United States of America v. Yu Xue, a/k/a JOYCE UNITED STATES OF AMERICA

PRECEDENTIAL
UNITED STATES COURT OF APPEALS
FOR THE THIRD CIRCUIT
_____________
Nos. 21-2227 & 21-2228
_____________
UNITED STATES OF AMERICA,
Appellant in No. 21-2227
v.
YU XUE, a/k/a JOYCE
____________
UNITED STATES OF AMERICA,
Appellant in No. 21-2228
v.
TAO LI
____________
On Appeals from the United States District Court
for the Eastern District of Pennsylvania
(D.C. Nos. 2:16-cr-00022-001 & 2:16-cr-00022-002)
District Judge: Honorable Joel H. Slomsky
____________
Argued: June 8, 2022

-- 1 of 18 --

2
____________
Before: CHAGARES, Chief Judge, AMBRO and FUENTES,
Circuit Judges
(Opinion filed: August 2, 2022)
____________
Robert J. Livermore [ARGUED]
Robert A. Zauzmer
Office of United States Attorney
615 Chestnut Street
Suite 1250
Philadelphia, PA 19106
Counsel for Appellant
Michael Dearington [ARGUED]
Peter Zeidenberg
ArentFox Schiff
1717 K Street, N.W.
Washington, DC 20006
Counsel for Appellee Yu Xue
John N. Joseph [ARGUED]
Abraham J. Rein
Post & Schell
1600 John F. Kennedy Boulevard
Four Penn Center, 14th Floor
Philadelphia, PA 19103
Counsel for Appellee Tao Li

-- 2 of 18 --

3
_____________
OPINION OF THE COURT
_____________
CHAGARES, Chief Judge.
Defendant Yu Xue was formerly a scientist at
pharmaceutical company GlaxoSmithKline (“GSK”). While
still employed at GSK, Xue formed a pharmaceutical company
with defendant Tao Li and others and proceeded to steal a
number of GSK documents containing trade secrets. Xue and
Li each pled guilty to a single count of conspiracy to steal trade
secrets, in violation of 18 U.S.C. § 1832(a)(5). The
Government, at sentencing, requested that the District Court
apply an enhancement under the United States Sentencing
Guidelines (the “Guidelines” or “U.S.S.G.”) based on the
“loss” attributable to the defendants’ conduct. While it is
undisputed that GSK did not suffer an actual monetary loss as
a result of the trade secret theft, under the commentary to
U.S.S.G. § 2B1.1, the definition of “loss” also includes losses
that the defendants intended. The District Court, however,
declined to apply an enhancement based on the intended loss
amount, finding that the Government failed to establish the
defendants purposely sought to inflict a pecuniary harm on the
victim, GSK.
The Government appeals and challenges the District
Court’s decision not to apply the enhancement. Because we
conclude that the District Court did not err in interpreting the
Guidelines or in its factual findings, we will affirm.

-- 3 of 18 --

4
I.
Until Xue’s arrest in 2016, she was a scientist employed
by GSK at its research facility in Upper Merion, Pennsylvania.
GSK is a global pharmaceutical company that researches and
develops vaccines, medicines, and consumer healthcare
products. While Xue was still a GSK employee, she formed a
pharmaceutical company in China called Renopharma with Li
and others. Xue stole approximately 200 GSK documents,
some of which contained trade secrets, and sent them to her co-
conspirators for the benefit of Renopharma. The stolen
documents concerned pharmaceutical products under
development, research data, and development and
manufacturing processes. At the time of the defendants’
arrests, Renopharma had neither made a profit nor developed
or sold any products using GSK’s information.
The defendants were charged in a superseding
indictment alleging that they engaged in a scheme to steal trade
secrets from GSK. Pursuant to plea agreements with the
Government, both Xue and Li pled guilty to a single count of
conspiracy to steal trade secrets, in violation of 18 U.S.C. §
1832(a)(5). The plea agreements specifically provided that
there was “no agreement between the parties as to fraud loss
under the Sentencing Guidelines” and that “the Court will
determine the fraud loss . . . prior to imposing a sentence.”
Appendix (“App.”) 102, 198.
After accepting the defendants’ guilty pleas, the District
Court held a three-day evidentiary hearing to determine the
“loss” amount attributable to the defendants’ conduct under the

