I. INTRODUCTION
A mudslide destroyed a commercial building (the Building) in Boulder, Colorado, owned by Paros Properties LLC (the Owner) and insured under a policy (the Policy) issued by Colorado Casualty Insurance Company (the Insurer). The Owner filed an insurance claim but the Insurer denied payment because damage from mudslides is excluded from policy coverage. The Owner then filed a state-court suit seeking payment under the Policy and damages for bad-faith breach of the insurance contract. It argued that the mudslide caused the building to explode, bringing the incident within the scope of an explosion exception to the Policy’s mudslide exclusion. The Insurer removed the action to federal court, which granted summary judgment to the Insurer. On appeal the Owner argues (1) that the district court lacked subject-matter jurisdiction because the Insurer’s removal from state court was untimely and (2) that the district court erred on the merits in holding that there *1267was no coverage. We have jurisdiction under 28 U.S.C. § 1291. We hold that the notice of removal was too late. But because the district court correctly ruled on the merits and the jurisdictional requirements were satisfied at that time, we affirm the judgment below rather than burden the state court and the parties by requiring relitigation.
II. BACKGROUND
Torrential rainfall hit Boulder in September 2013. On the night of September 12 a violent flow of water, mud, and debris thundered down a hillside into the Building, causing extensive damage. The Owner submitted an insurance claim but the Insurer denied it after a brief inspection. It explained in a letter: “The inspection revealed that there was a Mudslide that caused the damage to your buildings. Damages caused by Mudslides or Mudflows are specifically excluded under your policy.” Aplt. App., Vol. 2 at 268.
The Owner retained counsel to challenge the denial. Counsel hired an engineer, who inspected the Building on October 15, 2013, and (after a site inspection a year later) issued a report on his findings on November 3, 2014. According to the report, “The debris laden flow impacted the south elevation of the structure, causing a sudden reaction of the wall structure.” Report of Edward L. Fronapfel, Supp. Aplee. App., Vol. 1 at 65. The impact caused the property to “split into two separate structures along a north-south wall line. The eastern portion laterally displaced to the northeast, while the western portion laterally displaced to the northwest. The roof structure collapsed where the building separation occurred due to the sudden loss of the bearing walls.” Id. at 73. The Owner demolished the building on October 23, 2013.
On October 24, 2013, counsel for the Owner sent a letter to the Insurer claiming wrongful denial of coverage. It stated that the “force of the mudslide caused [the Owner’s] building, literally, to explode,” Aplt. App., Vol. 4 at 631, and that the resultant damage was therefore compensa-ble under the explosion exception to the mudslide exclusion in the Policy. It also accused the Insurer of failing to conduct a full and reasonable investigation of the Owner’s claim before denying coverage and, somewhat oddly given the budding’s demolition, requested that the Insurer “complete its investigation into the facts and circumstances surrounding the loss to the ... Building.” Id. at 634. On November 14, the Insurer, which had not been advised of the demolition, informed the Owner that it would continue to investigate the claim and had scheduled an engineer to inspect the site the next morning. When the engineer arrived, he saw that the Building had been demolished.
On January 29, 2014, the Owner’s attorney sent an email to the Insurer with “further explanation of the amounts of the losses suffered by our client.” Aplt. App., Vol. 4 at 625. The email indicated that the Building loss exceeded $1.1 million. (The policy limit was $907,600.) The record contains no indication that the Insurer responded to this email.
On February 26, 2014, the Owner filed its suit in Colorado state court. The complaint alleged “catastrophic damage to the Property and its contents” and claimed damages relating to cleanup, reconstruction, and loss of income and use. Complaint, Aplt. App., Vol. 1 at 21 ¶ 19. It did not quantify the damages other than to say that “[t]he amount in dispute is more than $15,000.” Id. at 20 ¶ 6. It also referenced, but did not attach, correspondence sent “[o]n or about January 30, 2014.” Id. at 23 ¶ 31. On appeal the Owner claims that this *1268was a reference to its January 29 email tallying losses exceeding $1.1 million.
On April 23, 2014, in its initial damages disclosures under Colorado Rule of Civil Procedure 26(a)(1)(C), the Owner claimed to be “in the process of calculating specific damage amounts” and listed the amount of all damages as “TBD.” Aplt. App., Vol. 1 at 30. Two days later, the Owner served its first supplemental disclosures, enumerating damages exceeding $1.3 million. Id. at 32-33. Three days after that, the Insurer removed the case to the United States District Court for the District of Colorado. The Owner moved that court to remand, but the court denied the motion and later granted summary judgment to the Insurer.
