Should the Pentagon have used Commerce Department funds for semiconductors?

The U.S. Defense Department has successfully, albeit surreptitiously, swiped $3.5 billion from the Commerce Department’s CHIPS Act funding to subsidize Intel’s creation of a classified advanced semiconductor manufacturing facility called Secure Enclave. Although being specifically built for defense and intelligence state-of-the-art computer chip needs, the funding for Secure Enclave was not included in the original CHIPS Act law. The Commerce Department, unsurprisingly, objected to losing control of almost 10% of the CHIPS Act’s manufacturing grants when this funding was commandeered by the Defense Department for Secure Enclave.

Advocates for Secure Enclave argued that Trusted Foundry, a program in existence since 2004 where Defense Department inspectors certify commercial semiconductor sector manufacturing facilities, was not secure enough for some types of chips, including state-of-the-art ones . Nevertheless, the reaction to this budget allocation legerdemain has been widespread disapproval.

“I don’t know how this happened, but it should not have,” said Charles Wessner, an expert on global innovation policy at the Center for Strategic International Studies think tank.

“There should be no Secure Enclave in the CHIPS program,” said Rep. Zoe Lofgren, D-Calif., the ranking member of the House Science, Space, and Technology Committee, “and any secure program that might be necessary should be funded by the Department of Defense … not from CHIPS funding that should be focused on revitalizing our domestic chip capacity.”

Is this simply a bureaucratic schoolyard fight where the Defense Department steals some of the Commerce Department’s lunch money?

While it is impossible to fairly evaluate the merits of both what has been done and how it has been done without access to the classified briefings, some semiconductor industry and national security program context helps explain what exactly is going on.

Semiconductors are more valuable and critical today than oil during the industrial age. Natural gas, coal, wind, solar, geothermal and nuclear power are all alternative energy sources that in some instances can replace the power supplied by oil. In addition, multiple countries on different continents have large sources of oil. Neither of these two key factors apply to semiconductors. There are no alternatives to semiconductors, and as a practical matter there is no real alternative to Taiwan Semiconductor Manufacturing Co. in Taiwan.

TSMC now manufactures approximately 90% of all advanced semiconductors in the world. When Russia invaded Ukraine , and Western countries heavily sanctioned Moscow’s oil and gas industry, Western countries had the ability — albeit in some cases with great difficulty — to turn to alternative producers of oil and gas as well as alternative sources of energy. If China attacks or blockades Taiwan, it is currently impossible for the lost volume of TSMC’s advanced semiconductors to be made up elsewhere. TINA, an investment and political acronym for “there is no alternative,” now has a place in the national security discourse.

Visitors look at screens at a museum of Taiwan Semiconductor Manufacturing Co. on July 5, 2023. (Sam Yeh/AFP via Getty Images)

Adm. Phil Davidson, then-chief of U.S. Indo-Pacific Command, predicted China will attempt an attack on Taiwan by 2027 . What is now known as the Davidson window provides additional context for Secure Enclave. Intel recently purchased from the semiconductor specialist ASML a high numerical aperture extreme ultraviolet wavelength semiconductor manufacturing system — its most advanced system. Intel expects that it will begin early production of some of the world’s most advanced computer chips from this system in the 2026-2027 time frame.

Congress’ notoriously erratic defense budget approval process means if Secure Enclave had not been funded by the CHIPS Act program, there likely would be no operational Secure Enclave until after the Davidson window had closed.

There is, though, another piece to this national security semiconductor puzzle: Trusted Foundry. A little-known agency within the Defense Department, the Defense Microelectronics Activity is the accrediting authority for the Trusted Foundry program, which ensures semiconductor components incorporated into military systems have not been compromised by foreign agents.

Several key weaknesses in the Trusted Foundry program provide context for why there may have been no desire to have this new secure advanced semiconductor manufacturing program simply folded into it.

Until Intel’s new extreme ultraviolet wavelength tool comes online in the Secure Enclave, the Defense Department can only buy chips from the trusted foundries, which are two generations behind what is available on the commercial state-of-the-art market. Part of the problem is that the Defense Department is not a large purchaser of microelectronics, so many semiconductor companies are unable to make a business case for participating in the Trusted Foundry program.

Insider threats are also a significant problem with this program, according to Mark Lewis, former acting deputy undersecretary of defense for research and engineering and former director of defense research and engineering for modernization. “We’ve seen a number of examples where the biggest threats that we face often are the insider threat. It’s the people inside the fence line, behind the guards, who we think we’ve cleared,” he said. “They’re the ones that pose the biggest threats to us.”

In light of these problems, the Trusted Foundry program has been deemed a failure and was supposed to be phased out, but that never happened. Defense Microelectronics Activity continued to run the Trusted Foundry program in a business-as-usual mode, with no announcement regarding its closure, no winding down of existing contracts, and no website notice of the Trusted Foundry program’s demise.

Instead, Defense Microelectronics Activity has focused on the next iteration of the Trusted Foundry program, which it calls the Trusted Foundry Access III program, by awarding contracts in 2023 “to ensure uninterrupted access to measurably secure, State-of-the-Art semiconductor foundry services over a 10-year period of performance.”

Pilfering funding from another department for one’s own is as old as bureaucracy itself. The semiconductor industry, national security and domestic political contexts to this particular interdepartmental commerce and defense bureaucratic wrangling indicate the national security stakes could not be higher.

André Brunel is an international technology attorney with Reiter, Brunel and Dunn. This commentary was adapted from his article published in the Journal of Business & Technology Law. The views and opinions expressed in this commentary are his and do not necessarily reflect the views or positions of the law firm or any clients it represents.