-- 4 of 18 --

5
Guidelines. We set forth a summary below of the evidence
introduced at the hearing.1
The Government’s appeal focuses on three categories of
stolen information related to HER3 monoclonal antibodies,
GSK development platforms, and specific GSK products under
development. While at GSK, Xue researched a monoclonal
antibody known as HER3, which was intended to treat certain
forms of cancer.2 Some documents created by Renopharma
also discussed HER3 products. According to the Government,
one of those documents showed that the defendants deleted
references to GSK and replaced them with references to
Renopharma. The defendants maintain, however, that they
never intended to develop a HER3 product.
The next type of stolen information concerned three
GSK platforms. A platform is a set of standardized procedures
1 The Government called five witnesses: (1) FBI Special Agent
Andrew Haugen; (2) Dr. Joseph Tarnowski, Senior Vice
President of GSK; (3) Dr. Joseph Villafranca, a scientific
expert; (4) Dr. Chester Meyers, a second scientific expert; and
(5) Dana Trexler, an economic expert. The defendants called
two witnesses: (1) Dr. Jeffrey Field, a scientific expert; and (2)
Dr. David Blackburn, an economic expert. The court also
received a number of documents as exhibits during the
evidentiary hearing, including documents stolen from GSK,
emails and documents prepared by the defendants, and expert
reports on the amount of loss and nature of the trade secrets.
2 Monoclonal antibodies are proteins that are made in a
laboratory and used to treat “a variety of serious diseases,
including cancer, asthma, autoimmune disorders, and
Alzheimer’s disease.” Gov’t Br. 8–9.

-- 5 of 18 --

6
used by GSK to develop monoclonal antibodies and other
products. GSK invested substantial time and money in
developing the platforms. According to the Government, the
platforms could help Renopharma develop and manufacture its
own monoclonal antibodies and reduce the cost of doing so.
And Renopharma could benefit from the platforms even if it
did not seek to develop the same products as GSK.
Lastly, the stolen documents contained information on
approximately twelve GSK products under development. This
included Investigational New Drug Applications, which are
non-public documents submitted to the United States Food and
Drug Administration in order to initiate clinical trials and that
contain detailed information on product design and testing.
The parties press widely different views regarding the
importance and nature of the information taken from GSK.
The Government asserts that Renopharma pursued various
strategies involving GSK’s information, such as marketing
products and soliciting investments using the information.
The defendants’ scientific expert, however, testified
that he did not see evidence of Renopharma using the GSK
information to research or develop any competing products.
The defendants further point to Renopharma’s annual report
and Li’s post-arrest statement as evidence that Renopharma
was developing products different from GSK’s targets.
Regarding the nature of the stolen information, the defendants’
scientific expert testified that much of it was publicly available
through well-known scientific textbooks or GSK’s patent
applications.

-- 6 of 18 --

7
The parties also disputed the value of the stolen
information for purposes of calculating the intended loss. The
Government’s economic expert witness asserted that the
intended loss amount exceeded $1 billion. The defendants,
conversely, argued that it should be $0.
On September 22, 2020, the District Court issued an
opinion finding that the Government failed to establish that the
defendants purposely sought to inflict a loss on the victim,
GSK. It therefore found the intended loss amount to be $0 and
declined to apply an enhancement under § 2B1.1 of the
Guidelines. The court found in the alternative that even if the
Government had established that the defendants intended to
harm GSK, the defendants had sufficiently rebutted the
Government’s proposed valuation.
The District Court subsequently determined that Xue’s
Guidelines range, without the enhancement, was 12 to 18
months of imprisonment. Xue was ultimately given a sentence
of 8 months of imprisonment. With respect to Li, the court
determined that the Guidelines range was 10 to 16 months, and
it imposed a sentence of time served, or 59 days of
imprisonment.3 The Government timely appealed the District
Court’s sentencing determinations.
3 The Government’s intended loss calculation, if adopted by
the District Court, would have resulted in a Guidelines range
of life imprisonment for each of the defendants. The
Government ultimately requested the court to impose the
maximum sentences permitted under the defendants’ plea
agreements, which were terms of imprisonment of ten years for
Xue and seven years for Li.

