Hartford Accident and Indemnity Company (“Hartford”) appeals the district court’s grant of summary judgment to Intel Corporation (“Intel”). The district court held that the insurance policy issued by Hartford to Intel covered expenses incurred by Intel pursuant to a consent decree it had entered into with the United States Environmental Protection Agency (the “EPA”); under the consent decree, Intel agreed to clean up toxic waste contamination at the site of a former manufacturing facility.

We affirm in part, reverse in part and remand.

BACKGROUND

I.

Intel is a manufacturer of semiconductors. Its headquarters are located in Santa Clara, California. During the late 1960’s through the early 1980’s, Intel maintained a number of production facilities in Northern California. This case involves an Intel facility located on Middlefield Road in Mountain View, California. The property was leased to Intel by Renault & Handley.

Intel engaged in manufacturing on the property from 1968 through 1980. As part of the manufacturing process Intel used chemical solvents which contained various combinations of the following substances: 1,1,1,-trichloroethane, trichloroethylene, trichlorobenzene, dichloroethylene, phenol, and xylene. Each of these substances has been classified as a “hazardous substance” within the meaning of the Comprehensive Environmental Response, Compensation, and Liability Act of 1980 (“CERCLA”).

Intel stored these hazardous chemicals in an underground storage tank. Evidently, the tank was meant only for temporary use. The chemicals were to be transported eventually to another location for permanent disposal.

In 1980 Intel discontinued manufacturing operations at the Mountain View facility. Because Intel’s lease continued through 1984, it sought to sublet the property. As part of this process Intel commissioned soil sampling and testing of the grounds encompassing the area. The tests revealed that the site was contaminated by hazardous waste solvents present both in the soil and in the groundwater beneath the soil.

Intel contacted several government agencies about the problem and initiated a more thorough investigation of the contamination. Soon thereafter Intel began cleanup efforts by excavating the storage tank, removing the adjoining soils, and employing a pump-absorption filtration system to decontaminate the underground water.

In August, 1985, Intel entered into a consent decree with the EPA for the purpose of decontaminating the site. The purpose of the consent decree was to ensure that Intel, a “potentially responsible party” (“PRP”), would clean up the waste quickly and to eliminate the need for further legal proceedings. As part of the decree, Intel accepted responsibility without admitting liability.

On January 10, 1985, the Middlefield Road site had been listed on the California State Priority List by the California Department of Health Services in accordance with the state’s Health and Safety Code Section 25356. In the spring of 1985, the California Regional Water Quality Control Board issued separate waste discharge requirements to Intel requiring the company to prepare and implement plans for interim containment and cleanup at its facility. The state notified the EPA at this time of its actions and requested federal intervention.

On May 15, 1985, the EPA contacted Intel and indicated its intent to conduct a remedial investigation and feasibility study at the site. The EPA, however, offered Intel the option of undertaking both the investigation and cleanup itself. Intel submitted a proposal; the EPA accepted it. It was embodied in a consent decree. In the decree the EPA made findings that the site had been contaminated with the chemicals mentioned above, as well as others, to a degree warranting the site’s inclusion on the National Priorities List. A hazard posed by the Middlefield Road site was its proximity to a major public water supply (the site lies within a half mile).

The EPA certified Intel’s work plan as being consistent with the relevant standards and also noted that “all costs reasonably incurred for such work are necessary costs of response.” Additionally, Intel and its co-respondents, Fairchild and Raytheon, were compelled to remit $50,000 to the EPA to cover the agency’s costs for oversight and response costs in connection with the cleanup.

II.

From April 1,1976 through April 1, 1983, Intel was insured under a series of insurance policies with Hartford; these policies covered the Mountain View facility. These “comprehensive general liability” (“CGL”) insurance policies covered a limited array of risks incurred by Intel with respect to the subject property. The policy in effect from April 1, 1981 to April 1, 1982 (the “Policy”) contained the following description of coverage:

It also contained a list of exclusions, including exclusion (f), under which the Policy did not apply

and exclusion (k), under which the Policy did not apply

Intel contacted Hartford in October 1981 and told its insurer about its discoveries at the Mountain View plant. Intel submitted a claim for reimbursement for the reasonable and necessary investigation and cleanup costs incurred by Intel in connection with the facility. The basis of Intel’s claim was the quoted insuring clause and another portion of the Policy defining an “occurrence.” The Policy defines an “occurrence” as “an accident, including continuous or repeated exposure to conditions, which results in bodily injury or property damage neither expected nor intended from the standpoint of the insured.”

