Before Higginbotham, Smith, and Higginson, Circuit Judges. Patrick E. Higginbotham, Circuit Judge: A jury found Ferrellgas, Inc. liable for a manufacturing defect and negligence after C. Sidney Johnston was injured using a propane tank it manufactured. During trial, Ferrellgas moved for a judgment as a matter of law (“JMOL”) on both the manufacturing defect and negligence claims, as well as, alternatively, for a new trial or a remittitur. The district court denied the motion for JMOL, conditionally denied the motion for a new trial, and granted the motion for a remittitur, reducing the jury award from $7 million to $1.7 million.[1] On appeal, Ferrellgas argues the district court erred by denying its JMOL motion. Because we find insufficient evidence to support the verdict, we REVERSE and RENDER judgment for Ferrellgas. I. A. On July 26, 2019, Johnston was injured when he used a propane gas tank manufactured and distributed by Ferrellgas. The tank was placed in circulation in 1999, requalified in 2017, and was within its “requalification date” at the time of the incident. Ferrellgas replaced the tank’s valve in 2002, and it refilled and inspected the tank in February 2019.[2] At that time, Ferrellgas placed a blue cap on the tank’s valve and shipped it to Lowe’s, where it sat in an outdoor display until Johnston purchased it approximately five months later on July 24, 2019. Two days later, Johnston attempted to connect the tank to his Char-Broil grill and pushed the igniter, but nothing happened. After he hit the ignitor a second time, a flash fire occurred. Johnston suffered first- and second-degree burns, covering 8% of his body, to his hands, stomach, and face, for which he received outpatient treatment. He returned to work part-time within a month, full-time within two months, and has since “excel[led] at his job.” Johnston’s burn physician testified that Johnston has made an “excellent recovery,” has “no significant scarring[,]” and “ is healed at this point.”[3] B. Johnston and his wife Danette sued Char-Broil and Ferrellgas in Texas state court. Relevant here, the Johnstons stated claims against Ferrellgas for strict products liability and negligence, arguing there was a manufacturing defect.[4] Specifically, the Johnstons alleged that the tank “contained one or more conditions which rendered it defective,” including “an older valve and seal.” Mrs. Johnston also sought damages for loss of consortium. Char-Broil removed the suit to federal court and settled. The Johnstons and Ferrellgas proceeded to trial in federal court with a seven-person jury. Ferrellgas moved for a JMOL pursuant to Federal Rule of Civil Procedure 50(b), contending that the Johnstons did not present sufficient evidence that the “face seal contained a manufacturing defect and was the proximate cause of the Plaintiffs’ injuries, or that Ferrellgas breached any duty.”[5] The district court denied the motion.[6] The jury found Ferrellgas liable for a manufacturing defect and negligence, considered Ferrellgas 93% responsible and Johnston 7% responsible, and returned a $7.5 million verdict.[7] Then, Ferrellgas renewed its JMOL motion and, alternatively, moved for a new trial or remittitur. It argued: (1) the Johnstons failed to prove “the face seal was defective when it left Ferrellgas or, stated otherwise, that any defect existed that constituted a producing cause of Plaintiffs’ injuries”; (2) the experts failed to “rule out plausible alternatives”; and (3) the experts’ opinions were based “solely on post-accident testing untethered to the pre-accident condition of the face seal.” Ferrellgas contended that the Johnstons effectively sought to invoke “res ipsa loquitur without pleading or proving that theory” and that the experts’ opinions were “speculative and conclusory.” The Johnstons disagreed, arguing there was evidence of a defect because there was “a leak in the seal,” the defect existed when Ferrellgas had possession of the tank because the seal was covered by a cap when it left Ferrellgas, and the seal was likely to have remained in the same condition. The district court denied Ferrellgas’s renewed JMOL motion. Regarding the manufacturing defect claim, the opinion outlined the arguments on both sides, stated the sufficiency of the evidence standard, and succinctly concluded: [T]here was sufficient expert testimony and circumstantial evidence to permit the jury to infer that a defect existed in the face seal when the tank left Ferrellgas’s possession. The Court rejects Ferrellgas’s characterization of the expert testimony as speculative and conclusory. Instead, the Court concludes that sufficient testimony with an adequate basis supported the jury’s findings.[8] The court also denied the JMOL motion as to the Johnstons’ negligence claim, concluding that it “was for the jury to determine whether Ferrellgas adequately inspected the seal” and that “a reasonable jury could independently find Ferrellgas breached its duty based on its policy of having no maximum lifespan for face seals and no policy for replacing old seals unless a defect was detected.”