-- 7 of 18 --

8
II.
The District Court had jurisdiction under 18 U.S.C. §
3231, and we have appellate jurisdiction under 28 U.S.C. §
1291 and 18 U.S.C. § 3742(b). We exercise plenary review
over a district court’s interpretation of the Sentencing
Guidelines and review its factual determinations for clear error.
United States v. Kirschner, 995 F.3d 327, 333 (3d Cir. 2021).
III.
The Government argues that the District Court, in
declining to apply a sentencing enhancement based on the
intended loss amount, committed a procedural error by failing
to calculate the value of the stolen trade secrets. The
Government further contends that the District Court failed to
consider properly the nature of offenses involving the theft of
trade secrets and their treatment under the Guidelines. We first
set forth below the legal principles relevant to applying the
Guidelines enhancement at issue and then consider the
defendants’ arguments in light of those principles.
A.
For offenses involving fraud or theft, § 2B1.1 of the
advisory Guidelines provides for an offense level enhancement
based on the value of the “loss” attributable to the defendant’s
conduct. The defendant’s base offense level, which is used to
calculate a recommended range of imprisonment, increases as
the loss resulting from the offense increases. See U.S.S.G. §
2B1.1(b)(1). The maximum increase is a 30-level
enhancement for conduct resulting in a loss exceeding $550
million. Id. The Government bears the burden of establishing

-- 8 of 18 --

9
the amount of loss under § 2B1.1 by a preponderance of the
evidence. United States v. Free, 839 F.3d 308, 319 (3d Cir.
2016); accord United States v. Diallo, 710 F.3d 147, 151 (3d
Cir. 2013). But the Guidelines do not mandate that a district
court find a loss amount of greater than $0. Free, 839 F.3d at
323 (“We agree with the proposition that the government is not
entitled to a punitive loss calculation, even in cases involving
fraud, absent evidence of actual or intended pecuniary loss.”).
The Guidelines’ commentary details how a district court
should calculate the amount of loss. U.S.S.G. § 2B1.1 cmt.
n.3. The general rule defines “loss” as “the greater of actual
loss or intended loss.” Id. at cmt. n.3(A).4 There is no dispute
that GSK’s actual loss here was $0. The District Court was
therefore required to calculate the intended loss, which the
application note defines, in the pertinent part, as “the pecuniary
harm that the defendant purposely sought to inflict . . . .” Id. at
cmt. n.3(A)(ii). The commentary makes clear that “pecuniary
4 Principles of administrative law inform “[t]he extent to which
the guidelines’ commentary controls our interpretation of the
guidelines.” United States v. Nasir, 17 F.4th 459, 469 (3d Cir.
2021) (en banc). Discussing these principles in light of Kisor
v. Wilkie, ––– U.S. –––, 139 S. Ct. 2400 (2019), we noted
commentary that “expanded and did not merely interpret” the
Guidelines may not be entitled to deference. Nasir, 17 F.4th at
470–71. We have previously suggested that the commentary
to U.S.S.G. § 2B1.1 defining intended loss might sweep more
broadly than the Guideline itself but declined to address the
issue. See Kirschner, 995 F.3d at 333. Because neither the
Government nor the defendants argue that the commentary to
U.S.S.G. § 2B1.1 defining intended loss should not apply, we
will similarly not address the issue.

-- 9 of 18 --

10
harm” includes only “harm that is monetary or that otherwise
is readily measurable in money,” and not “non-economic
harm.” Id. at cmt. n.3(A)(iii). In calculating the amount of
loss, a district court “need only make a reasonable estimate of
the loss,” and on appeal, we must give its analysis “appropriate
deference.” Id. at cmt. n.3(C) (citing 18 U.S.C. §§ 3742(e),
(f)). The commentary sets forth six factors that the court, “as
appropriate and practicable under the circumstances,” should
use in calculating the intended loss amount. Id. Two of those
factors are relevant here: the fair market value of the
information taken and the cost of developing the information.
Id. at cmt. n.3(C)(i)–(ii).
We recently reaffirmed that a district court is required
to “conduct a ‘deeper analysis’ before inferring that a
defendant intended to cause a particular loss” for purposes of
§ 2B1.1. Kirschner, 995 F.3d at 337; see also Diallo, 710 F.3d
at 151–52. This requirement applies to “any loss methodology
the government elects to adopt.” Kirschner, 995 F.3d at 337.
Our decision in United States v. Kirschner illustrates this
deeper analysis. The district court in Kirschner calculated the
amount of the loss from the defendant’s intended sales of
counterfeit coins. Id. at 331–32. The vast majority of the
intended loss amount was based on the fair market value of six
types of rare coins. Id. The Government’s intended-loss
calculation, adopted by the district court, multiplied the
estimated fair market value of these rare coins by the historical
markdown at which the defendant sold other more common
counterfeit coins. Id. We vacated the sentence because the
district court “never found that [the defendant] intended to sell
the coins as counterfeits for the prices the government
claimed.” Id. at 335. In reaffirming that a district court must
perform a “deeper analysis,” we noted that the Government