III. TIMELINESS OF REMOVAL A. General Principles
The Insurer was served with the complaint on February 26, 2014, and filed its notice of removal on April 28, 2014. Under the federal removal statute the notice ordinarily must be filed “within 30 days after the receipt by the defendant, through service or otherwise, of a copy of the initial pleading setting forth the claim for relief upon which such action or proceeding is based.” 28 U.S.C. § 1446(b)(1).1 This deadline was clearly not met. The Insurer relies, however, on an escape hatch. Under another provision of the removal statute, “if the case stated by the initial pleading is not removable, a notice of removal may be filed within 30 days after receipt by the defendant, through service or otherwise, of a copy of an amended pleading, motion, order or other paper from which it may first be ascertained that the case is one which is or has become removable.” Id. § 1446(b)(3).2 Most courts, including this one, have consistently interpreted the term other paper broadly to include state-court filings and discovery. See, e.g., Huffman v. Saul Holdings Ltd. Partnership, 194 F.3d 1072, 1079 (10th Cir. 1999) (30-day clock triggered by deposition testimony); DeBry v. Transamerica Corp., 601 F.2d 480, 488-89 (10th Cir. 1979) (clock not triggered by ambiguous deposition testimony); 14G Charles Alan Wright and Arthur R. Miller, Federal Practice and Procedure § 3731, at 524 (“The federal courts have given the reference to ‘other paper’ an expansive construction and have included a wide array of documents within its scope.”) And this position has now been codified in 28 U.S.C. § 1446(c)(3)(A).3
The Insurer points out that it could not remove the case to federal court under *1269diversity jurisdiction (it is not disputed that the Owner and the Insurer are citizens of different states) unless the amount in controversy was at least $75,000, see 28 U.S.C. § 1332(a) (requirements for diversity jurisdiction); id. § 1441(a) (requirements for removal), and it contends that the initial complaint left in doubt whether the Owner’s claim equaled or exceeded $75,000. It says that it was not put on notice until April 25, when the Owner disclosed its damages computation of over $1.3 million.
The question before us, then, is when could the Insurer “first ... ascertain[ ] that the case [was] one which [was] or [had] become removable.” Id. § 1446(b)(3). In particular, could the Insurer have ascertained removability from the complaint itself and, if not, could it have done so from some “other paper” it received before the Owner disclosed its damages computation?
This circuit has been very strict in assessing whether the grounds for removal are ascertainable. We require a specific allegation that damages exceed the federal jurisdictional amount of $75,000.4 The 30-day clock does not begin to run until the plaintiff provides the defendant with “clear and unequivocal notice” that the suit is removable. Akin v. Ashland Chemical Co., 156 F.3d 1030, 1036 (10th Cir. 1998). In Akin federal jurisdiction hinged upon whether exposure to a toxic substance occurred within a federal enclave. See id. at 1034. The complaint alleged “injuries sustained ‘while working at’ Tinker Air Force Base.” Id. at 1035. We held that this allegation “did not provide unequivocal notice of the right to remove” because the -phrase while working at was ambiguous; it could have been a “geographical modifier” (meaning the exposure occurred on the base) or a “durational modifier” (exposure occurred during the time the plaintiffs were employed by the base). See id.
Huffman, 194 F.3d 1072, is similar. The plaintiffs leased space to operate a retail furniture store, but the defendant landlord failed to fix a leaking roof for a number of months. See id. at 1075. The lessees filed a complaint containing numerous allegations from which the defendant might well have surmised that damages exceeded the jurisdictional floor: that the “constant problem with the roof leaking made it almost impossible to carry on a business,” that it “was impossible to avoid or prevent the moldy, damp, smelly atmosphere of the store,” and that “[c]ustomers complained and often left quickly without taking time to browse or really shop the furniture items because of these intolerable conditions.” Id. (internal quotation marks omitted). Further, the complaint sought actual and punitive damages. Id. at 1077. But as to dollar amount the complaint said only that the plaintiffs sought “in excess of $10,000.” Id. at 1077 (internal quotation *1270marks omitted). We ruled that based on the complaint, the defendant “could only guess as to whether the claim exceeded $75,000” and that the complaint therefore failed to put the defendant on notice of removability. Id. The 30-day clock was not triggered until the store owner’s deposition testimony that the plaintiffs sought more than $300,000. See id. at 1079.