美国国防部成功地（尽管是秘密地）从商务部的《芯片技术创新法案》（CHIPS Act）中挪用了35亿美元，用于补贴英特尔公司建造一座名为“安全飞地”（Secure Enclave）的机密先进半导体制造工厂。尽管“安全飞地”是专门为满足国防和情报部门对尖端计算机芯片的需求而建造的，但其资金并未包含在最初的《芯片技术创新法案》中。不出所料，当国防部将这笔资金挪用于“安全飞地”项目时，商务部对失去对《芯片技术创新法案》近10%制造拨款的控制权表示反对。

安全隔离区的支持者认为，自2004年以来一直存在的“可信代工厂”项目（该项目由国防部检查员对商业半导体制造工厂进行认证）对于某些类型的芯片（包括最先进的芯片）而言安全性不足。然而，这种预算分配上的花招遭到了广泛的反对。

“我不知道这是怎么发生的，但这本不应该发生，”战略国际研究中心智库的全球创新政策专家查尔斯·韦斯纳说。

“CHIPS 计划中不应该有安全隔离区，”众议院科学、空间和技术委员会资深成员、加州民主党众议员佐伊·洛夫格伦表示，“任何必要的安全计划都应该由国防部资助……而不是从 CHIPS 计划的资金中支出，CHIPS 计划的资金应该集中用于振兴我们的国内芯片产能。”

这难道只是一场官僚主义的闹剧，国防部抢走了商务部的部分“午餐钱”吗？

虽然在没有查阅机密简报的情况下，不可能公正地评价已经做的事情以及做事的方式的好坏，但一些半导体行业和国家安全计划的背景有助于解释到底发生了什么。

如今，半导体比工业时代的石油更有价值、更至关重要。天然气、煤炭、风能、太阳能、地热能和核能都是可替代能源，在某些情况下可以取代石油提供的电力。此外，不同大陆的多个国家都拥有丰富的石油资源。然而，这两个关键因素都不适用于半导体。半导体没有替代品，实际上，台湾的台积电也没有真正的替代品。

台积电目前生产全球约90%的先进半导体。当俄罗斯入侵乌克兰，西方国家对莫斯科的石油和天然气行业实施严厉制裁时，西方国家有能力——尽管在某些情况下举步维艰——转向其他石油和天然气生产商以及其他能源来源。如果中国攻击或封锁台湾，目前台积电先进半导体的产量损失无法在其他地方得到弥补。“别无选择”（TINA）——这个投资和政治术语的缩写——如今已成为国家安全话语中的重要议题。

2023年7月5日，参观者在台湾积体电路制造股份有限公司博物馆观看屏幕。（Sam Yeh/AFP via Getty Images）

时任美国印太司令部司令菲尔·戴维森海军上将曾预测，中国将在2027年前对台湾发动攻击。如今被称为“戴维森窗口”的事件为“安全飞地”（Secure Enclave）提供了更多背景信息。英特尔近期从半导体专家ASML公司购入了一套高数值孔径极紫外波长半导体制造系统——这是其最先进的系统。英特尔预计将在2026年至2027年间开始利用该系统生产一些世界上最先进的计算机芯片。

国会出了名的反复无常的国防预算审批程序意味着，如果安全飞地没有得到 CHIPS 法案计划的资助，那么在戴维森窗口关闭之前，安全飞地可能都不会投入使用。

然而，国家安全半导体难题还有另一块拼图：可信代工厂（Trusted Foundry）。国防微电子活动中心（Defense Microelectronics Activity）是国防部下属一个鲜为人知的机构，也是可信代工厂项目的认证机构，该项目旨在确保军用系统中采用的半导体元件未被外国势力篡改。

Trusted Foundry 计划中的几个关键缺陷，解释了为什么人们可能不愿意将这个新的安全先进半导体制造计划简单地并入其中。

在英特尔的新型极紫外波长工具在安全隔离区投入使用之前，国防部只能从受信任的代工厂采购芯片，而这些代工厂的产品比目前市场上最先进的商用芯片落后两代。部分原因是国防部并非微电子产品的主要采购方，因此许多半导体公司无法证明参与“受信任代工厂”计划的商业价值。

据曾任国防部负责研究与工程的代理副部长兼国防研究与工程现代化项目主任的马克·刘易斯称，内部威胁也是该项目面临的一大问题。“我们已经看到很多例子，我们面临的最大威胁往往来自内部。这些人就在围栏内，在警卫后面，我们以为已经排除了他们的嫌疑，”他说。“他们才是对我们构成最大威胁的人。”

鉴于这些问题，“可信代工厂”项目已被认定为失败，本应逐步终止，但这种情况并未发生。国防微电子活动公司继续照常运营“可信代工厂”项目，既未宣布项目关闭，也未终止现有合同，其网站上也未发布任何关于该项目终止的通知。

相反，国防微电子活动部门将重点放在了可信代工厂计划的下一阶段，即可信代工厂接入 III 计划上，并在 2023 年授予合同，“以确保在 10 年的执行期内不间断地获得可衡量的安全、最先进的半导体代工厂服务”。

从其他部门挪用资金为自己所用，这种做法与官僚机构本身一样古老。半导体行业、国家安全和国内政治背景，以及此次商务和国防部门间官僚机构的权力斗争，都表明其事关国家安全，意义重大。

安德烈·布鲁内尔是Reiter, Brunel & Dunn律师事务所的国际技术律师。本文改编自他发表在《商业与技术法杂志》上的文章。文中表达的观点和意见仅代表布鲁内尔本人，并不一定反映律师事务所或其任何客户的观点或立场。