-- 10 of 18 --

11
had not presented evidence that the defendant was aware of the
value of the rare coins or would actually be able to sell them in
a manner similar to his past sales of much more common coins.
Id. at 336 (“And again, the principal question is not whether
[the defendant] could have sold the high-value counterfeits at
the prices claimed by the government. The question is whether
he intended to.”).
B.
The District Court interpreted the Guidelines’
“definition of intended loss [to] include[] the mens rea
requirement that the defendant ‘purposefully sought to inflict’
pecuniary harm on the victim.” App. 972 (quoting U.S.S.G. §
2B1.1 cmt. n.3(A)(ii)). Because the government failed to make
this showing, and instead, “attempted to prove only the
development cost of the stolen information and its fair market
value,” the District Court declined to apply an enhancement
under § 2B1.1. App. 942.
The Government suggests that the District Court was
required “to calculate the loss, based on either development
costs or [the defendants’] intended gain.” App. 29.5 By failing
to do so, the Government asserts that the court committed a
procedural error. We disagree and conclude that it properly
5 The defendants briefly claim that the Government “waived”
any argument that it satisfied the “purposely sought to inflict
pecuniary harm standard.” Def’s Br. 39 (quotation marks
omitted). This “waiver” argument amounts to a bare
conclusion, and we therefore will consider the Government’s
argument on appeal.

-- 11 of 18 --

12
interpreted the Guidelines’ commentary defining intended
loss.
The plain language of the commentary to § 2B1.1 limits
the definition of intended loss to “the pecuniary harm that the
defendant purposely sought to inflict.” U.S.S.G. § 2B1.1 cmt.
n.3(A)(ii). The District Court’s task was therefore to determine
whether the defendants purposely sought to inflict a loss on the
victim in the amount claimed by the government. Kirschner,
995 F.3d at 336–37 (noting that a district court fails to perform
a “deeper analysis” if “it adopts an intended-loss methodology
without demonstrating that the defendant’s ‘purpose’ was to
inflict the losses the government claims he intended to inflict”);
see also United States v. Yeaman, 194 F.3d 442, 460 (3d Cir.
1999) (“Intended loss refers to the defendant’s subjective
expectation, not to the risk of loss to which he may have
exposed his victims.”). By arguing that the District Court erred
in failing to value the trade secrets, the Government
presupposes that the defendants intended to inflict losses on
GSK equal to the cost of developing the trade secrets or their
fair market value. The District Court found that the
Government failed to establish that the defendants had the
purpose to inflict a pecuniary loss on GSK, and accordingly,
we hold that the court did not err in declining to value the trade
secrets.
Although the defendant in Kirschner challenged the
valuation methodology used to calculate the amount of loss,
we explained that in conducting a “deeper analysis,” a district
court must also find that the defendant had the required mental
state to apply an enhancement based on intended loss. See 995
F.3d at 337 (“[A] district court must conduct a ‘deeper
analysis’ to make sure the defendant purposely sought to inflict

