concurring in part and dissenting in part.
While I join the majority opinion insofar as it holds that the 35 U.S.C. § 271(e)(1) safe harbor does not immunize Amphas-tar’s accused use of the '886 patent in its manufacturing process, I respectfully dissent from the majority’s holding that Teva does not infringe the '886 patent under 35 U.S.C. § 271(g).1 The majority reasons that a patent related to quality control testing cannot be infringed under § 271(g), which states, “[w]hoever without authority imports into the United States or offers to sell, sells, or uses within the United States a product which is made by a process patented in the United States shall be liable as an infringer.” 35 U.S.C. § 271(g) (2012) (emphasis added). Quality control, according to the majority, is not used to “make” a product. This seems to me too limited a construction of § 271(g).
I
The central question here is whether quality control is part of the process of “manufacturing” a product. The majority holds that it is not, relying primarily on Bayer AG v. Housey Pharmaceuticals, Inc., 340 F.3d 1367 (Fed.Cir.2003). There *623we held that § 271(g) “contemplates that ‘made’ means ‘manufactured.’ ” Id. at 1372. We also held that “in order for a product to have been ‘made by a process patented in the United States’ it must have been a physical article.” Id. at 1377. Finally, Bayer held that “the process must be used directly in the manufacture of the product, and not merely as a predicate process to identify the product to be manufactured.” Id. at 1378. Thus in Bayer we held that a method for screening substances to identify promising products was not a method used in the manufacture of a product. Id. at 1369, 1378. “A drug product, the characteristics of which were studied using the claimed research processes ... is not a product ‘made by’ those claimed processes.” Id. at 1378.2
The patent here, however, is not utilized to identify the product to be made, but rather is used in the manufacturing process. The quality control process of the '886 patent is an intermediate step to determine which batches of putative enoxa-parin must be discarded, and which batches may be incorporated in the final drug product. It is distinctly part of the manufacturing process of the product.
The dictionary definitions of “make” and “manufacture” relied on by the majority at most suggest that quality control, standing alone, is not making or manufacturing. But they hardly suggest that quality control is not part of making or manufacturing. Nor can there be any suggestion that the processes described in § 271(g) are limited to those that cover the entire manufacturing process. The majority opinion cites no authority that quality control is not a part of manufacturing, other than our non-precedential decision in Phillip M. Adams & Associates, LLC v. Dell Computer Corp., 619 Fed.Appx. 998 (Fed.Cir.2013). In fact, quality control is, as a general matter, considered to be a part of the drug manufacturing process. That is the view of the Food and Drug Administration (“FDA”). The FDA, in its Good Manufacturing Practice regulations, 21 C.F.R. §§ 210.1-210.3, defines “[mjanufac-ture” as “including] packaging and labeling operations, testing, and quality control of drug products.” 21 C.F.R. § 210.3(12) (2011) (emphasis added).
Similarly, statutes and regulations in other areas have recognized that quality control is inherent in the manufacturing process. For example, in the manufacture of chemicals, the Toxic Substances Control Act provides that the Administrator of the Environmental Protection Agency may “require ... [a] manufacturer or processor to submit a description of the relevant quality control procedures followed in the manufacturing or processing of [a] chemical substance or mixture.” 15 U.S.C. § 2605(b). So, too, in the manufacture of medical devices. A medical device manufacturer, in order to obtain approval of a device under the Investigational Device Exemption, must submit an application with, inter alia, a “description of the methods, facilities, and controls used for the manufacture ... of the device ... so that a person generally familiar with good manufacturing practices can make a knowledgeable judgment about the quality *624control used in the manufacture of the device.” 21 C.F.R. § 812.20 (2015). In other words, quality control is “used in the manufacture of the device.” Id,.; see also United States v. Castillo, 928 F.2d 1106, 1108 (11th Cir.1991) (“A device that is used for ‘quality control’ in the manufacture of any item can be considered a device used in the manufacture of the product.”).