According to the district court, it is not clear when Intel actually filed its claim, but it is clear that Hartford denied the claim in a letter dated May 19, 1982. Hartford cited exclusion (k), the exclusion precluded coverage for damage to “property owned or occupied by or rented to the insured.” Earlier, Hartford had informed Intel that other exclusions might apply: in a letter dated March 31, 1982, it directed Intel’s attention to exclusion (f), as well as to exclusion (k) and others. In addition, according to Intel’s complaint and Hartford’s answer, on November 19, 1982, Hartford informed Intel that exclusion (f) would bar coverage.

After subsequent negotiations as to coverage failed, Intel brought suit against Hartford in California’s Superior Court in the County of Santa Clara. Hartford successfully removed the case to federal court on November 20, 1986. The district judge dismissed the action without prejudice because of a lack of derivative jurisdiction under the applicable removal statute and lack of diversity. See Intel Corp. v. The Hartford Accident and Indemnity, Co., 662 F.Supp. 1507 (N.D.Cal.1987).

On June 2, 1987, Intel refiled in California Superior Court in Santa Clara County. The case was again removed to federal court. On August 24, 1987, Intel filed its original motion for summary judgment on the issue of coverage. The district court granted Intel’s motion for summary judgment on April 28, 1988. The district court held, in essence, that “as a matter of law ... all expenses incurred by Intel pursuant to the consent decree are governmentally-mandated cleanup expenses which are fully compensable under the terms of the Hartford Comprehensive General Liability policy.” Intel Corp. v. Hartford Acc. and Indem. Co., 692 F.Supp. 1171, 1194 (N.D.Cal.1988) {“Intel”).

Hartford filed a motion for reconsideration. In August 1988, the court issued an amended opinion again granting Intel summary judgment. This appeal followed.

DISCUSSION

We review de novo a district court's grant of summary judgment. Kruso v. International Tel. and Tel. Corp., 872 F.2d 1416 , 1421 (9th Cir.1989), cert. denied, — U.S.-, 110 S.Ct. 3217 , 110 L.Ed.2d 664 (1990). In our review, we may consider any ground supported in the record. Swenson v. United States Postal Serv., 890 F.2d 1075 , 1077 n. 1 (9th Cir.1989).

Because this case was removed to federal court on the basis of diversity jurisdiction, we apply California substantive law. James B. Lansing Sound Inc. v. National Union Fire Ins. Co. of Pittsburg, PA, 801 F.2d 1560, 1564 (9th Cir.1986). We apply California law as we believe the California Supreme Court would apply it. Insurance Co. of North America v. Howard, 679 F.2d 147, 149 (9th Cir.1982).

I. Propriety of Summary Judgment Grant

We address below each issue in the district court’s grant of summary judgment. However, Hartford also argues that the grant of summary judgment was improper and unfair in a more general sense.

Hartford claims that summary judgment was awarded precipitously, on issues not briefed by either party. Quoting Cool Fuel, Inc. v. Connett, 685 F.2d 309, 312 (9th Cir.1982), it argues that it has been deprived of a “full and fair opportunity to litigate the issues involved in the motion.”

In the Notice of its Motion for Summary Judgment, Intel presented the grounds for its motion thus:

Intel described the issues on which it sought decision in expansive terms. Yet the Notice clearly advised Hartford that Intel sought an answer to the broad question of whether “insurance coverage exists” under the policy. As its opinion shows, to answer this question the district court was required to answer a number of smaller questions. Neither party chose to break the broad question down and focus on every one of these sub-questions in its brief. However, Hartford cannot complain that it was not warned of what was at stake in the summary judgment, and that it had no opportunity fully to present its position.