-- 12 of 18 --

13
each component of the losses the government claims he
intended to inflict.”). The record must support a finding that
the defendant’s purpose was to inflict a pecuniary loss on the
victim to apply this enhancement. Id. at 337–38. And as we
explained in Kirschner, direct evidence of the defendant’s
mental state is not required; a district court “is free to make
reasonable inferences about the defendant’s mental state from
the available facts.” Id. at 337.
Only one other Court of Appeals’ decision has
addressed the application of U.S.S.G. § 2B1.1 and the analysis
for intended loss in the theft of trade secrets context: United
States v. Yihao Pu, 814 F.3d 818 (7th Cir. 2016). The
defendant in Pu worked at two high-frequency trading firms
and stole files from those firms’ proprietary trading algorithms.
Id. at 821–22. After the defendant pled guilty to offenses
involving the theft of trade secrets, the District Court found that
the defendant’s intended loss amount was the cost to develop
the high frequency trading algorithms. Id. at 822–23. The
Court of Appeals for the Seventh Circuit disagreed. While
recognizing that the Guidelines’ commentary permits a district
court to estimate intended loss in a trade secret case by
considering the cost of development, the court noted that this
alone does not establish that the cost of development is the
correct loss figure absent evidence of the defendant’s intent to
cause such a loss. See id. at 826. It identified the core issue as
“whether the record supports a finding that it was more likely
than not that [the defendant] intended to cause a loss to the
victims that equaled the cost of development.” Id. Because
there was no evidence that the defendant intended to inflict a
loss on the victim equal to the cost to develop the stolen trade
secrets, the court vacated the defendant’s sentence. Id. at 826–
27.

-- 13 of 18 --

14
We agree with the framework set forth by the Court of
Appeals for the Seventh Circuit in Pu.6 The commentary to §
2B1.1 defining intended loss directs a district court to estimate
loss based on considerations such as the cost of developing the
trade secrets and the fair market value of the information.
U.S.S.G. § 2B1.1 cmt. n.3(C). But absent evidence of a
defendant’s purpose to inflict a pecuniary loss equal to the
stolen information’s cost of development or its fair market
value, a district court cannot find that intended loss equals
these amounts.
We conclude that the District Court did not err by
declining to value the stolen trade secrets where the
Government failed to establish that the defendants had the
required mental state for the enhancement based on intended
loss.
C.
The Government next raises a variety of challenges to
the District Court’s intended-loss analysis, arguing that it
incorrectly considered the nature of offenses involving the
theft of trade secrets and their treatment under the Guidelines.
The Government first asserts that trade secrets have
“independent economic value” that, when stolen, deprive the
owner of its exclusive ability to control the information,
diminish the comparative value of its investment, and
6 The Government attempts to distinguish Pu on the ground that
the defendant’s conduct in that case was factually dissimilar to
that of Xue and Li. But the Government offers no persuasive
reason as to why the legal framework in Pu is incorrect.

-- 14 of 18 --

15
demonstrate to others that it is unable to protect trade secrets.
Gov’t Br. 32–33. While the fact that the object of the theft is a
trade secret may factor into the analysis, we disagree that the
District Court erred by failing to account for the trade secrets’
“independent economic value.” What counts under the
Guidelines’ commentary is loss in the form of pecuniary harm.
See U.S.S.G. § 2B1.1 cmt. n.3(a)(iii) (defining “pecuniary
harm”); see also Free, 839 F.3d at 321. Harms such as loss of
exclusive control of trade secrets or publicity regarding a
company’s inability to protect trade secrets, without more, do
not constitute “pecuniary harm.” And contrary to the
Government’s assertion that the “theft of a trade secret by a
would-be competitor necessarily exposes the victim to market
losses such as lost profits, lost royalties, or diminished share
price,” Gov’t Br. 34, a potential for pecuniary harm standing
alone likewise does not establish an intended loss, see Diallo,
710 F.3d at 153 (noting that a district court errs when it equates
“potential loss and intended loss without deeper analysis”
(cleaned up)).
The Government relatedly argues that if there is no
finding of loss in this case, then an enhancement based on
intended loss would never apply in offenses involving the theft
of trade secrets.7 The Government’s contention is better
understood as a disagreement with the District Court over the
proper inferences to be drawn from the record. While the
7 At oral argument, the defendants disputed this point and
suggested hypothetical cases that might raise the inference that
an individual who steals trade secrets also intends to inflict a
pecuniary harm on the victim, such as publishing the trade
secrets on the internet for others to copy or using the secrets to
compete directly against the victim.