II
However, we need not reach the question here of whether quality control is always part of a manufacturing process. Our precedent suggests that we should resolve the question of whether a product was “made by” a process on a case-by-case basis. See Bio-Tech. Gen. Corp. v. Genentech, Inc., 80 F.3d 1553, 1561 (Fed.Cir.1996). Under the facts of this case, the quality control testing of the '886 patent is clearly an integral part of the manufacturing process of enoxaparin. In order to understand why, it is helpful to understand how the final enoxaparin drug product is made.
Heparin is a naturally occurring anticoagulant consisting of a mixture of long chains of sugar molecules. Heparin may be cleaved, using different methods, into shorter sugar chains (“oligosaccharides”) to create different low molecular weight heparins (“LMWHs”), each of which is a different heterogeneous collection of oligo-saccharides. The different heterogeneous collections of oligosaccharides give each LMWH a different therapeutic effect.
Enoxaparin is one type of LMWH, and was first sold under the brand name Love-nox. As with any LMWH, the sugar chains in enoxaparin may differ slightly from batch to batch, but they have structural similarities determined to be unique to that LMWH. One such signature structural feature is a 1, 6-anhydro ring structure that is present at approximately 20% of the reducing ends of sugar chains in the collection. The molecular diversity of enoxaparin creates special problems for the manufacturing of a generic version of the drug, which must be bioequivalent to and contain the same active ingredients as the branded drug. Thus, as we previously described,
the FDA identified five criteria, or standards for identity, that together provide sufficient information to conclude that generic enoxaparin has the ‘same’ active ingredient as Lovenox. These criteria included, inter alia, [equivalence in disaccharide building blocks, fragment mapping, and sequence of oligosacchar-ide species.... Detecting the presence of a 1, 6 anhydro ring structure is particularly important for proving equivalence ....
Momenta Pharm., Inc. v. Amphastar Pharm., Inc. (Momenta I), 686 F.3d 1348, 1350-51 (Fed.Cir.2012) (citations and quotation marks omitted). As required by the FDA, only batches in which 15-25% of the sugar chains contain a 1, 6-anhydro ring structure at the reducing end may be released and combined for further processing to become the finished drug product.
Momenta’s '886 patent claims a method of analyzing and selecting batches of intermediate enoxaparin drug substance, based on the appropriate quantity of sugar chains containing the 1, 6-anhydro ring structure. The patent contemplates the usage of its methods during the manufacturing process, teaching, for example, a method that “provides a way to both streamline manufacturing and reduce costs while ensuring a more consistent, higher quality product,” U.S. Patent No. 7,757,886 col. 34 ll. 43-52. The specification also notes that the methods of the claimed invention allow for the creation of “LMWH preparations with low batch-batch variabil*625ity and a desired structural signature,” id. at col. 60 l.66-col. 61 l.3. It compares the claimed method of conducting a structural characterization of LMWHs with the prior art “current manufacturing practices for ... LMWHs [which] use functional assays ... and gross physical characterization to provide quality control,” id. at col. 48 ll. 1-7.
As the majority characterizes it, “ ‘ma[king]’ does not extend to testing to determine whether an already synthesized drug product possesses existing qualities or properties.” Maj. Op. at 615-16. While I do not agree with the majority’s cabining of the term “making,” even under the majority’s test, the quality control process is an integral part of the manufacturing of the enoxaparin drug product. The enoxaparin drug substance that is tested using the method of the '886 patent is far from being a finished product. The FDA defines a “drug product” as the “finished dosage form, for example, tablet, capsule, solution, etc.” 21 C.F.R. § 210.3(4) (2015). Even after the identity of the drug substance is confirmed utilizing the quality control steps of the '886 patent, further processing steps remain: “weighing, combining the enoxaparin in one batch with other batches of enoxaparin that have been similarly processed and selected by use of the claimed method, compounding the resulting mixture with specially-purified water, sterilizing this compound, placing it into syringes, and labeling and packaging the finished product.” J.A. 12440. Only after these additional processing steps are completed is the drug product ready for commercial sale. See 21 C.F.R. § 210.3(4). Thus, the quality control testing method of the '886 patent is a necessary intermediate step in the manufacture of enoxaparin.