[9] The district court also conditionally denied Ferrellgas’s motion for a new trial; however, it offered a remittitur to reduce the damages to $1.7 million. The court noted that it had never “seen a jury go off the rails like this,” opined the “award was totally unsupported by the evidence that [it] saw,” and thought it was “duty-bound to intervene.” The Johnstons did not initially respond to the remittitur offer, so a new trial on damages was scheduled. During a pretrial conference, the district judge commented that “[y]ou can make a good argument that the evidence is not there and the testimony was not there to connect the damaged product to the actual explosion.” Following the conference, the Johnstons moved to reopen the period to accept the remittitur, which the district court granted. On December 9, 2022, the district court entered final judgment, awarding the Johnstons $1,846,000. Ferrellgas appeals the district court’s denial of its JMOL motion, arguing there was no legally sufficient evidence of a manufacturing defect that existed when the tank left Ferrellgas’s possession, nor was there sufficient evidence of negligence. II. The district court had jurisdiction over this diversity case under 28 U.S.C. § 1332(a),[10] and there is no challenge to the application of Texas law in this case. This Court has jurisdiction over the district court’s final judgment pursuant to 28 U.S.C. § 1291. A district court should grant a JMOL motion only if “a reasonable jury would not have a legally sufficient evidentiary basis to find for the party on that issue,” which occurs when “the facts and inferences point so strongly and overwhelmingly in the movant’s favor that jurors could not reasonably have reached a contrary verdict.”[11] This Court reviews a district court’s denial of a JMOL motion de novo and applies the same deferential standard as the district court does in reviewing the jury’s verdict.[12] Courts of appeals are “wary of upsetting jury verdicts”[13] and will do so only if there is no “substantial evidence” that supports the verdict.[14] “Substantial evidence is defined as evidence of such quality and weight that reasonable and fair-minded men in the exercise of impartial judgment might reach different conclusions.”[15] This Court “must consider all of the evidence in the light most favorable to the nonmovant, drawing all factual inferences in favor of the non-moving party, and leaving credibility determinations, the weighing of evidence, and the drawing of legitimate inferences from the facts to the jury.”[16] III. Ferrellgas argues there was insufficient evidence of a specific manufacturing defect, extant when the tank left its possession, that was a producing cause of the fire and injuries at issue. Because, after reviewing the record, we find no substantial evidence to support the jury’s finding that any defect existed at the time the tank left Ferrellgas, we REVERSE.[17] A. “A manufacturing defect exists when a product deviates, in its construction or quality, from the specifications or planned output in a manner that renders it unreasonably dangerous.”[18] These elements are “essential,” as otherwise “the jury is invited to find liability based on speculation as to the cause of the incident in issue.”[19] “To prove a manufacturing defect under Texas law, a specific defect must be identified by competent evidence and other possible causes must be ruled out.”[20] The defect must have existed at the time the product left the manufacturer’s possession, and it must have been “a producing cause” of the plaintiff’s injuries.[21] “Causation-in-fact is common to both proximate and producing cause, including the requirement that the defendant’s conduct or product be a substantial factor in bringing about the injuries in question.”[22] “[C]ourts are to rigorously examine the validity of facts and assumptions on which the [expert] testimony is based . . . .”[23] Conclusory or speculative expert opinions cannot support a verdict.[24] B. The Johnstons argued at trial that the tank was defective because it had a “bad” face seal that leaked propane gas. After reviewing the record, we find no substantial evidence that this purported defect existed when the tank left Ferrellgas. Fatally, no expert or witness could identify when this purported defect arose. The Johnstons’ expert, Scott Buske, opined that the face seal “probably dried out,” “degraded,” or “shrunk” at some point, but he could offer no opinion on whether the alleged defect existed when the product left Ferrellgas. He initially agreed there was no reason to believe the seal’s condition had changed from the time Ferrellgas placed the blue cap on the valve to when Johnston removed it shortly before the incident. But this statement proved merely speculative, as he expressly and repeatedly conceded on cross-examination that the seal does not always malfunction and that he could not say it was defective when it left Ferrellgas: Q: So, you’re not able to testify under oath that the face seal was in a condition where it had shrunk to the point where it wouldn’t provide a seal when it left Blue Rhino [Ferrellgas], because you don’t know when that happened; fair statement? A. That’s—that’s a fair statement. Of course, if you tighten it up all the way to the very end, then it was holding a seal. So, it was kind of intermittent. … Q: [Y]ou