DeBry, 601 F.2d 480, illustrates that our strict standard also applies to notice provided after the complaint. We considered whether the plaintiffs deposition testimony gave notice that he was a citizen of Utah (for purposes of establishing diversity of citizenship). The complaint alleged that he had been a citizen of Utah until June 1972. See id. at 481-82. He testified that after “mov[ing] to California [from Salt Lake City] the first part of July of 1972,” he had “returned [to Salt Lake City] in August of [1974] and purchased a unit.” Id. at 495. We held that this testimony did not give notice. Because the plaintiff “did not say that he had permanently moved,” the defendant did not “learn with certainty” from the deposition that the plaintiff was a citizen of Utah. Id. at 488-89.
Our strict rule is justified by important practical considerations. Litigation concerning when the defendant realized or should have realized the amount in controversy or the like can expend considerable resources of the court and the parties. And the expenditure is wholly unnecessary. Any time the plaintiff wishes to start the 30-day clock, it can provide the defendant with an unambiguous notice of what is being claimed. For similar reasons, other circuits have held that they will not inquire into a defendant’s subjective knowledge or what the defendant should have inferred from an investigation or a review of its files. See, e.g., Moltner v. Starbucks Coffee Co., 624 F.3d 34, 38 (2d Cir. 2010) (“[T]he removal clock does not start to run until the plaintiff serves the defendant with a paper that explicitly specifies the amount of monetary damages sought.... Requiring a defendant to read the complaint and guess the amount of damages that the plaintiff seeks will create uncertainty and risks increasing the time and money spent on litigation.”); Foster v. Mut. Fire, Marine & Inland Ins. Co., 986 F.2d 48, 54 (3d Cir. 1993), overruled on other grounds by Murphy Bros., Inc. v. Michetti Pipe Stringing, Inc., 526 U.S. 344, 119 S.Ct. 1322, 143 L.Ed.2d 448 (1999) (“[T]he relevant test is not what the defendants purportedly knew, but what [the plaintiffs] documents said.”); Lovern v. Gen. Motors Corp., 121 F.3d 160, 162 (4th Cir. 1997) (“[W]e will not require courts to inquire into the subjective knowledge of the defendant, an inquiry that could degenerate into a mini-trial regarding who knew what and when.... [The grounds for removal must] be apparent within the four corners of the initial pleading or subsequent paper.”); Mumfrey v. CVS Pharmacy, Inc., 719 F.3d 392, 399 (5th Cir. 2013) (seeking to “promote efficiency by preventing courts from expending copious time determining what a defendant should have known or have been able to ascertain at the time of the initial pleading” and to “avoid encouraging defendants to remove cases prematurely for fear of accidentally letting the thirty-day window to federal court close when it is unclear that the initial pleading satisfies the amount in controversy”); Graiser v. Visionworks of Am., Inc., 819 F.3d 277, 285 (6th Cir. 2016) (“[A] defendant is not required to search its own business records or perform an independent investigation into a plaintiffs indeterminate allegations to determine removability.” (internal quotation marks omitted)); Walker v. Trailer Transit, Inc., 727 F.3d 819, 825 (7th Cir. 2013) (requiring “specific and unambiguous notice that the case satisfies *1271federal jurisdictional requirements and therefore is removable. Assessing the timeliness of removal should not involve a fact-intensive inquiry about what the defendant subjectively knew or should have discovered through independent investigation.”); In re Willis, 228 F.3d 896, 897 (8th Cir. 2000) (per curiam) (“[T]he thirty-day time limit of section 1446(b) begins running upon receipt of the initial complaint only when the complaint explicitly discloses the plaintiff is seeking damages in excess of the federal jurisdictional amount.”); Carvalho v. Equifax Info. Servs., LLC, 629 F.3d 876, 886 (9th Cir. 2010) (“We adopted this bright-line approach to avoid the spectre of inevitable collateral litigation over whether defendant had subjective knowledge, or whether defendant conducted sufficient inquiry.” (alterations and internal quotation marks omitted)); cf. Romulus v. CVS Pharm., Inc., 770 F.3d 67, 80 (1st Cir. 2014) (“The timeliness inquiry is limited to the information in the plaintiffs’ papers.... The defendant has no duty to perform significant investigation of its own data to ascertain removability.”).5
We now turn to the specifics of this case.
