WYNN, Circuit Judge

ARGUED: Gregg S. Levin, MOTLEY RICE LLC, Mount Pleasant, South Carolina, for Appellants. Jamie L. Wine, LATHAM & WATKINS, New York, New York, for Appellees. ON BRIEF: Aaron S. Book, WEBSTER BOOK LLP, Alexandria, Virginia; Christopher F. Moriarty, MOTLEY RICE LLC, Mount Pleasant, South Carolina; John C. Browne, Lauren A. Ormsbee, Jesse L. Jensen, BERNSTEIN LITOWITZ BERGER & GROSSMANN LLP, New York, New York, for Appellants. Kevin M. McDonough, New York, New York, Melissa Arbus Sherry, Stephen P. Barry, Margaret A. Upshaw, LATHAM & WATKINS LLP, Washington, D.C., for Appellees.

Before WYNN, THACKER, and RUSHING, Circuit Judges.

Affirmed by published opinion. Judge Wynn wrote the opinion, in which Judge Thacker and Judge Rushing joined.

WYNN, Circuit Judge Plaintiffs KBC Asset Management NV and Arbejdsmarkedets Tillaegspension appeal the dismissal of their class action suit alleging securities fraud under Sections 10(b) and 20(a) of the Securities Exchange Act of 1934 and a regulation promulgated thereunder known as Rule 10b-5 against Defendants DXC Technology Company and its two principal executives, J. Michael Lawrie and Paul N. Saleh. Specifically, Plaintiffs allege that they purchased shares of DXC at inflated prices after DXC, Lawrie, and Saleh made false and misleading statements concerning DXC's financial health.

The district court dismissed their complaint, ruling that Plaintiffs failed to allege that Defendants made actionable false and misleading statements and failed to allege facts leading to the strong inference that Defendants acted with the requisite scienter. We affirm.

I.

In reviewing the district court's dismissal under Federal Rule of Civil Procedure 12(b)(6), "we accept all factual allegations in the complaint as true." Yates v. Mun. Mortg. & Equity, LLC , 744 F.3d 874, 881 (4th Cir. 2014) (internal quotation marks omitted) (quoting Matrix Cap. Mgmt. Fund, LP v. BearingPoint, Inc. , 576 F.3d 172, 176 (4th Cir.2009) ).

DXC is a publicly traded information-technology company formed in 2017 from a merger of Computer Science Corporation and Hewlett Packard Enterprise Company. The new company initially succeeded in meeting its strategic financial goals by instituting cost-cutting measures, and on February 8, 2018, it issued a press release announcing its continued financial success. Soon, however, the company found itself needing to revise its projected revenue guidance to shareholders downward by an estimated $800 million, a decision it announced on November 6 of the same year. As a result, DXC's shareholders incurred losses when its stock price decreased following that announcement. Plaintiffs represent a class of shareholders who purchased or otherwise acquired DXC stock from February 8, 2018 through November 6, 2018.

Plaintiffs filed suit alleging violations of Sections 10(b) and 20(a), 15 U.S.C. §§ 78j(b), 78t(a) and Rule 10b-5, 17 C.F.R. § 240.10b-5. In their complaint, Plaintiffs allege that Defendants knew the cost-cutting measures implemented in 2018 undermined DXC's ability to generate revenue and that this was contrary to information the Defendants were telling the public. As such, the Plaintiffs allege the Defendants fraudulently induced them to purchase or acquire stock in DXC by making material misstatements and omissions regarding the financial health of the company and that they did so with the requisite scienter for such fraud.

The Defendants successfully moved to dismiss the complaint pursuant to Rule 12(b)(6). In re DXC Tech. Co. Sec. Litig. , No. 1:18cv01599, 2020 WL 3456129, at *1, 13 (E.D. Va. June 2, 2020). The district court determined that the statements issued by DXC or made by its employees were either forward-looking statements protected under the safe-harbor provision of the Private Securities Litigation Reform Act of 1995 ("PSLRA"), 15 U.S.C. § 78u-5, or non-actionable puffery. See id. at *6–10. Further, the district court concluded that the Plaintiffs' complaint, viewed as a whole, did not contain factual allegations sufficient to give rise to the "strong inference" of scienter required by the PSLRA, 15 U.S.C. § 78u-4(b)(2)(A), and applicable precedent. See id. at *11–13. The Plaintiffs timely appealed.