don’t know what condition it was in when it left Blue Rhino [Ferrellgas], fair? A: Fair. Portions of the deposition of James Petersen, Char-Broil’s expert, were admitted at trial, but he did not opine on this issue. Additionally, Jim McGrath (a Ferrellgas plant manager, testifying as Ferrellgas’s corporate representative) said he had “no idea what has occurred between the time we ship [a tank] and the time that it arrives in someone’s backyard,” “can’t speculate as to what condition it might be in,” and said to his knowledge, a defective tank has never left his facility. Given this testimony and Buske’s concessions, there was no basis by which a reasonable juror could conclude that the defect existed at the time the tank left Ferrellgas. We are unpersuaded by the Johnstons’ attempt to avoid this evidentiary deficiency by relying on Texas’s sealed container doctrine, which states that “[w]hen it is shown that the product involved comes in a sealed container, it is inferable that the product reached the consumer without substantial change in the condition in which it was sold.”[25] They argue this inference is permitted because there was evidence that the seal, after being inspected by Ferrellgas, was covered by a blue cap before it left the facility; Buske agreed during trial that there was no reason to believe the condition of the seal had changed before Johnston removed the cap; and Johnston testified he bought the tank with the blue cap on it two days before the incident. Although all inferences are to be drawn in favor of the Johnstons as the nonmoving party, the fault in their argument lies in the fact that the propane tank does not come in a sealed container—nor is the blue cap on the tank’s valve a seal. Rather, the cap is “placed over” the valve on the tank containing the seal; it is not a seal itself. Further, the cap has indentations that would allow contaminants to enter. Moreover, the Johnstons did not produce evidence showing why the blue cap should be considered a seal. As Ferrellgas notes, “[t]he ‘sealed container’ presumption, therefore, did not arise as a matter of law—particularly given the Johnstons’ case theory that the face seal may have ‘become expired’ in ‘however many months it took from the time it got from Blue Rhino [Ferrellgas] to Mr. Johnston.’” We agree with our brother in dissent that jury verdicts deserve substantial deference and we do not take lightly the decision to reverse them. But this Court cannot affirm if, as here, there is insufficient evidence as a matter of law to support a jury finding that the tank suffered a manufacturing defect. As we explained, the circumstantial evidence on which the Johnstons rely does not cure the want of proof that the tank was defective when it left Ferrellgas’s possession. This is not a res ipsa case. Indeed, the Johnstons did not advance that theory of liability before the district court or before us. In sum, the Plaintiff’s expert admitted that he could not say the tank was defective at the time it left Ferrellgas, making his prior comments about the tank’s condition at that time pure speculation; the tank functioned properly before Johnston used it; the tank and seal are not sealed containers; and both parties agree Ferrellgas successfully refilled the tank with gas under high-pressure months before the accident.[26] There is no reasonable basis on which the jury could find the Johnstons met their burden. IV. Second, Ferrellgas argues there was insufficient evidence that it was negligent. We agree. Under Texas law, a party alleging a manufacturer was negligent for producing a defective product must first prove that there was a manufacturing defect: [A] manufacturer logically cannot be held liable for failing to exercise ordinary care when producing a product that is not defective because: (1) if a product is not unreasonably dangerous because of the way it was manufactured, it was not negligent to manufacture it that way and (2) even if the manufacturer was somehow negligent in the design or production of the product, that negligence cannot have caused the plaintiff’s injury because the negligence did not render the product “unreasonably dangerous.”[27] The district court recognized this, instructing the jury to reach the negligence question only if it first found Ferrellgas liable for a manufacturing defect. This charge was correct, as the Johnstons’ negligence theory was dependent on the tank having a manufacturing defect. Thus, because we find there was insufficient evidence to support the jury’s finding that Ferrellgas was liable for a manufacturing defect, the negligence claim must also fail. V. There was legally insufficient evidence that any manufacturing defect existed at the time the tank left Ferrellgas. As a result, we REVERSE the denial of Ferrellgas’s JMOL motion as to both the manufacturing defect and negligence claims, and we RENDER judgment for Ferrellgas. Stephen A. Higginson, Circuit Judge, dissenting: Two days after he purchased a propane gas tank that was manufactured, distributed, and—in this case, also refilled—by Ferrellgas, Appellant C. Sidney Johnston attempted to use it in his Char-Broil grill. Johnston, who grilled regularly—about two or three times a