B. Application to this case
The Owner relies on two sources to establish that the Insurer should have ascertained the amount in controversy more than 30 days before it filed its notice of removal: (1) an email sent before suit was filed and (2) the civil cover sheet filed with the complaint. We first address the email.
1. Presuit email
On January 29, 2014, the Owner’s counsel sent an email to the Insurer setting forth an assessment of the losses suffered from destruction of the Building. The assessment, which totaled almost $1.2 million, arose from damage to the building, lost rental income, and debris removal. We first consider whether the email can be considered an “other paper” and then whether it would suffice to provide notice as part of the complaint.
A written demand from counsel may be an “other paper” under § 1446(b)(3) “from which it may first be ascertained that the case is one which is or has become removable.” Several circuit courts have 'so held. See Romulus, 770 F.3d at 78; Babasa v. LensCrafters, Inc., 498 F.3d 972, 975 (9th Cir. 2007); Addo v. Globe Life & Ace. Ins. Co., 230 F.3d 759, 762 (5th Cir. 2000). See generally Wright & Miller, Federal Practice and Procedure § 3731, at 524, 545 (“The federal courts have given the reference to ‘other paper’ an expansive construction and have included a wide array of documents within its scope ... [including] correspondence between the parties and their attorneys or between the attorneys.”). But those cases have involved only postsuit demands. This limitation makes sense because the statutory language clearly does not contemplate “papers” submitted before the complaint *1272was filed. After all, the removal clock under § 1446(b)(3) starts ticking upon receipt of the “paper,” so the time for removal after receipt of a presuit paper could well expire before service of the complaint and would never extend beyond the time permitted by § 1446(b)(1)' — which is 30 days after service of the complaint or summons. We agree with the other circuits to have considered the matter that a presuit communication is not an “other paper.” See Chapman v. Powermatic, Inc., 969 F.2d 160, 164 (5th Cir. 1992) (“By its plain terms the statute requires that if an ‘other paper’ is to trigger the thirty-day time period of the second paragraph of § 1446(b), the defendant must receive the ‘other paper’ only after it receives the initial pleading.”); Carvalho, 629 F.3d at 885-86 (same).
Perhaps, however, presuit demand could provide adequate notice if it were incorporated into the complaint. The Owner points out that its complaint contains the allegation that “[o]n or about January 30, 2014, [Owner] followed up with [the Insurer] again to provide updated information regarding [the Owner’s] damages.” Complaint, Aplt. App., Vol. 1 at 23 ¶ 31. But this reference fails for two reasons. First, there is ambiguity about the date. The allegation in the complaint refers to an email sent “on or about January 30,” which is not a “clear and unequivocal” reference to an email actually sent on January 29. Second, the email does not communicate a demand. It sets forth alleged damages without stating how much the Insurer must pay. Indeed, the total damages asserted in the email exceed the policy limits, so the Owner was presumably not seeking payment of all the listed damages. Also, if the Owner’s claim was limited to the loss caused by an explosion resulting from the mudslide, the claim would not include property damage caused by the mudslide alone.
2. Civil Cover Sheet
The Owner also alleges that the Insurer was on notice of the amount in controversy because of the civil cover sheet filed in state court with the complaint. For cases seeking $100,000 or less, Colorado uses a simplified procedure with detailed early disclosures and limited discovery. See C.R.C.P. 16.1. So that the court can determine whether the procedure applies, civil complaints (except in certain types of excluded cases not present here) must be accompanied by a civil cover sheet indicating whether the amount in controversy exceeds $100,000. See C.R.C.P. 16.1(b)(3). In this case the Owner checked the box indicating that the suit sought more than $100,000.
If the cover sheet is not considered part of the “initial pleading” under § 1446(b)(1), it is at least properly considered an “other paper” under § 1446(b)(3). The question is whether its content provides adequate notice.