II.

We review de novo the district court's determination that the Plaintiffs' complaint failed to state a claim for securities fraud. Singer v. Reali , 883 F.3d 425, 437 (4th Cir. 2018). In reviewing the dismissal, "we accept all factual allegations in the [c]omplaint as true, and we consider the [c]omplaint in its entirety." Id. We draw all reasonable inferences in favor of the Plaintiffs. Id. We may also take judicial notice of the content of relevant Securities and Exchange Commission ("SEC") filings and other publicly available documents included in the record. See In re PEC Sols., Inc. Sec. Litig. , 418 F.3d 379, 390 n.10 (4th Cir. 2005).

To be actionable, fraud claims brought under Section 10(b) must satisfy six elements: "(1) a material misrepresentation or omission by the defendant; (2) scienter; (3) a connection between the misrepresentation or omission and the purchase or sale of a security; (4) reliance upon the misrepresentation or omission; (5) economic loss; and (6) loss causation." Singer , 883 F.3d at 438 (quoting Stoneridge Inv. Partners, LLC v. Sci.-Atlanta, Inc. , 552 U.S. 148, 157, 128 S.Ct. 761, 169 L.Ed.2d 627 (2008) ). These elements can be addressed in any order, and the failure to adequately allege scienter is enough to doom the claim. See In re PEC Sols. , 418 F.3d at 388 n.6. We train our analysis in this appeal only on this second element, scienter.

The requirements for pleading scienter in a securities fraud claim are set forth in the PSLRA. 15 U.S.C. § 78u-4(b)(2). When enacting the PSLRA, Congress imposed a heightened pleading requirement for the element of scienter "[a]s a check against abusive litigation by private parties" in securities fraud actions. Tellabs, Inc. v. Makor Issues & Rts., Ltd. , 551 U.S. 308, 313, 127 S.Ct. 2499, 168 L.Ed.2d 179 (2007). Thus, the PSLRA mandates that, "with respect to each act or omission alleged " to constitute securities fraud, any prospective plaintiff must "state with particularity facts giving rise to a strong inference that the defendant acted with the required state of mind." 15 U.S.C. § 78u-4(b)(2)(A) (emphases added). In addition to analyzing each act or omission alleged, "we ultimately evaluate [the] plaintiff['s] allegations of scienter holistically" and afford them "the inferential weight warranted by context and common sense." Yates , 744 F.3d at 885 (4th Cir. 2014) (quoting Matrix Cap. , 576 F.3d at 183 ).

The "required state of mind" under Section 10(b) is "a mental state embracing intent to deceive, manipulate, or defraud." Yates , 744 F.3d at 884 (first quoting 15 U.S.C. § 78u-4(b)(2) ; then quoting Tellabs , 551 U.S. at 319, 127 S.Ct. 2499 ). To satisfy this standard at the motion to dismiss stage, a complaint "alleging either intentional or severely reckless conduct is sufficient." Id. "Recklessness is ‘an act so highly unreasonable and such an extreme departure from the standard of ordinary care as to present a danger of misleading the plaintiff.’ " Maguire Fin., LP v. PowerSecure Int'l, Inc. , 876 F.3d 541, 547 (4th Cir. 2017) (quoting Ottmann v. Hanger Orthopedic Grp. , 353 F.3d 338, 343 (4th Cir. 2003) ).

Further, the inference of scienter must be "cogent and compelling." Id. (quoting Tellabs , 551 U.S. at 324, 127 S.Ct. 2499 ). This is "necessarily a comparative inquiry." Yates , 744 F.3d at 885. Courts must "compare the malicious and innocent inferences cognizable from the facts pled ... and only allow the complaint to survive a motion to dismiss if the malicious inference is at least as compelling as any opposing innocent inference." Id. (quoting Zucco Partners, LLC v. Digimarc Corp. , 552 F.3d 981, 991 (9th Cir. 2009) ).

