Recommended Free Tools
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
On April 14, 2025, Nvidia said its Blackwell AI chips had entered production at TSMC’s Arizona fab, while AMD announced that fifth-generation EPYC products had been brought up and validated there. Both were meaningful U.S. manufacturing milestones—but neither announcement showed that the companies had moved their leading-edge chip supply chains out of Taiwan. The timing came as Washington was threatening semiconductor-specific tariffs, making Arizona a potential hedge as well as a political signal.
What Nvidia and AMD actually announced
Nvidia’s announcement concerned Blackwell wafer production at TSMC’s Arizona Fab 21. “Entered production” establishes that production activity had begun; the public announcement did not disclose wafer volumes, yields, or what share of Blackwell output would come from Arizona. It should not be read as a claim that all Blackwell chips—or Nvidia’s overall chip supply—were now made in the United States. The White House’s account of Nvidia’s announcement also described a separate plan to build AI supercomputers in Texas with manufacturing partners.
Nvidia said it expected to produce up to $500 billion of AI infrastructure in the United States over four years. That figure refers to AI infrastructure, not $500 billion in Nvidia chip sales or semiconductor output. Chip fabrication in Arizona and system manufacturing in Texas are different stages of a broader plan.
AMD’s April 14 statement bundled two distinct milestones. The company said its fifth-generation EPYC products had been successfully brought up and validated at TSMC Arizona. Separately, it said its next-generation EPYC processor, code-named Venice, had been taped out and brought up using TSMC’s N2 process. The Arizona validation is evidence of U.S.-based manufacturing and product qualification; the N2 milestone is not evidence that Venice was being manufactured in Arizona. AMD’s release describes both developments.
#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
Neither company published enough information to establish high-volume Arizona output. Production entry, bring-up, and validation are important steps, but they are not the same as disclosing sustained volume, commercial contribution, or a complete manufacturing chain.
Why tariffs were part of the story
The announcements landed amid uncertainty about U.S. trade policy. In April 2025, semiconductor-specific tariffs were threatened or anticipated, not a settled tariff already fully applied to Nvidia and AMD products on April 14. Separately, the administration’s reciprocal-tariff order set a general additional duty of 10% from April 5 and scheduled country-specific rates from April 9, subject to the order’s terms and exceptions. The distinction matters: the general framework and a potential semiconductor-specific measure were not the same policy. The executive order lays out the reciprocal-tariff framework.
A tariff on imported chips could raise their landed cost in the United States. A wafer fabricated domestically might reduce some exposure, depending on the measure’s scope and how customs rules determine the product’s country of origin. But domestic fabrication alone does not guarantee a tariff exemption: packaging, testing, and other steps may occur elsewhere, and the applicable legal treatment depends on the specific policy and product.
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
There is a plausible business case for Arizona as insurance against both trade disruption and geopolitical concentration. Yet the public announcements do not prove that tariffs caused the underlying decisions. A fab takes years to build, equip, and qualify; customer allocation and process planning also precede a news release. Analysts quoted by EE Times saw the timing as a way to manage tariff uncertainty and demonstrate alignment with U.S. manufacturing priorities. That is a reasonable interpretation of the timing and messaging—not a documented admission by either company that tariffs initiated the work.
For a successful tariff hedge, the right product must actually be made in Arizona, qualifying origin rules must apply, usable capacity and yields must be available, and any savings must justify the added cost. A U.S. wafer could lower exposure to one risk while leaving costs or cross-border dependencies elsewhere in the chain.
Arizona is a real manufacturing site, not a replacement for Taiwan
TSMC’s first Arizona fab was designed around 4nm-class production, with production targeted for the first half of 2025. TSMC’s broader Arizona roadmap calls for later fabs to bring more advanced 3nm- and 2nm-class manufacturing to the United States later in the decade. These “4nm” and “2nm” labels describe process generations; they are not literal measurements of a single transistor feature. See TSMC’s Arizona roadmap.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
That staged buildout is a substantial commitment, but it does not instantly reproduce Taiwan’s scale, process maturity, supplier network, or concentration of leading-edge production. Capacity must be installed and qualified, yields must mature, and production must compete for allocation alongside other customers’ needs. EE Times cited an analyst estimate that Arizona’s advanced-node capacity was roughly 20% of corresponding Taiwan capacity. That is an analyst estimate, not a TSMC-reported operating statistic, and it should not be mistaken for a direct measure of capacity available to Nvidia or AMD.