Hartford also suggests that summary judgment was improper because it did not have sufficient opportunity to conduct discovery and develop the facts in the case. Yet Intel filed its original suit against Hartford in May, 1986. It had first notified Hartford of contamination problems at the Middlefield Road site in 1981. Hartford thus had ample time to undertake more informal factfinding and, once the suit was filed, to conduct formal discovery. Hartford has not been “shortchanged” on discovery. See Beneficial Standard Life Ins. Co. v. Madariaga, 851 F.2d 271, 277 (9th Cir.1988) (affirming summary judgment where there was “more than six months” between appellant’s initial appearance and grant of summary judgment). Additionally, if it could have identified specific information it needed to respond adequately to the summary judgment motion, Hartford could have moved for a continuance of discovery under Federal Rule of Civil Procedure 56(f).

Thus, because Hartford had sufficient notice of what was to be decided, and opportunity to conduct discovery or request additional time for discovery, we find no merit in its claims of unfairness.

II. Existence of an “occurrence” within the meaning of the Policy

Under the Policy, Hartford was obliged to pay damages Intel became liable to pay that were “caused by an occurrence.”

Although its opinion does not focus on this point, in finding that coverage existed under the Policy, the district court implicitly found that there was no dispute that an “occurrence” had taken place.

A. Burden of Proof

In insurance litigation, “[w]hile the burden is on the insurer to prove a claim covered falls within an exclusion, the burden is on the insured initially to prove that an event is a claim within the scope of the basic coverage.” Royal Globe Ins. Co. v. Whitaker, 181 Cal.App.3d 532 , 226 Cal.Rptr. 435, 437 (1986) (citations omitted). Here, the parties disagree as to who bears the burden of proof on the “occurrence” issue. Hartford suggests that as the phrase “caused by an occurrence” is found in the general description of policy coverage, Intel had the burden of proving that the contamination constituted an “occurrence,” and that Intel failed to meet that burden in its motion for summary judgment. Intel responds suggesting that we follow the lead of Clemco Industries v. Commercial Union Ins. Co., 665 F.Supp. 816 (N.D.Cal.1987) (“Clemco ”), aff'd, 848 F.2d 1242 (9th Cir.1988) and find that the “occurrence” language is an exclusion, for which Hartford bears the burden of proof.

In Clemco, the court rejected the argument that the location of the “occurrence” language within the policy is decisive. It reviewed the background of the clause. The “occurrence clause” came into use in standard insurance policies in about 1966; it was meant to serve the same “ ‘exclusion’ objective” as a clause, previously in use, which excluded coverage for damages arising from “ ‘bodily injury or property damage caused intentionally by or at the direction of the insured.’ ” James L. Rigel-haupt, Jr., Annotation, Construction and Application of Provision of Liability Insurance Policy Expressly Excluding Injuries Intended or Expected by Insured, 31 A.L.R. 4th 957 , 972 (1984). The Clemco court concluded, “The placement of the phrase, however, in no way changed the effect or character of the phrase; ‘expected or intended’ remained an exclusion of the coverage grant by the very operation of its terms.” Clemco, 665 F.Supp. at 820 .

The Clemco court thus took a “functional” view of the occurrence clause. However sound its reasoning may be, Clemco’s support in California caselaw is not so sturdy. Clemco relies principally on U.S. Fidelity & Guar. v. Am. Employers Ins., 159 Cal.App.3d 277 , 205 Cal.Rptr. 460 (1984) (“U.S. Fidelity”) for the proposition that “those California courts that have had occasion to interpret the same language have also found that the ‘expected or intended’ phrase is in fact an exclusion.” Clemco, 665 F.Supp. at 820 . In U.S. Fidelity, at issue was section 533 of the California insurance code, which provides that “[a]n insurer is not liable for a loss caused by the willful act of the insured.” Cal.Ins.Code § 533 (West 1972). The court characterized this statute as “an implied exclusionary clause which by statute is to be read into all insurance policies.” U.S. Fidelity, 205 Cal.Rptr. at 464 . However, the reasoning of U.S. Fidelity does not apply to the Clemco facts merely because a reference to intentional or willful acts of the insured appears in both Insurance Code section 533 and the occurrence clause.