We previously declined to decide in an unpublished opinion whether a defendant could support its removal notice by reference to a plaintiffs check mark on the Colorado civil cover sheet. See Warner v. CitiMortgage, Inc., 533 Fed.Appx. 813, 816 (10th Cir. 2013). But at least one federal district judge in Colorado has determined that the cover sheet is notice that starts the removal clock. See Henderson v. Target Stores, Inc., 431 F.Supp.2d 1143, 1144 (D. Colo. 2006) (civil cover sheet is an “other paper” that put the defendant on notice that the amount in controversy exceeded $75,000). We think that view is sound. There is no ambiguity in the cover sheet. And we see no reason not to credit an assertion by an officer of the court on a matter of significant consequence in the state proceeding (whether or not simplified *1273procedures will apply). Because the Insurer’s notice of removal was filed more than 30 days after it received the cover sheet, the notice was untimely.
That does not, however, end the matter. Although the district court should have remanded the case to state court, a lot of water has now flowed under the bridge. Must we now order that the litigation start anew in state court? The Supreme Court tells us no. In Caterpillar Inc. v. Lewis, 519 U.S. 61, 64, 117 S.Ct. 467, 136 L.Ed.2d 437 (1996), a unanimous Supreme Court declared that “a district court’s error in failing to remand a case improperly removed is not fatal to the ensuing adjudication if federal jurisdictional requirements are met at the time judgment is entered.” In that case the federal district court (erroneously) denied a motion to remand a removed case when the parties were not completely diverse; the nondiverse parties later settled and the case proceeded to trial and judgment. See id. at 64, 117 S.Ct. 467. The Supreme Court ruled that the defective removal did not require that the judgment be vacated. Untimeliness is a “statutory flaw,” not a “jurisdictional defect.” Id. at 73, 117 S.Ct. 467 (emphasis omitted). And “[ojnce a diversity case has been tried in federal court,” the Court wrote, “considerations of finality, efficiency, and economy become overwhelming.” Id. at 75, 117 S.Ct. 467. “To wipe out the adjudication postjudgment, and return to state court a case now satisfying all federal jurisdictional requirements, would impose an exorbitant cost on our dual court system, a cost incompatible with .the fair and unprotracted administration of justice.” Id. at 77, 117 S.Ct. 467.
Here, it was clear by the time the Insurer filed its notice of removal that the amount in controversy exceeded $75,000. Although there was no trial in this case, this circuit has confirmed that the rule of Caterpillar applies not only after a trial but also when “judgment is based on ... a district court’s ruling on a dispositive motion.” Huffman, 194 F.3d at 1080. Remand to state court would therefore be improper if we upheld the summary judgment. On the other hand, the efficiency rationale of Caterpillar would disappear if we reversed the district court’s summary judgment, because the case would need to proceed anew anyway, and it might as well proceed in state court as in federal. See id. But, as we proceed to explain, we affirm the judgment below. •
IV. POLICY COVERAGE
Turning to the merits, we agree that the Owner failed to raise a genuine issue of material fact that the damage to the Building is covered by the Policy. The critical policy provision is the “Water Exclusion Endorsement,” which excludes from coverage any damage caused by the following water-based sources:
1. Flood, surface water, waves (including tidal wave and tsunami), tides, tidal water, overflow of any body of water, or spray from any of these, all whether or not driven by wind (including storm surge);
2. Mudslide or mudflow;
3. Water that backs up or overflows or is otherwise discharged from a sewer, drain, sump, sump pump or related equipment;
4. Water under the ground surface pressing on, or flowing or seeping through:
a. Foundations, walls, floors or paved surfaces;
b. Basements, whether paved or not; or
c. Doors, windows or other openings; or
*12745. Waterborne material carried or otherwise moved by any of the water referred to in Paragraph 1., 3., or 4., or material carried or otherwise moved by mudslide or mudflow.
This exclusion applies regardless of whether any of the above, in Paragraphs 1. through 5., is caused by an act of nature or is otherwise caused.
Supp. Aplee. Appx., Vol. 8 at 555. There is, however, an exception to this exclusion for damage from explosions:
But if any of the above, in Paragraphs 1. through 5., results in fire, explosion or sprinkler leakage, we will pay for the loss or damage caused by that fire, explosion or sprinkler leakage (if sprinkler leakage is a Covered Cause of Loss).
Id. In short, the. Policy does not cover damage caused by water, but does cover damage caused by an explosion caused by water.