week, every week, for twenty years—did exactly as he always did: he connected the propane tank and then hit the igniter. When nothing happened, he hit the igniter a second time, only to be engulfed in flames, ultimately resulting in first- and second- degree burns to eight percent of his body. Johnston and his wife brought suit against both Char-Broil and Ferrellgas. While Char-Broil settled, the case against Ferrellgas proceeded to trial, where the jury heard testimony from, among others, a Ferrellgas representative, James Magrath; the Johnstons; and two expert witnesses— the Johnstons’ expert, Scott Buske, and Char-Broil’s expert, James Petersen, via video.1 After the three-day trial, the unanimous jury found Ferrellgas liable for both manufacturing defect and negligence, awarding a $7.5 million verdict but attributing 7% of the fault to Johnston (which would reduce the recovery accordingly). After an eventual remittitur, the district court reduced the judgment to $1,846,000. Before us is the appeal of the district court’s denial of Ferrellgas’ motion for judgment as a matter of law (JMOL) on both the manufacturing defect and negligence claims. As the panel majority acknowledges, JMOL should be granted “only . . . if the facts and inferences point so strongly and overwhelmingly in the movant’s favor that jurors could not reasonably have reached a contrary verdict.” Janvey v. Dillon Gage, Inc. of Dallas, 856 F.3d 377, 385 (5th Cir. 2017) (citation omitted). This evaluation must “consider all of the evidence in the light most favorable to the nonmovant, drawing all factual inferences in favor of the non-moving party, and leaving credibility determinations, the weighing of evidence, and the drawing of legitimate inferences from the facts to the jury.” Id. (citation omitted). Crucially, though a district court’s denial of a JMOL is reviewed de novo, such review is “especially deferential” after a jury trial, id. (citation omitted), as this court is “wary of upsetting jury verdicts.” Goodner v. Hyundai Motor Co., Ltd., 650 F.3d 1034, 1039 (5th Cir. 2011) (citation omitted). Yet in reversing this jury verdict, which was both unanimous and upheld after scrutiny by the trial judge, the majority fails to adhere to the required deference. The majority opinion reverses due to its determination that there is no substantial evidence to support a finding that any defect existed at the time the tank left Ferrellgas. Maj. Op. at 9–10. “Fatally,” the majority finds, “no expert or witness could identify when th[e] purported defect arose.” Maj. Op. at 10. Texas law does not, however, require plaintiffs to pinpoint the moment a purported defect arises. The Supreme Court of Texas has explained: To exclude circumstantial evidence that the product was de-fective at the time of the sale would frustrate the beneficial purposes of the doctrine. It would be equally difficult, if not impossible, for the plaintiff to rebut by direct evidence all of the conceivable possibilities which would account for the defective condition other than the existence of the defect at the time of the sale. Such direct evidence should not be required, particularly when dealing with a latent defect. Darryl v. Ford Motor Co., 440 S.W.2d 630, 632 (Tex. 1969). The ability to rely on circumstantial evidence “to support an inference that the defect existed when it left the manufacturer” is similarly echoed and expounded upon in caselaw from the state. Shaun T. Mian Corp. v. Hewlett-Packard Co., 237 S.W.3d 851, 863 (Tex. App. 2007); see also Ford Motor Co. v. Ridgway, 135 S.W.3d 598, 602 (Tex. 2004) (Hecht, J., concurring) (explaining that “Texas law would allow proof of products liability by circumstantial evidence in certain cases”). As one Texas Court of Appeals has explained: Sometimes this basis can be provided by evidence of the age of the product and its history of usage up to the time of failure. “The age and use of [a] product during the time intervening between [its] purchase and malfunction will tend to support or defeat the circumstantial weight of the malfunction as proof of original defect.” [General Motors Corp. v. Hopkins, 548 S.W.2d 344, 350 (Tex. 1977).] New or nearly new products “typically have not been modified or repaired, therefore making a product defect the likely cause of an accident.” Ridgway, 135 S.W.3d at 601. Thus, an inference of original product defect may be war-ranted from the malfunction of a relatively new or sealed prod-uct. See Darryl, 440 S.W.2d at 631 (three-month-old truck with 600 to 700 miles and no repairs); [Sipes v. General Motors Corp., 946 S.W.2d 143, 146, 155 (Tex. App. 1997, writ denied)] (sealed airbag system in new car). Shaun T. Mian Corp., 237 S.W.3d at 863. The majority’s focus is therefore misplaced; plaintiffs are only required to “provide a reasonable basis for con-cluding the defective condition did not arise subsequent to the manufacturer’s exercise of control over the product.” Id. Here, the inference that the defect existed when the tank left Ferrellgas is reasonable due to, among other factors, “the age and use of [the] product.” Id. (citation omitted). The gas tank in question was effectively new: it was purchased only two days prior and had