Thus, the question before us is whether "the allegations in the complaint, viewed in their totality and in light of all the evidence in the record, allow us to draw a strong inference, at least as compelling as any opposing inference," that Defendants knowingly or recklessly defrauded investors by making false or misleading statements about DXC's financial health. Id. (quoting Pub. Emps.' Ret. Ass'n of Colo. v. Deloitte & Touche LLP , 551 F.3d 305, 313 (4th Cir. 2009) ). If the more compelling inference is that the Defendants merely acted negligently or non-fraudulently, we must affirm the district court's dismissal for lack of scienter.

III.

Plaintiffs rely on five categories of allegations that they claim demonstrate scienter. They allege that they can show Defendants acted with scienter based on (1) allegations made by a former executive of DXC; (2) statements of unnamed former DXC employees; (3) massive stock sales by Defendants Lawrie and Saleh during the class period; (4) the core-operations theory; and (5) the temporal proximity between the company's statements painting a sunny picture and its ultimate admission that it had been overly optimistic. We conclude that none of these categories individually gives rise to an inference of scienter that is as compelling as a non-fraudulent inference. See id. We then evaluate the allegations holistically, but conclude that even viewed together, the allegations cannot establish the requisite scienter. Accordingly, we affirm the district court's dismissal of the complaint.

A.

We begin with Plaintiffs' allegations related to a lawsuit that DXC's former Executive Vice President and Head of Global Delivery, Stephen Hilton, filed against DXC. According to Plaintiffs, Hilton's complaint in that lawsuit includes allegations that DXC and its executives knew of the falsity of the alleged misstatements and omissions at issue here. Hilton claimed he warned Lawrie that cutting costs at DXC harm customer satisfaction. DXC terminated Hilton noting, among other things, Lawrie's dissatisfaction with Hilton's performance, including his failure to adequately cut costs. Plaintiffs contend this complaint showed that Defendants deliberately deceived investors by making statements about the success of Hilton's division while knowing that they could not meet their revenue projections. But these allegations support only a weak inference of scienter. Perhaps DXC or Lawrie realized that Hilton's division as a whole was underperforming. But this inference is less persuasive than the competing inference that Hilton was fired because Hilton —not his division—was performing poorly due to his "material misconduct" and "substantial and willful failure to render services." J.A. 85 (internal quotation marks omitted).

Further, Hilton's allegations in his lawsuit tend to show only that Hilton believed the cuts he was being asked to make were misguided—not that Lawrie or Saleh, the executives in question here, agreed with that assessment. These allegations suggest a mere business disagreement among the executives, which does not amount to securities fraud. See Yates , 744 F.3d at 887 (finding employee statements supported "[t]he more plausible inference" of "an honest disagreement" over a business decision because the fact that the executives "were ultimately wrong is not enough to support an inference of scienter").

B.

Plaintiffs' second category of scienter allegations concerns the statements of unnamed former DXC employees. Plaintiffs offer statements from former employees who believed DXC was heading in the wrong direction before and during the class period. They rely on these allegations to establish that (1) the company's outlook was deteriorating, (2) Defendants were aware of that, and thus (3) they must have been intentionally deceiving investors when they made positive statements about the company's outlook.

Generally, plaintiffs are free to use the allegations of confidential witnesses to support an inference of scienter. See id. at 885–88. However, such allegations will only be afforded the weight they are due given their indicia of reliability, and any " ‘[o]missions and ambiguities count against’ an inference of scienter." Id. at 885–86 (quoting Tellabs , 551 U.S. at 326, 127 S.Ct. 2499 ).