Moreover, there are California cases which would support a result contrary to Clemco. For example, in Royal Globe Ins. Co. v. Whitaker, 181 Cal.App.3d 532 , 226 Cal.Rptr. 435 (1986) (“Royal Globe”), the insured argued that the insurer was required to defend and indemnify him when he was sued for fraud, emotional distress and breach of contract. The policy at issue contained an occurrence clause. The court held that the insured was required to show that the damages for which he was liable were caused by an accident. Thus, although the Clemco rationale was not addressed in Royal Globe, Royal Globe took the more “bright line” view that the location of the occurrence clause in the description of basic policy coverage was decisive. See also Hartford Fire Ins. Co. v. Karavan Enterprises, 659 F.Supp. 1075 (N.D.Cal.1987) (citing Royal Globe; insured failed to show termination of employee was an “occurrence”); Dyer v. Northbrook Property & Cas. Ins., 210 Cal. App.3d 1540 , 259 Cal.Rptr. 298 (1989) (citing Royal Globe; insured must show termination of employee was an “occurrence”).

Thus, existing caselaw provides no clear answer as to how a California court would allocate the burden of proof on the “occurrence” issue. However, we need not decide this question as we hold that, even if it bears the burden of showing the contamination constituted an “occurrence,” Intel has met that burden.

B. Intel’s showing that the contamination constituted an “occurrence”

Summary judgment will be granted “if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.” Fed. R.Civ.P. 56(c). If the moving party meets its initial burden of showing “the absence of a material and triable issue of fact,” “the burden then moves to the opposing party, who must present significant probative evidence tending to support its claim or defense.” Richards v. Neilsen Freight Lines, 810 F.2d 898, 902 (9th Cir.1987). The non-moving party has failed to meet its burden if “the record taken as a whole could not lead a rational trier of fact to find for the non-moving party.” Matsushita Elec. Industrial Co. v. Zenith Radio, 475 U.S. 574, 587 , 106 S.Ct. 1348, 1356 , 89 L.Ed.2d 538 (1986).

As noted above, the district court did not focus on the “occurrence” issue; the parties, too, did not argue it in detail. However, the materials supporting Intel’s motion clearly present a scenario of unexpected and unintended environmental damage. Affidavits adduced by Intel indicate company employees learned of contamination only after a prospective sublessor requested Intel to conduct soil sampling and testing. Thus, Intel met its initial burden.

The district court properly granted summary judgment on this issue because Hartford, when the burden shifted to it, failed to show that there remained genuine issues of fact. In opposing Intel’s motion, Hartford neither presented any additional evidence disputing Intel’s showing, nor pointed to facts already before the court that revealed the existence of a dispute.

Hartford asks the court to “infer” that “Intel’s contamination of its property was expected, if not intended.” As the non-moving party, Hartford is entitled to have all reasonable inferences drawn in its favor. Matsushita, 475 U.S. at 587 , 106 S.Ct. at 1356 . However, Hartford’s proposed “inferences,” presented in its brief as a series of insinuating questions, resemble tenuous speculations rather than potentially valid conclusions that could be grounded in evidence in the record. Hartford also suggests that “[i]n similar cases discovery has consistently revealed that industrial polluters did know that their toxic chemical handling practices were causing, and would in the future cause, pollution.” Hartford in effect asks this court to take judicial notice that all industrial polluters are aware that they are polluting. But the facts of other cases cannot help Hartford’s effort to fend off summary judgment in this one.

In conclusion, the district court did not err in granting summary judgment on the “occurrence” issue.

III. Applicability of Exclusion (f)

Two exclusions in the policy were arguably relevant to coverage of the contamination at the Middlefield Road site: exclusion (f) and exclusion (k). We discuss first the former.