In this respect this case is similar to Bio-Technology, where we considered whether a manufacturer’s importation of human growth hormone (“hGH”) could infringe two Genentech patents under § 271(g). 80 F.3d at 1558. The first patent was a method of producing hGH in bacterial hosts by inserting a semi-synthetic gene (e.g., a “plasmid”), encoding for hGH and one additional amino acid, into bacterial cells that could then express the hGH product. Id. at 1556-57. The second patent’s claims were directed to the method for constructing a plasmid, in other words, a method for creating information that the bacterial cells could use to generate the product. Id. at 1557. Notably, there was no doubt that the “plasmid product of the claimed process and hGH are entirely different materials.” Id. at 1561. Nonetheless, we noted that the manufacturer “use[d] the claimed process of making a [plasmid] as an essential part of an overall process for producing hGH,” and held that “it cannot be said as a matter of law that the production of hGH is too remote from the claimed process of making a replicable cloning vehicle.” Id.
In Bio-Technology, the practice of the plasmid patent was an essential intermediate component of the overall process for producing hGH. Similarly, here, the quality control step of the '886 patent is an essential intermediate step in the overall production of enoxaparin. In this case, the majority states that “[n]o assertion is made ... that the enoxaparin samples on which tests are performed are themselves incorporated into the finished product or imported into the United States.” Maj. Op. at 616-17. But this was also true in Bio-Technology, and provides no ground for distinction.
Ill
Finally, limiting “made” in § 271(g) to “the creation or transformation of a product, such as by synthesizing, combining components, or giving raw materials new *626properties,” Maj. Op. at 615-16, would lead to anomalous results. Patents on purification methods or the quality control method at issue here, which may be integral to the regulatory or commercial viability of a product, but which do not create or transform' a product, combine components, or confer new properties, could be freely infringed simply by outsourcing those processes abroad. Congress could not have intended to create this loophole when it sought to protect process patent owners from foreign competitors using U.S. manufacturing processes abroad. See generally Eli Lilly & Co. v. Am. Cyanamid Co., 82 F.3d 1568, 1571-72 (Fed.Cir.1996).
I respectfully dissent.
. The majority also determined that Amphas-tar does not infringe under § 271(g). This has little practical consequence since the majority holds that the § 271(e)(1) safe harbor does not shield Amphastar from liability under § 271(a).
However, the parties dispute whether § 271(g) can apply to products made in the United States. While the primary purpose of § 271(g) was to impose infringement liability for products shipped to the United States but made abroad by a United States patented process, the plain language of § 271(g) admits of no such geographic limitation. And the legislative history is clear that § 271(g) in- , eludes situations where the process is practiced in the United States. As the Senate Judiciary Committee report stated, "the process patent bill was crafted to apply equally to the use or sale of a product made by a process patented in this country whether the product was made ... in this country or in a foreign country.” S.Rep. No. 100-83, at 46 (1987).
The cases on which Amphastar relies as suggesting that the statute is limited to practicing a process abroad hold no more than that § 271(g) applies to that circumstance. See, e.g., Ajinomoto Co. v. Archer-Daniels-Midland Co., 228 F.3d 1338, 1347 (Fed.Cir.2000). They do not suggest that the sale of a product made by the practice of a process in the United States would not be an infringement under § 271(g).
. Sharafabadi v. University of Idaho, No. C09-1043JLR, 2009 WL 4432367 (W.D.Wash. Nov. 27, 2009), relied on by the majority, is similar to Bayer and is equally beside the point. Maj. Op. at 616. In Sharafabadi, the district court found that the patent holder "alleg[ed] only that the Universities used the [patent] as a research tool to test the characteristics of various yellow mustard seeds” in the course of developing a new IdaGold mustard seed and "[did] not allege that [any defendant] used the [patent] to directly manufacture or produce the IdaGold seeds.” Sharafabadi, 2009 WL 4432367, at *5 (citing Bayer, 340 F.3d at 1378).