Here, the allegations do not raise a strong inference of scienter because the former employees, for the most part, do not allege that they passed their concerns on to Lawrie or Saleh or that the individual Defendants were otherwise aware of the problems alleged. See Nolte v. Cap. One Fin. Corp. , 390 F.3d 311, 316 (4th Cir. 2004) (affirming dismissal of a complaint in part because the complaint did not allege "that management was ever informed of" the concerns at the heart of the shareholders' allegations). This general lack of direct contact with Defendants weakens the inference of scienter, as Lawrie and Saleh may have been unaware of the problems, the causes of the problems, or the extent of the problems. Cf. Metzler Asset Mgmt. GmbH v. Kingsley , 928 F.3d 151, 162 (1st Cir. 2019) (concluding that investors' allegations failed to establish a strong inference of scienter where plaintiffs did not allege that confidential witnesses spoke with the executives and the witnesses were "several levels removed from the company's executive team"). Further, to the limited extent the employees allege that they did notify Lawrie or Saleh of their concerns, Plaintiffs' allegations are vague and conclusory, and lack particularized facts demonstrating Defendants intentionally or recklessly misled investors. The far more plausible inference here is that there was a disagreement within DXC over the proper course forward for the company and that Defendants made a business decision to make cuts that their former employees found questionable. Even if those employees were ultimately correct that Defendants made unwise business decisions by erring too aggressively on the side of cutting costs, that mistake does not support a strong inference of scienter. See, e.g., In re Triangle Cap. Corp. Sec. Litig. , 988 F.3d 743, 754 (4th Cir. 2021) (holding there was no strong inference of scienter in light of a stronger competing inference that the "[d]efendants had an honest debate about the merits of a subjective business judgment, and in hindsight, simply made the wrong choice with some investments").

C.

Third, Plaintiffs raise the specter of insider trading to support their scienter allegations, pointing to Lawrie's and Saleh's sales of shares of DXC's stock during the class period. We are not persuaded.

"Insider trading allegations will only support an inference of scienter ‘if the timing and amount of a defendant's trading were unusual or suspicious.’ " Yates , 744 F.3d at 890 (quoting Tchrs.' Ret. Sys. of La. v. Hunter , 477 F.3d 162, 184 (4th Cir. 2007) ). "To determine whether an insider's sales were ‘unusual in scope’ we consider factors such as ‘the amount of profit made, the amount of stock traded, the portion of stockholdings sold, or the number of insiders involved.’ " Id. (quoting In re Suprema Specialties, Inc. Sec. Litig. , 438 F.3d 256, 277 (3d Cir. 2006), abrogated on other grounds by Tellabs , 551 U.S. at 322–23, 127 S.Ct. 2499 ).

Plaintiffs compare Lawrie's and Saleh's sales during the nine-month class period with their sales during the preceding nine months (the "control period"). According to Plaintiffs, while Lawrie did not sell any shares during the control period, he sold 110,549 shares—valued at over $10.2 million—during the class period. These sales amount to over 17% of Lawrie's holdings and were concentrated in August and September of 2018—shortly before DXC revised its revenue projections in early November. As for Saleh, Plaintiffs allege he sold 77% of his shares for more than $9.5 million in proceeds during the class period and that these sales were also concentrated in August and September. Plaintiffs acknowledge, however, that Saleh sold $14.8 million of stock during the control period.

Plaintiffs' allegations fairly give Defendants a strong motive: inflating DXC's stock and cashing out while the stock was high and before the company's poor performance was reported. Saleh sold a robust 77% of his shares. And while Lawrie did not sell a particularly high percentage of his shares, the 17% he did sell was arguably noteworthy because he did not sell any shares during the control period. Further, the fact that both Defendants sold a substantial amount of stock—valued at around $10 million apiece—is also somewhat unusual and suspicious. See id. And both executives concentrated their sales in the months before the unfavorable news broke about the company.

Nevertheless, the insider-trading allegations do not raise a strong inference of scienter, as they must to support Plaintiffs' claim. For a start, we did not previously find a strong inference of scienter from a CFO's "nearly de minimis " sale of 13% of his holdings in the company during the class period. In re PEC Sols. , 418 F.3d at 390 ; see also Cozzarelli v. Inspire Pharms. Inc. , 549 F.3d 618, 628 (4th Cir. 2008) (describing sales of "13%, 12%, and 3% of [the directors'] holdings, respectively" as "modest to de minimis"); Proter v. Medifast, Inc. , No. GLR-11-720, 2013 WL 1316034, at *21 n.20 (D. Md. Mar. 28, 2013) (collecting cases from this and other courts discussing similar or greater sales percentages that did not give rise to a strong inference of scienter). Here, Plaintiffs have not explained why Lawrie's slightly larger sale—17% of his holdings—would cause us to jump from labeling the 13% sale in In re PEC Solutions "nearly de minimis " to finding a strong inference of scienter here.