A. Waiver

The district court found that Hartford had waived reliance on exclusion (f). When Hartford, in its letter of May 19, 1982, informed Intel that it was denying coverage, it cited only exclusion (k). Relying on McLaughlin v. Connecticut General Life Ins. Co., 565 F.Supp. 434 (N.D.Cal.1983) {“McLaughlin”), the district court held that Hartford had thus waived reliance on any other grounds not mentioned in that letter. In McLaughlin , which will be discussed further below, the court held that “an insurance company which relies on specified grounds for denying a claim thereby waives the right to rely in subsequent litigation on any other grounds which a reasonable investigation would have uncovered.” McLaughlin, 565 F.Supp. at 451 .

Under California law, waiver is a question of fact. Waiver is an affirmative defense, for which the insured bears the burden of proof. Insurance Co. of the West v. Haralambos Beverage Co., 195 Cal.App.3d 1308 , 241 Cal.Rptr. 427 (1987).

California courts will find waiver when a party intentionally relinquishes a right, or when that party’s acts are so inconsistent with an intent to enforce the right as to induce a reasonable belief that such right has been relinquished. Rheem Mfg Co. v. United States, 57 Cal.2d 621 , 21 Cal.Rptr. 802 , 371 P.2d 578, 581 (1962). The doctrine of waiver looks to “the act, or the consequences of the act, of one side only,” in contrast to the doctrine of estoppel, which “is applicable where the conduct of one side has induced the other to take such a position that it would be injured if the first should be permitted to repudiate its acts.” 30 Cal.Jur.3d Estoppel and Waiver 2 (1987).

In the insurance context, however, the waiver doctrine has developed somewhat differently, and the distinction between waiver and estoppel has been blurred. In cases where waiver has been found, there is generally some element of misconduct by the insurer or detrimental reliance by the insured. For example, in Miller v. Elite Ins. Co., 100 Cal.App.3d 739 , 161 Cal.Rptr. 322 (1980) {“Miller”), the court held that the insurer had waived its right to deny coverage for damages caused by an accident in which the insured was involved. The court found that, “by its conduct in proceeding to deal with [the other party’s insurer] as though [the insurer] intended to represent and defend [the insured], created the impression that there was no coverage dispute.” Moreover, the insurer never informed the insured that it intended to reserve its rights to assert noncoverage. Miller, 161 Cal.Rptr. at 328 ; see also Elliano v. Assurance Co. of America, 3 Cal.App.3d 446 , 83 Cal.Rptr. 509 (1970) (insurer waived requirement that insured commence any action on insurance policy within one year when insurer engaged in settlement negotiations with insured that lasted beyond the deadline for filing suit, admitted at all times that insured had suffered a loss, and never notified insured of its intention to rely on the limitation); Dalzell v. Northwestern Mut. Ins. Co., 218 Cal.App.2d 96 , 32 Cal.Rptr. 125, 128 (1963) (insurer waived requirement that insured forward summons to it where insurer knew suit had been filed against insured and told insured “everything would be taken care of”); Ruffino v. Queen Ins. Co. of America, 138 Cal.App. 528 , 33 P.2d 26 (1934) (insurer waived requirement that insured present sworn proof of loss within sixty days of loss when insurer repeatedly told insured, in response to insured’s inquiries, that insured was required to do nothing further to obtain settlement of claim).

Federal cases applying California law also suggest such misconduct or reliance must be shown. See Zumbrun v. United Services Auto. Ass’n, 719 F.Supp. 890 (E.D.Cal.1989) {“Zumbrun") (in lawsuit filed after it denied a claim, insurer waived affirmative defenses based on policy exclusions insurer failed to assert in letter denying coverage issued fifteen months after insured filed claim); Stinson v. Home Ins. Co., 690 F.Supp. 882 (N.D.Cal.1988) (refusing to find waiver of limitation of actions clause where there was no evidence that the insurance company had taken affirmative steps to deter insured from filing suit within the limitations period); Becker v. State Farm Fire and Cas. Co., 664 F.Supp. 460 (N.D.Cal.1987) (refusing to find waiver of limitation of actions clause where insured did not inform insured of its loss until after the limitations period had expired); McLaughlin (discussed below).