-- 15 of 18 --

16
Government maintains that the defendants intended to develop
drugs based on the GSK information or to use the information
to solicit investment, as discussed below, the District Court did
not clearly err in finding that the record failed to support such
an inference.
The Government next argues that proof of the
defendants’ intent to gain from the theft of trade secrets
establishes that the defendants intended to inflict a pecuniary
harm on GSK. Gov’t Br. 34.8 The District Court, however,
considered evidence of how the defendants used, and intended
to use, the stolen information and found that the record did not
support an inference of an intent to inflict a pecuniary harm on
GSK. This finding was not clearly erroneous.
While the defendants took information for the benefit of
their pharmaceutical company, the District Court found that the
defendants did not in fact use the trade secrets to compete with
GSK. It found that the defendants did not use the information
to solicit investment, develop competing products, or develop
the same products that GSK was developing. The District
Court, for example, found that the “financial statements” used
for soliciting investment in Renopharma were mere puffery
and not credible evidence of a realistic expectation of gain.
App. 981–82. It further noted that an email sent by Li to solicit
8 The commentary to § 2B1.1, relatedly, provides that “[t]he
court shall use the gain that resulted from the offense as an
alternative measure of loss only if there is a loss but it
reasonably cannot be determined.” U.S.S.G. § 2B1.1 cmt.
n.3(B). But we do not see how this provision could apply here
absent evidence that the defendants experienced any gains
resulting from their Renopharma scheme.

-- 16 of 18 --

17
investment in Renopharma “did not show that it was more
likely than not that [the] Defendants were shopping a GSK-
developed product, rather than their own, independent
creation, or simply engaging in puffery to solicit investment,
which appears to have been Renopharma’s modus operandi.”
App. 983. We see no clear error in the District Court declining
to infer intended loss from evidence of the defendants’
intended gain.9
The Government finally argues that the defendants
admitted during their plea allocutions that they acted with the
required mental state for applying the enhancement based on
the intended loss. Both Xue and Li pled guilty to a single count
of conspiracy to steal trade secrets, in violation of 18 U.S.C. §
1832(a)(5). This statute, in the relevant part, prohibits an
individual who, “with intent to convert a trade secret . . . to the
economic benefit of anyone other than the owner thereof, and
intending or knowing that the offense will, injure any owner of
9 To the extent the District Court suggested that in the context
of theft of trade secrets a defendant’s intended gain can never
further an inference of intended loss, we disagree with this
suggestion. An inference of intent, for purposes of the
intended loss analysis, may be based on the defendant’s
conduct and purpose in committing the offense. See Kirschner,
995 F.3d at 337 (“In conducting [the deeper] analysis, the court
is free to make reasonable inferences about the defendant’s
mental state from the available facts.”); cf. United States v.
Feldman, 338 F.3d 212, 223 (3d Cir. 2003) (identifying that a
district court “must look at what [the defendant] sought to gain
from committing the crime” as part of the intended loss
analysis). This disagreement, however, does not alter the result
herein.

-- 17 of 18 --

18
that secret, knowingly . . . conspires” to steal trade secrets. 18
U.S.C. § 1832(a)(5) (emphasis added).
The Government’s argument fails to account for
important differences between the elements of the offense and
the required mental state for purposes of the analysis for
intended loss. A defendant who intends or knows that his or
her conduct will injure the owner of the trade secret does not
necessarily intend to inflict pecuniary harm on that secret’s
owner. Cf. Pu, 814 F.3d at 828 (“The statute of conviction, 18
U.S.C. § 1832, does not explicitly require an economic loss to
the victim.”). The Government itself identifies examples of
potential non-pecuniary injuries that result from the theft of
trade secrets, such as loss of the exclusive use of the
information and the possible public disclosure that a company
cannot protect the information. A second element of the
offense, “inten[ding] to convert a trade secret . . . to the
economic benefit of anyone other than the owner thereof,” 18
U.S.C. § 1832(a), likewise does not establish an intent to inflict
pecuniary harm on GSK. The District Court, as explained
above, did not credit the Government’s theories that
Renopharma was developing competing products or soliciting
investment based on the GSK information. We conclude that
the District Court did not clearly err in finding that the
defendants’ plea allocutions failed to establish the required
intent for purposes of the intended loss analysis.
IV.
For the foregoing reasons, we will affirm the judgments
of the District Court.

-- 18 of 18 --