winners and Losers in the CRE Data Center Sector
The Section 232 AI chip tariff creates clear competitive dynamics within the data center real estate ecosystem.
- Hyperscale operators with domestic chip partnerships benefit. Companies like Microsoft, Google, and Amazon, which are already investing billions in custom chip development and domestic manufacturing partnerships, would qualify for tariff offset programs. Amazon recently committed over $200 billion in AI infrastructure capex, while Anthropic is exploring custom chip design. These operators can absorb or offset tariff costs that smaller competitors cannot.
- Colocation and wholesale data center landlords face margin pressure. Operators like Digital Realty, Equinix, and CoreWeave rely on tenants who procure their own GPUs. If those tenants face higher chip costs, lease negotiations tighten and expansion timelines stretch. Digital Realty recently closed a $3.25 billion hyperscale fund, but returns on that capital could compress under a broader tariff regime.
- Edge and secondary market data centers may see relative advantage. Facilities in markets like Phoenix, Dallas, and Northern Virginia with access to domestic power and proximity to chip packaging facilities could command premium positioning as operators seek to minimize tariff exposure through supply chain localization.
- Intel and domestic foundry investments gain strategic value. Intel's 18A process node, central to the Terafab megaproject with Elon Musk, becomes more commercially relevant if tariff offset programs reward domestic fabrication. CRE investors with exposure to semiconductor manufacturing campuses could see upside. For more on this, read our coverage of the Intel Terafab foundry and its CRE implications.
What CRE Investors Should Do Before April 14
The Phase 2 report is not just a trade policy document. It is a signal that will influence capital allocation decisions across the data center sector for the remainder of 2026 and beyond. Here is how CRE investors should prepare.
First, stress-test GPU cost assumptions in active underwriting. If you are evaluating a data center acquisition or development deal, model scenarios where GPU costs increase by 10%, 15%, and 25%. Assess how each scenario affects tenant economics, lease-up timelines, and stabilized NOI. A cap rate of 5.5% on a data center deal assumes stable operating costs; a tariff-driven GPU price spike could compress NOI by 8% to 12% depending on tenant mix.
Second, evaluate tenant exposure to tariff risk. Hyperscale tenants with domestic chip programs (Microsoft with its Maia chips, Google with TPUs, Amazon with Trainium and Graviton) carry lower tariff risk than AI startups relying entirely on imported NVIDIA hardware. Tenant credit analysis should now include supply chain resilience as a factor.
Third, monitor the tariff offset program details. If Phase 2 includes incentives for domestic semiconductor investment, data center properties adjacent to or integrated with chip fabrication facilities gain strategic value. Markets like Columbus (Intel), Phoenix (TSMC Arizona), and Austin (Samsung) could see accelerated demand. For personalized guidance on positioning your portfolio for these shifts, connect with The AI Consulting Network.
The Bigger Picture: Semiconductor Sovereignty and CRE
The Section 232 AI chip tariff is part of a broader policy push toward semiconductor sovereignty. The Stanford 2026 AI Index, released today, highlights that one company in Taiwan, TSMC, fabricates almost every leading AI chip. That concentration creates national security risk and supply chain fragility that the administration is actively working to address.
For CRE investors, this means the convergence of AI infrastructure and trade policy is no longer a distant concern. With 92% of corporate occupiers having initiated AI programs and CRE sales volume forecast to increase 15% to 20% in 2026 (Source: PwC Emerging Trends in Real Estate), the cost of the chips powering AI workloads directly influences the viability of data center investments.
The AI in real estate market is projected to reach $1.3 trillion by 2030 at a 33.9% CAGR. But that growth trajectory assumes stable input costs. If Phase 2 tariffs significantly increase GPU procurement expenses, some planned data center developments may be delayed, downsized, or restructured entirely. CRE investors looking for hands-on guidance navigating these trade policy dynamics can reach out to Avi Hacker, J.D. at The AI Consulting Network.
What Happens After April 14
Regardless of the Phase 2 outcome, two additional milestones will shape the landscape. The July 1, 2026 Commerce Department update on the data center semiconductor market could trigger further modifications. And ongoing country-specific negotiations with Taiwan, South Korea, and Japan will determine whether allied nations receive preferential rates or face the full weight of broader tariffs.
Investors should also watch for secondary effects. If tariffs drive up GPU costs domestically, demand for nuclear-powered data centers and energy-efficient AI architectures could accelerate as operators seek to offset hardware cost increases with lower operating expenses.
Frequently Asked Questions
Q: What chips are currently subject to the Section 232 AI chip tariff?
A: The 25% tariff currently applies to advanced semiconductors meeting specific performance thresholds, including the NVIDIA H200 and AMD Instinct MI325X. These are the highest-end AI training and inference chips, classified by Total Processing Performance above 14,000 and defined DRAM bandwidth parameters.
Q: Are data center operators exempt from the Section 232 tariff?
A: Under Phase 1, yes. Chips imported for U.S. data center operations, domestic R&D, repairs, startups, and public sector use are exempt. However, Phase 2 could modify or narrow these exemptions, which is why the April 14 report is critical for CRE investors to monitor.
Q: How would broader Phase 2 tariffs affect data center lease rates?
A: If GPU procurement costs increase by 15% to 25%, tenants may push for longer lease terms, lower base rates, or tenant improvement allowances to offset capital expenditure increases. Data center landlords should expect more aggressive lease negotiations in markets with high AI workload concentration.
Q: Which CRE markets benefit most from domestic semiconductor manufacturing incentives?
A: Markets with active semiconductor fabrication facilities stand to gain: Columbus, Ohio (Intel), Phoenix, Arizona (TSMC), Austin, Texas (Samsung), and potentially sites associated with the Terafab megaproject. These markets could see accelerated data center demand as operators co-locate near domestic chip sources.
Q: What is the tariff offset program mentioned in the Section 232 proclamation?
A: The tariff offset program, expected in Phase 2, would allow companies investing in U.S. semiconductor production to receive preferential tariff rates on imports. This creates a competitive advantage for vertically integrated operators and could influence where data center developers choose to build new facilities.