McLaughlin is consistent with this trend. The language cited above, and relied on in the district court’s decision, suggests a more absolute rule: whatever exclusions an insurer fails to include in a letter denying coverage are lost forever. However, the facts in McLaughlin , and the policies behind the decision, suggest that a finding of waiver should not be so automatic.

In McLaughlin , the insured sought coverage for experimental cancer treatment. The insurer denied coverage; in the period leading up to litigation it based its denial on the fact that the treatment was not approved by the FDA, and informed the insured that this was the only basis for its denial. Apparently, the insurer conducted only minimal investigation of the insured’s treatment. When the insured sued, however, the insurer invoked a series of other potentially applicable exclusions it had not previously mentioned. The district court held that the insured had waived these exclusions, which it had not investigated before denying coverage, and had not raised when it denied coverage. McLaughlin, 565 F.Supp. at 451 .

Thus, in McLaughlin it was necessary to find waiver to protect insureds who had been mislead by the insurer’s statements as to the denial of coverage. The policy grounds set forth by the McLaughlin court, and subsequently by the court in Zumbrun , make clear the rationale of protecting insureds and providing incentives for insurers to proceed properly. McLaughlin relied on Egan v. Mutual of Omaha, 24 Cal.3d 809 , 169 Cal.Rptr. 691 , 620 P.2d 141 (1979), in which the California Supreme Court found that an insurer has a duty to investigate thoroughly the claims of an insured. The McLaughlin waiver doctrine is a means of enforcing this requirement: “If an insurance company could deny a claim without thoroughly investigating it and then defend a subsequent lawsuit on grounds which it develops during discovery for trial, the company’s incentive to fulfill its duty to investigate would be significantly diminished.” McLaughlin, 565 F.Supp. at 451 . Additionally, the waiver doctrine prevents an insurer from “sandbagging” the insured by constantly raising additional exclusions. The waiver rule “denies [the insurer] multiple opportunities to avoid liability under terms of the policy that are ambiguous or disputed.” Zumbrun, 719 F.Supp. at 896 .

Applying the waiver doctrine as it has been applied by California courts and federal courts applying California law, we find that the district court erred in granting summary judgment on the exclusion (f) issue on grounds of waiver. Intel has failed to make the required showing as to waiver by Hartford. The record showed that Intel was informed on March 31, 1982 that exclusion (f) might apply. While there was no mention of that exclusion in the May 19, 1982 letter denying coverage, Hartford mentioned the exclusion again when it reiterated its denial of coverage on November 19, 1982. Intel points to Hartford in-house communications in which employees expressed doubt as to whether exclusion (f) would apply; however, Intel did not know of such communications at the time they were made and thus could not have relied on them to its detriment. There is no evidence Hartford attempted to “sandbag” or mislead Intel with the belated announcement of a new grounds for denial of coverage, nor has Intel shown it was prejudiced by the failure to mention exclusion (f) in the May 19, 1982 letter.

Thus, on these facts, the district court should not have found Hartford waived reliance on exclusion (f). However, we nonetheless affirm the grant of summary judgment on the application of exclusion (f). We find Hartford failed to meet its burden of showing that summary judgment on the issue of coverage was precluded because there was a genuine issue of fact as to the application of exclusion (f).

