Incorporator The Journal of U.S. Formation
Start Free Quiz
Editorial 6 MIN READ

The CHIPS Act on the President's desk: what it means for how you stand up a fab

A $52.7 billion appropriation, a new 25% manufacturing credit, and a ten-year ban on the companies that take the money from building leading-edge capacity in China

The Editors
Staff, Incorporator.org
6 min read Updated
Contents 5 sections
  1. What the money actually is
  2. The guardrail is a formation constraint
  3. How new fabs are actually being structured
  4. What remains unclear until the NPRM
  5. Sources

he CHIPS and Science Act cleared the Senate 64 to 33 on July 27 and the House 243 to 187 the next day. It is on the President's desk this week. For anyone planning to stand up a US semiconductor fab, the document that arrives at the signing ceremony is both a cash appropriation and a tax statute, and the entity you form to catch the money has to be built for both.

The headline is $52.7 billion for semiconductor manufacturing, research, and workforce, most of it routed through the Department of Commerce. The quieter, and for many fabs more valuable, piece is a new advanced manufacturing investment tax credit at IRC § 48D, set at 25 percent of qualified investment in a facility that manufactures semiconductors or the equipment used to make them. The credit and the grants share a single deterrent clause: if you take the money, you cannot materially expand leading-edge semiconductor manufacturing capacity in China, Russia, Iran, or North Korea for ten years.

What the money actually is

The Commerce Department gets $39 billion for a financial-assistance program to incentivize the construction, expansion, or modernization of semiconductor facilities in the United States, split across fiscal years 2022 through 2026, with the largest tranche in the first year. An additional $2 billion is carved out inside that total for mature-node production, which is the political compromise that kept senators from automotive states on board. The law adds $11 billion for a Department of Commerce R&D program including the National Semiconductor Technology Center, a National Advanced Packaging Manufacturing Program, and Manufacturing USA institutes. Defense gets $2 billion for a Microelectronics Commons; State and the Commerce Department split $500 million for an international technology security fund. Workforce and wireless-supply-chain programs account for the rest.

The $52.7 billion figure cited on every front page is the sum of those buckets. It is not a pot that any one fab will draw from alone. A single leading-edge facility costs more than the entire first-year appropriation on its own, which is why the tax credit is load-bearing.

The § 48D credit is a direct 25 percent of qualified investment placed in service after December 31, 2022, in a facility whose primary purpose is semiconductor or semiconductor-manufacturing-equipment production. The credit is refundable to most taxpayers via a direct-pay election at § 48D(d), which is the design choice that makes it usable by companies without enough US tax liability to absorb it. Construction must begin before 2027. The "applicable taxpayer" definition excludes any "foreign entity of concern," which imports the guardrail directly into eligibility rather than leaving it to contract.

The guardrail is a formation constraint

The ten-year guardrail is not a covenant you sign once at closing. It is a statutory condition that runs with the credit and the grant. An "applicable taxpayer" that engages in a "significant transaction" involving material expansion of semiconductor manufacturing capacity in China, Russia, Iran, or North Korea during the ten years after the credit is claimed must recapture the full credit. The grant side has a parallel clawback written into the program authority. Legacy-node production (28nm and older on logic, with equivalent carve-outs for memory expected in the Department of Commerce implementing rules) is excluded from the prohibition, but the line between leading-edge and legacy is exactly where the Commerce NPRM to come will do its real work.

For a US operating company with a pre-existing Chinese subsidiary that makes chips, this means the formation question is not only "what entity takes the grant" but "what is in the controlled group, and does any piece of it sit on the wrong side of the guardrail." The statute reaches members of a consolidated group and certain partnership allocations. A company considering the credit is effectively being asked to commit its entire consolidated footprint, not just the fab that applies.

How new fabs are actually being structured

The reference structure for a domestic fab under this law looks like this. A US parent C corporation holds the fab through a single-member Delaware LLC. The LLC is disregarded for federal tax purposes under Treas. Reg. § 301.7701-3, so the fab's income, deductions, and credits roll up to the parent and the § 48D credit is claimed on the parent's return. The SMLLC gives the fab its own liability shield, its own state registration, and a clean contract signer without introducing a second taxpayer. The parent is the applicable taxpayer for § 48D purposes.

When the fab is funded by more than one party, the structure gets harder. A common pattern is a limited partnership or multi-member LLC taxed as a partnership, with a C-corp general partner affiliated with the operator and limited partners taking equity tranches. Partnership allocations of the credit follow § 704(b) and the regulations under § 48D once Treasury writes them. Direct pay for partnerships under § 48D(d) is expected to follow the same mechanical path as other refundable credits, which means the partnership receives the payment and distributes cash consistent with the partners' capital accounts. Until the Treasury regulations are out, JV documents are being drafted with placeholder provisions that reallocate economic value if the credit partially disqualifies one partner (for example, a partner that later becomes a foreign entity of concern).

The choice between SMLLC under a C-corp and a multi-member partnership is not ideological. A single strategic investor writing the full check takes SMLLC. A consortium of equipment makers, pension funds, or sovereign wealth co-investors takes the partnership, accepting the complexity to share the economics. The one thing almost no new fab is doing is forming the top-level entity as an LLC that elects S-corp status; the ownership restrictions in IRC § 1361 are incompatible with the investor base, and the credit mechanics are cleaner at the C-corp layer.

What remains unclear until the NPRM

The Department of Commerce is required to publish implementing rules for the financial-assistance program, including pre-award application requirements, clawback triggers, and the definition of leading-edge, in a forthcoming Notice of Proposed Rulemaking. Treasury will separately issue guidance on § 48D, including the direct-pay mechanics, the treatment of progress expenditures, the definition of qualified investment, and how the applicable-taxpayer rule applies to partnerships and consolidated groups. Neither document existed as of the Senate vote last week.

Three open questions will move deal terms. First, what counts as "material expansion" of leading-edge capacity abroad. A de minimis allowance inside the guardrail, even a narrow one, changes how global fabs think about routine tool upgrades in existing Chinese plants. Second, what exactly is "legacy" beyond 28nm logic, particularly for memory and analog. Third, how the NPRM handles pre-existing joint ventures that have both US and Chinese operating subsidiaries. Commerce has flexibility here that the statute does not spell out.

For a fab that hopes to break ground in 2023, the operational advice is to form the holding C corporation now, form the SMLLC now, file the Delaware or Arizona or Texas registrations now, and hold the credit application for the Treasury regulations. The formation work is cheap and reversible. The credit application is neither.

Sources

Keep reading

More from the journal.