TSMC announced an additional $100 billion U.S. investment in 2025, bringing its planned U.S. total to $165 billion and including three additional fabs, two advanced-packaging facilities, and an R&D center. Those plans point toward greater domestic capability over time, not an immediate end-to-end U.S. supply chain. TSMC’s announcement also described the Arizona site as employing more than 3,000 people on 1,100 acres and said it had been in volume production since late 2024.
The packaging gap limits what “U.S.-made” means
For an AI accelerator, making a wafer is only one step. High-end products can require advanced 2.5D or 3D packaging, the integration of chiplets and high-bandwidth memory (HBM), large interposers, specialized substrates, testing, and final assembly. Nvidia’s CoWoS packaging and comparable advanced processes are part of the capacity equation, not an optional finishing touch.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
In April 2025, advanced packaging remained heavily concentrated in Taiwan. An Arizona-fabricated wafer might still need to travel there for packaging, then move again for testing or system integration. In that case, domestic wafer production diversifies one stage but leaves the product dependent on cross-border logistics, Taiwan-based packaging capacity, and Asian suppliers of memory, substrates, materials, and equipment. TSMC’s planned U.S. packaging facilities were a longer-term part of the expansion, not an immediate remedy for this gap. EE Times reported on the packaging constraint and the possibility that Arizona wafers would still go to Taiwan.
“Made in the U.S.” can therefore mean several different things: wafer fabrication, packaging, testing, server assembly, or some combination. These announcements supported claims of U.S. manufacturing activity, but they did not establish that every critical stage for either company’s products was domestic.
What the Arizona moves mean for costs and resilience
Adding U.S. production can diversify supply and reduce dependence on a single geography. It may also improve access to capacity for certain products or customers if the Arizona site meets the required schedule, yield, and qualification standards. Those benefits matter to companies planning around geopolitical risk, shipping interruptions, and unpredictable trade policy.
Best Value
- Powered by the NVIDIA Blackwell architecture and DLSS 4 OC mode: 2640MHz/Default mode: 2610MHz (Boost Clock)
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.125-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
But diversification is not automatically cheaper. A newer fab may face higher operating costs, lower initial utilization, yield-ramp expenses, and the cost of building a local workforce and supplier ecosystem. If wafers travel to Asia for packaging and return for integration, extra logistics remain. Domestic production may be worth paying for as risk insurance, but the public evidence does not show that it lowers Nvidia’s or AMD’s total cost of manufacturing.
Nor can Arizona instantly take over every product made in Taiwan. Demand for AI accelerators was growing, and a limited pool of qualified capacity must serve multiple chip customers. Companies may use Arizona for selected products, variants, qualification work, or geographic diversification without shifting their highest-volume or newest designs there. Capacity, process fit, yield, and delivery timing—not just tariff exposure—determine what can move.
Could Intel Foundry take some of the work?
Intel offers another U.S.-based foundry possibility and has domestic manufacturing and advanced-packaging capabilities. In principle, that could provide a second source for selected designs and reduce concentration risk. But it is not a plug-and-play substitute for TSMC. A customer must have suitable design enablement, qualify a process, gain confidence in yields, and secure commercial capacity. Moving an advanced product family to another foundry is a multiyear engineering effort.
Analysts cited by EE Times were skeptical that Intel could immediately absorb substantial advanced Nvidia or AMD volume. Intel may be more plausible for selected products that fit its process and schedule than as an instant source for the newest AI accelerators. The April 2025 announcements did not establish that Nvidia or AMD had committed production to Intel Foundry.
How to read the announcements
The strongest interpretation is neither “the companies moved to America” nor “the announcements were only political theater.” Nvidia reported Blackwell production activity at Arizona; AMD reported EPYC bring-up and validation there. Those are operational milestones that make the U.S. a more credible part of their manufacturing footprint. The timing also gave both companies a visible response to Washington’s priorities and a possible hedge against trade uncertainty.
The limits are equally important: public figures for Arizona wafer volumes, yields, and the share of either company’s output were not disclosed; AMD’s Venice N2 milestone was separate from Arizona; and packaging, memory, materials, and much of the established leading-edge ecosystem still depended on Taiwan and Asia. Arizona adds resilience incrementally. It does not make Taiwan unnecessary.
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.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