B. Absence of Issue of Fact as to Applicability of Exclusion (f)

As the district court noted, “courts have differed” as to the construction and application of pollution exclusions such as exclusion (f). The trend appears to be towards an application of the exclusion according to the “plain, everyday meaning” of its terms. See, e.g., FL Aerospace v. Aetna Casualty and Sur. Co., 897 F.2d 214 (6th Cir.) (applying Michigan law) (pollution exclusion barred coverage), cert. denied, — U.S. -, 111 S.Ct. 284 , 112 L.Ed.2d 238 (1990); United States Fidelity and Guar. Co. v. Star Fire Coals, Inc., 856 F.2d 31 (6th Cir.1988) (applying Kentucky law) (language of pollution exclusion was not ambiguous; pollution exclusion barred coverage); see also C.P.C. Intern, v. Northbrook Excess & Surplus Ins., 759 F.Supp. 966 (D.R.I.1991) (applying New Jersey law) (language of pollution exclusion was not ambiguous; listing other cases which have also so held). However, a significant number of courts have refused to apply the exclusion, holding that because its language is ambiguous, the clause must be construed against the insurer. See, e.g., New Castle Cty. v. Hartford Accident and Indem. Co., 673 F.Supp. 1359 (D.Del.1987) (applying Delaware law) (term “sudden” was ambiguous, and ambiguity should be resolved in favor of the insured; pollution exclusion did not bar coverage); National Grange Mut. Ins. Co. v. Continental Ins. Co., 650 F.Supp. 1404 (S.D.N.Y.1986) (applying New York law) (language of pollution clause was ambiguous and ambiguity should be resolved in favor of the insured; pollution clause did not bar coverage); see also Claussen v. Aetna Casualty & Sur. Co., 865 F.2d 1217 (11th Cir.1989) (certifying question of application of pollution clause under Georgia law to Georgia Supreme Court; listing cases in which the clause has been applied on its terms and in which courts have found it ambiguous). California courts have apparently not had occasion to rule on this issue.

We need not enter this controversy, however, for we find that even if exclusion (f) is applied on its terms, Hartford has failed to meet its burden of showing that there is an issue for the trier of fact as to whether it bars coverage in this case.

Here, after Intel made its showing that there remained no genuine issues of material fact as to coverage under the policy, the burden shifted to Hartford to show that such issues did exist. As noted above, the insurer has the burden of proof as to the application of exclusions. Hartford failed to show that there existed evidence such that a reasonable jury might find for it on the application of exclusion (f). In the materials supporting its opposition, it made no showing that the contamination at the Middlefield Rd. site fell within the exclusion. “[T]he plain language of Rule 56(c) mandates the entry of summary judgment, after adequate discovery and upon motion, against a party who fails to make a showing sufficient to establish the existence of an element essential to that party’s case, and on which that party will bear the burden of proof at trial.” Celotex Corp. v. Catrett, 477 U.S. 317, 322-23, 106 S.Ct. 2548, 2552-53 , 91 L.Ed.2d 265 (1986). Admittedly, Intel made no showing that the exclusion did not apply. However, there is “no express or implied requirement in Rule 56 that the moving party support its motion with affidavits or other similar materials negating the opponent’s claim.” Id.

Therefore, although on grounds other than those relied on by the district court, we find that the district court properly granted summary judgment on the issue of exclusion (f).

IV. Application of Exclusion (k)

A. Cleanup costs as “damages”

The Policy provides coverage for “all sums which the insured shall become legally obligated to pay as damages because of ... property damage to which this insurance applies.” In holding that Intel was entitled to summary judgment on the issue of coverage, the district court determined that the expenses Intel incurred pursuant to the consent decree constituted “damages” within the meaning of the policy. We affirm the district court on this issue. Relying on the decision of the California Supreme Court in AIU Ins. Co. v. FMC Corp., 51 Cal.3d 807 , 274 Cal.Rptr. 820 , 799 P.2d 1253 (1990) (“AIU’), issued subsequent to the district court’s decision in this case, we hold that these expenses are “damages” within the meaning of the policy.

1. AIU Ins. Co. v. FMC Corp.

FMC was insured by AIU under a series of policies which contained “damages” language identical or similar to the clause in Intel’s policy with Hartford. The U.S. and local administrative agencies (collectively, the “agencies”) filed suit against FMC for alleged violations of CERCLA and other environmental statutes. The agencies sought injunctive relief compelling decontamination of FMC waste disposal sites, and reimbursement for the agencies’ costs of investigating and monitoring waste disposal and initiating cleanup. FMC sought recovery from its insurer for costs incurred pursuant to these suits.