Although the Insurer has the burden of establishing that the Owner’s loss fell within the exclusion, see Colorado Intergovernmental Risk Sharing Agency v. Northfield Ins. Co., 207 P.3d 839, 842 (Colo. App. 2008), the Owner does not dispute that the damage was caused by a mudslide. But it makes two arguments why it is still covered: (1) the damage to the Building was not caused by surface water and is therefore not within the water-exclusion provision; (2) the Building was destroyed not by the mudslide itself, but by a resulting explosion. We are not persuaded. .
A. Surface Water
The Owner’s surface-water argument is easily resolved. The Owner relies on Heller v. Fire Insurance Exchange, a Division of Farmers Insurance Group, 800 P.2d 1006, 1007 (Colo. 1990), in which runoffs of melted snow damaged a mountain residence in Vail, Colorado. The insurance policy excluded loss caused by “flood, surface water, waves, tidal water, overflow of a body of water, or spray from any of these, whether driven by wind or not.” Id. at 1008 n.2. The insurer denied coverage on the ground that surface water had caused the damage. See id. at 1008. The Colorado Supreme Court ruled for the plaintiffs. It defined surface water in part as water that “follows no defined course or channel.” Id. at 1009. But in that case the “path of the water had been diverted ... by three parallel trenches.... fifteen to twenty feet long, three feet wide, six inches deep, and lined with plastic sheets, rocks and tree limbs.” Id. at 1007. The court reasoned that “the runoff lost its character as surface water when it was diverted' by the trenches and therefore was not within the surface water exclusion.” Id. at 1009.
The Owner argues that, as in Heller, the water here lost its character as surface water because it “was diverted by man-made parking lots, roadways, drainpipes, and culverts on Flagstaff Mountain uphill from the Building.” Aplt. Br. at 57. This diversion seems less dramatic than the one in Heller, but the factual distinctions are of no consequence because the terms of the policies are different. In Heller the relevant exclusion was for surface water. See Heller, 800 P.2d at 1008. The exclusion here encompasses “surface water,” but it also includes “mudslide or mudflow.” Supp. Aplee. App., Vol. 8 at 555. The Owner concedes the damage to have been caused by “an avalanche of mud, rocks, and debris ... [that] careened into the Building.” Aplt. Br. at 2. The Owner does not, and could not, make any argument that the avalanche was not a mudslide. And under the Policy it makes no difference whether any water causing the avalanche had been diverted by manmade features; mudslide damage is excluded whether the mudslide “is caused by an act *1275of nature or is otherwise caused.” Supp. Aplee. App., Vol. 8 at 555.
B. Explosion
As for the Owner’s second argument, the facts do matter. The Owner contends that the mudslide “caus[ed] the Building to violently and suddenly burst apart with a loud boom,” that is to say, caused the building to explode, thus bringing the damage within the explosion exception to the water exclusion. Aplt. Br. at 2. We disagree that demolition by an external cascade of water, mud, and debris is an explosion under the Policy.
“In interpreting a contract, we seek to give effect to the intent and reasonable expectations of the parties.” Thompson v. Maryland Cas. Co., 84 P.3d 496, 501 (Colo. 2004) (en banc). “Accordingly, unless the parties intend otherwise, terms in an insurance policy should be assigned their plain and ordinary meaning.” Id. To comport with public policy and principles of fairness, we construe ambiguous terms in an insurance policy against the insurer. See id. at 501-02. We also “read the provisions of the policy as a whole, rather than reading them in isolation.” Cyprus Amax Minerals Co. v. Lexington Ins. Co., 74 P.3d 294, 299 (Colo. 2003) (en banc); see Sec. Ins. Co. of Hartford v. Houser, 191 Colo. 189, 552 P.2d 308, 310 (1976) (en banc) (interpreting an insurance-policy clause based on the purpose of the clause within the policy).
The Policy itself does not define the term explosion. But that does not mean that the term must be construed to encompass all possible meanings, even if found in the dictionary. Context matters. We would be reluctant, for example, to construe policy language to include figurative meanings. See Rhinelander v. Ins. Co. of Pennsylvania, 8 U.S. (4 Cranch) 29, 44, 2 L.Ed. 540 (1807) (“Commercial contracts have but little connection with figurative language.”). Although a football player may “explode” off the line of scrimmage, we would not construe the exception to the exclusion to include damage to a wall from someone (even someone who is 6'6" tall and weighs 330 pounds) fleeing a flash flood. Nor does it make sense to construe the term explosion in a way that would undermine the exclusion to which it is an exception. The Owner urges us to find an explosion any time an external impact transfers sufficient kinetic energy to a structure to destroy it. But the exclusion includes tidal waves, tsunamis, and mudslides, which all typically produce extreme forces that can smash anything in their paths; to adopt the Owner’s conception would be to read those exclusions out of the Policy.