Moreover, Saleh's sale of $9.5 million worth of shares during the class period is less than the $14.8 million of stock he sold during the control period. When the sale of stock during the class period is not substantially out of line with the sale of stock during the control period, the inference of scienter is lessened. See Zucco Partners , 552 F.3d at 1005–06 (refusing to find an inference of scienter from insider trades made by defendants where plaintiffs failed to allege facts showing that the sales were inconsistent with defendant's usual trading practices). Here, not only were Saleh's sales during the class period not substantially greater than during the control period, they were, in fact, substantially smaller. That is, Plaintiffs ask us to draw a strong inference of scienter from that fact that Saleh sold much less stock during the period in which he was allegedly defrauding investors than during the period in which he is not alleged to have done so. We decline to draw such an inference.

Defendants contend that this cluster of allegations faces another hurdle because it is undisputed that both individual Defendants have at some point entered into Rule 10b5-1 plans. "Under Rule 10b5-1, corporate insiders can set up trading plans to sell company shares at predetermined times and amounts to avoid accusations of illegal insider trading." Yates , 744 F.3d at 891. "[I]t is an affirmative defense in insider trading cases that the defendant's purchases or sales were made pursuant to a ‘written plan for trading securities.’ " Id. (quoting 17 C.F.R. § 240.10b5-1(c) ). Thus, the fact that a defendant trades shares pursuant to such a plan "weakens any inference of fraudulent purpose." Id.

Indeed, the record includes SEC filings by Lawrie and Saleh noting the existence of 10b5-1 trading plans. And we may take judicial notice of those filings. See In re PEC Sols. , 418 F.3d at 390 n.10. The problem for Defendants, however, is that the record is silent as to when they entered their plans. When "neither side has pointed to any information about when ... [D]efendants entered into the 10b5-1 trading plans[,] [we] ... cannot say that the trading plans mitigate a suggestion of motive" for suspicious trading. Alaska Elec. Pension Fund v. Asar , 768 F. App'x 175, 182 (5th Cir. 2019). Nevertheless, we conclude that Plaintiffs have failed to allege a strong inference of scienter related to the suspicious sales for the reasons given above.

D.

Plaintiffs' fourth category of allegations used to support the inference of scienter amounts to a "core operations" theory. Under this theory, plaintiffs contend that because some of Defendants' alleged misstatements related to DXC's core operations, Defendants were more likely to know that their statements were false. In Yates , we acknowledged that core-operations allegations "are relevant to the court's holistic analysis of scienter." Yates , 744 F.3d at 890. But we cautioned that "bare allegations" that officers have "knowledge of key facts" because of their positions or because such knowledge relates to the business's core operations are not enough, standing alone, "to support a strong inference of scienter." Id. (quoting Zucco Partners , 552 F.3d at 1000 ).

Here, Plaintiffs argue we should draw an inference of scienter from the fact that Defendants allegedly made false statements concerning DXC's move into the digital space and investments in its human capital after representing that those issues were key concerns for the company. Even if such broad corporate strategies can constitute core operations, no core-operations inference can be appropriately drawn from these allegations. Without particularized allegations regarding Defendants' knowledge of shortcomings regarding the digital space and investments in employees, Plaintiffs are simply asking us to assume scienter because part of DXC's business plan was not going smoothly. This we cannot do. We therefore conclude there is no inference of scienter based on the core-operations theory.

E.

Finally, Plaintiffs point to the temporal proximity between some of Defendants' allegedly false statements or omissions and the subsequent disclosure of the truth. Plaintiffs argue that we should infer scienter from the fact that Defendants "offered positive revenue projections and reassuring guidance as late as mid-August of 2018" and then announced a substantial decline in revenue less than three months later. Opening Br. at 40–41. Compare J.A. 101–02 (collecting optimistic revenue statements made by both individual Defendants in August 2018), with J.A. 76 (announcing an $800 million reduction in guidance for company revenue on November 6).