never been used between the point of purchase and the date of the incident.2 Johnston testified that a Ferrellgas blue cap (bearing the brand Blue Rhino) was affixed on it the entire time—from the moment the Lowe’s employee retrieved the tank for John-ston, to immediately before Johnston attached it to his grill. This is thus a far cry from a “tire with 30,000 miles and nail puncture,” a “two-year-old truck with 54,000 miles and three repairs to fuel system,” an “allegedly defective valve in 30-year-old aircraft engine,” or other circumstances under which Texas courts find inference of defect unwarranted. Shaun T. Mian Corp., 237 S.W.3d at 863 (collecting cases and explaining that “such an inference may not be warranted if the product was worn, misused, damaged, repaired, or altered after it left the manufacturer’s control”). The majority also fails to consider all the evidence presented to the jury in the light most favorable to the Johnstons and does not credit other reasonable inferences in their favor. For instance, the majority opinion focuses on the fact that Johnston’s expert Buske agreed that it was “fair” to say that he did not know what condition the face seal was in when it left Ferrellgas. Map. Op. at 10. Yet there is also testimony from the only two experts who testified at trial that the face seal on the gas tank was “bad,” and testimony from Buske that “because there [had] been a blue cap over the face seal,” there was no “reason to believe that the condition of that face seal [underneath the cap] ha[d] changed” from the time that the tank left Ferrellgas.3 This, in conjunction with evidence introduced at trial that Ferrellgas had previously seen loose seals and fires resulting from those bad seals on tanks in testing, and that the National Fire Protection Association (NFPA)4 had amended its policies after finding an increasing number of reported deterioration of face seals, furnish a sufficient basis fora reasonable juror to conclude in the Johnstons’ favor. Perhaps the majority deviates because it rejects the Johnstons’ sealed container theory. Based on its review of photographs submitted into the rec-ord, the majority adopts Ferrellgas’ argument that the blue cap on the tank’s valve is not a seal because “the cap has indentations that would allow contaminants to enter,” Maj. Op. at 12, including during the time in which the tank was stored outside at Lowe’s. In conducting its independent assessment, the majority not only assumes the role of the jury—which was presented with the same facts and theory of defense, and necessarily rejected it5—but also embraces the same speculation for which it faults the Johnstons. The Johnstons’ claims apparently must fail as a matter of law because of the possibility that contaminants entered the blue cap and touched the face seal. But what contaminants? The jury heard no testimony as to presence of con-taminants in the geographic area, or as to weather conditions that might have made it likely for the face seal to deteriorate. Nor did Ferrellgas present any evidence about the average length of time a tank typically resides at a retailer, or why it was not foreseeable for the tank in question to have spent time at Lowe’s. Without evidence in the record suggesting that five months is an ab-errantly long inventory period for a refilled tank—which, as a matter of com-mon sense, is necessarily kept outside—the majority today appears to insulate gas refillers like Ferrellgas for defects in any products not immediately purchased. With no evidence, this alternative theory of causation (contamination) is not enough to find against the Johnstons as a matter of law, let alone con-trary to a unanimous jury and the trial judge. Texas law makes clear that “for circumstantial evidence to support inferences that a product was defective and the defect existed at the time it left the manufacturer, the evidence need not disprove all other possible causes for the injury”; indeed, “[t]he plaintiff is not required to exclude an appreciable chance that the event might have occurred in some other way.” Shaun T. Mian Corp., 237 S.W.3d at 863 (emphasis added). “Expressed otherwise, a conclusion of causal connection may be inferred by a balance of probabilities.” Id. Speculation as to unnamed con-taminants constitutes no more than a “reduced” likelihood that does not prevent “the fact-finder [to] reasonably find by a preponderance of the evidence that the cause of the product failure lies at the manufacturer’s door.” Id. “A manufacturing defect exists when a product deviates, in its con-struction or quality, from the specifications or planned output in a manner that renders it unreasonably dangerous.” Ridgway, 135 S.W.3d at 600. As the district court who assessed the trial evidence firsthand held, the Johnstons presented sufficient evidence for a jury to reasonably conclude that the face seal deviated in quality from the planned output in a manner that was unreasonably dangerous, and the verdict should stand. Because the majority opinion fails to consider the evidence in the light most favorable to the Johnstons and misapplies Texas law—construing it to make recovery nearly impossible for injured consumers—I respectfully dissent.