What makes most sense in the present context is the classical notion of an explosion, as from a bomb or leaking gas. Such an explosion involves a buildup of internal pressure and a sudden bursting outward in all available directions. The exception would apply, for example, if a mudslide damaged a gas pipe, creating a leak of gas that was ignited and exploded. Our understanding of the term is consistent with that of other courts construing “explosion” in an insurance policy. See Pre-Cast Concrete Prods., Inc. v. Home Ins. Co., 417 F.2d 1323, 1328 (7th Cir. 1969) (“[A]n explosion occurs when the pressure inside the container exceeds the strength of the container and results in a sudden release of the pressure.”); Jersey Ins. Co. of N.Y. v. Heffron, 242 F.2d 136, 139 (4th Cir. 1957) (finding an explosion where “the roof, falling intact like a huge piston ... gradually built a compression of air ... great enough to burst the first story windows”); Commercial Union Fire Ins. Co. v. Bank of Ga., 197 F.2d 455, 457 (5th Cir. 1952) (In a bursting-fire-hydrant case, “the bursting [was] caused by excessive pressure, and *1276the pressure [was] caused by pent-up energy.”); Bower v. Aetna Ins. Co., 54 F.Supp. 897, 898 (N.D. Tex. 1944) (“The application of a force from within the radiators which the radiators, or the pipes, could not resist, and burst, or exploded is apparently what happened.”); Sperling v. Allstate Indemnity Company, 182 Vt. 521, 944 A.2d 210, 217 (2007) (“In the absence of the release of energy through an ignitable substance, decisions require a buildup of internal pressure preceding the rupture in order to define the event as an explosion”).
To convince us that the term explosion includes causes not driven by internal pressure, the Owner points to Stone Container Corp. v. Hartford Steam Boiler Inspection & Ins. Co., 165 F.3d 1157, 1160 (7th Cir. 1999), which states in dictum that a watermelon explodes when shot through with a bullet. But that statement does not help the Owner. The court was asked to limit the term explosion to an event caused by “combustion or some other chemical reaction.” Id. at 1159. It declined, explaining that to limit explosion in that way would improperly exclude from the term’s scope an atomic bomb or “volcanic explosions, as well as the ‘explosion’ of a tire caused by a blowout, the explosion of a melon caused by a bullet....” Id. at 1160. This comment reflects a (seemingly correct) belief that the melon explodes due to some source other than combustion or chemical reaction. It does not suggest that the source is something other than a buildup of internal pressure. And that is, in fact, the case, as far as we can discern: a bullet entering a watermelon compresses the fluid within it and creates a hydrostatic shock wave through the fluid that presses in all directions against the rind until it bursts. See “Hydrostatic sliock,” available at https://en.wikipedia.org/wiki/ Hydrostatic_shock (last visited August 2, 2016); “Why do watermelons explode,” available at http://bit.ly/lqC0OfW (last visited August 2, 2016).
Given our understanding of the policy language, we must affirm the district court. The Owner does argue briefly that a buildup of internal pressure did damage the building. Drawing an analogy to the popping of an overfilled balloon, the Owner suggests “that- mud, water, and debris may have filled up the Building to the point of failure, at which time the walls burst outward in a catastrophic and sudden explosion.” Aplt. Br. at 55. The problem for the Owner is that there is no evidence to support this argument. The Owner’s own engineer found that the north-traveling mudslide displaced the building’s walls upon impact, not after filling the building with mud, See Fronapfel Report, Supp. Aplee. App., Vol. 1 at 65 (“The debris laden flow impacted the south elevation of the structure, causing a sudden reaction of the wall structure.... ”).6 The walls moved laterally to the north, not outward in all directions. And the roof did not burst outward as from an explosion, but rather “collapsed.” Id. at 73.
V. CONCLUSION
We AFFIRM the district court’s orders denying the Owner’s motion to remand *1277and granting summary judgment to the Insurer. Our affirmance moots the Owner’s other arguments on appeal.