While such temporal proximity is relevant to the scienter inquiry, we follow our sister circuits in declining to draw a strong inference of scienter based solely on there being a short amount of time between positive announcements and statements announcing negative financial information. See Elam v. Neidorff , 544 F.3d 921, 930 (8th Cir. 2008) (declining to draw a strong inference of scienter where plaintiffs' allegation that the defendant "experienced financial difficulties within a month of [making] positive statements" amounted to "nothing more than hindsight" (internal quotation marks omitted)); Arazie v. Mullane , 2 F.3d 1456, 1467–68 (7th Cir. 1993) ("[T]emporal proximity between positive statements stressing a firm's strengths and announcements of poor economic performance do not create an inference that the earlier statements were fraudulent.").

Here, Plaintiffs' use of the period between Defendants' allegedly false statements or omissions and the subsequent disclosure of the truth "amounts to little more than pleading fraud by hindsight." In re Triangle , 988 F.3d at 753 (quoting In re Biogen Inc. Sec. Litig. , 857 F.3d 34, 44 (1st Cir. 2017) ). We therefore decline to draw a strong inference of scienter from the proximity of DXC's optimistic statements in the summer and its more sobering news in the fall.

F.

Accordingly, none of Plaintiffs' five categories of allegations alone can support a strong inference of scienter. That leaves the question of whether, "evaluat[ing] the complaint holistically," the combined allegations can do what their individual parts failed to do. Yates , 744 F.3d at 893 (noting that "allegations of scienter that would not independently create a strong inference of scienter might [complement] each other to create an inference of sufficient strength to satisfy the PSLRA" (quoting Matrix Cap. , 576 F.3d at 187–88 )). We hold that they cannot. Holistically analyzing Plaintiffs' allegations as to scienter, the non-fraudulent inference—that DXC unexpectedly stumbled, leading to reduced revenue and other greater-than-expected operational difficulties—is more compelling than the requisite inference that Defendants knowingly or recklessly misled investors about the company's financial health.

While there is some reason to believe Defendants acted with the requisite intent—including the individual Defendants' stock sales and the speed with which DXC's business outlook changed—there is not enough to create a strong inference of scienter. Furthermore, Plaintiffs' own allegations fully explain why DXC did not hit all of its business-growth goals. As Plaintiffs acknowledge in their complaint, DXC told investors it stumbled because of a combination of "a stronger dollar, completion of several large transformation projects, and slower ramp-up on a few large Digital contracts." J.A. 35 (emphasis omitted). Plus, DXC experienced a decline in its application and maintenance business as customers contemplated systems upgrades, made an unsuccessful switch to a "generalist sales model," and had trouble executing its workforce optimization plan. J.A. 78. While Plaintiffs argue that these roadblocks were the inevitable result of Defendants' business cuts and should have been more thoroughly disclosed, the existence of a plausible—and largely uncontested—innocent narrative explaining DXC's struggles further convinces us that Plaintiffs have failed to adequately plead scienter.

Additionally, on multiple prior occasions, we have found that when defendants disclose risks and newly discovered weaknesses to investors this counts against an inference of scienter. See Yates , 744 F.3d at 892–93 ; In re Triangle , 988 F.3d at 755–56. Here, we note that Defendants announced DXC's significant revenue decline to its investors in a timely manner. Further, while Defendants announced a downward revision in revenue projections, they also announced an increase in earnings per share. This undercuts Plaintiffs' arguments that the company financials soured overnight and were entirely unfavorable to investors, and that the Defendants engaged in the fraudulent misdirection of investors by withholding information.

Accordingly, we hold that Plaintiffs have not satisfied the PSLRA's heightened burden for pleading scienter. This failure to plead scienter sufficiently is fatal to both Plaintiffs' securities fraud claim and their director liability claims against the individual Defendants.

IV.

For the foregoing reasons, the judgment of the district court is AFFIRMED.