The California Supreme Court held that FMC’s costs of complying with the injunction and the reimbursement it was required to pay were “sums ... [FMC became] legally obligated to pay as damages because of ... property damage.” It looked first at the phrase “legally obligated,” and found that although “both injunctions and awards of response costs under CERCLA reasonably can be viewed as equitable relief,” “as a matter of plain meaning, the term ‘legally obligated’ covers injunctive relief and recovery of response costs.” AIU, 799 P.2d at 1266 .

The court then examined the two forms of relief the agencies sought. It first explored more generally the term “damages.” It noted that, while statutory and dictionary definitions of “damages” were relatively straightforward, the application of this term to the remedies available to the agencies under CERCLA creates some ambiguity:

Id. at 1268-69.

The court then held that reimbursement of government response costs constituted “damages” within the meaning of the policy:

Id. at 1269.

The court rejected several arguments against its holding. In particular, it rejected the argument that the distinctions or definitions contained in CERCLA with regard to the types of available relief were relevant to its determination: at issue was the interpretation of the insurance policy under state law. It also rejected the argument that the agencies’ remedial action was “prophylactic in nature”:

Id. at 1272.

The court then turned to the issue of the costs of compliance with injunctions. It acknowledged that “[t]he costs of injunc-tive relief, whether incurred for prophylactic, mitigative, or remedial purposes, do not readily satisfy the statutory or dictionary definitions of ‘damages.’ ” Id. at 1276. However, it observed that the relationship between injunctive and reimbursement relief under CERCLA “is not the same as that between damages and injunctive remedies traditionally available at common law and in equity.” Id. Government agencies seek injunctive relief rather than incur costs themselves because “government cleanup efforts are generally considerably more expensive than cleanups performed by the responsible party.” Id. at 1275. “Under CERCLA and similar statutes, in-junctive relief and reimbursement of response costs serve substantially the same purpose.” Id. at 1278. “It is unlikely ... that the parties to CGL policies intended to cover reimbursement of response costs but not the costs of injunctive relief, at least where the latter costs are incurred — generally at a lower total cost — for exactly the same purposes addressed through governmental expenditure of response costs.” Id. Thus, the court found: “CGL policy language is ambiguous as applied to remedial and mitigative costs incurred pursuant to injunction under CERCLA and similar statutes, and therefore must be construed in favor of coverage to satisfy the reasonable expectations of the insured.” Id. It would apply this construction “regardless of any formal or technical difficulties this reading poses.” Id. at 1278 n. 18.

In concluding, the court interpreted the phrase “because of property damage.” It held that all costs arising from the contamination at issue were incurred “ ‘because of’ property damage,” whether the cleanup took place on the property of FMC or of third parties: “The provisions at issue here do not specify that coverage hinges on the nature or location of property damage. We therefore construe them to encompass damages because of property damage in general, regardless of by whom it is suffered.” Id. at 1279. However, the court also noted one exception to the insurer’s obligation to cover cleanup costs: costs incurred to pay for prophylactic measures, “measures taken in advance of any release of hazardous waste,” were not covered by the policies. Id.

2. Application of AIU to consent decrees

Under AIU, costs incurred to comply with an injunction mandating cleanup or to reimburse a government agency for cleanup expenses the agency has incurred are “damages” within the meaning of the insurance policy. We must now determine, however, whether the California Supreme Court would extend this holding and find that costs incurred to comply with a consent decree are “damages.” We conclude that it would.

The first issue is whether such costs are sums the insured are “legally obligated” to pay. That the insured voluntarily assumed the obligation to conduct cleanup, rather than forcing the government to assume the expenses of a coercive suit or of cleanup itself, should not change the analysis that a legal obligation to be responsible for cleanup does exist.

Our analysis of whether consent decree costs are “damages” parallels the AIU court’s discussion of injunctive relief. Consent decrees are an alternative means for government agencies to ensure that cleanup takes place. They are a favored method, as they provide for more rapid, and less costly, cleanup than does full scale litigation. See 42 U.S.C. § 9622 (a) (authorizing the President to “facilitate agreements” with potentially responsible persons “in order to expedite effective remedial actions and minimize litigation”). Faced with an obligation that it will have to meet, sooner or later, an insured would reasonably expect that, if it acts responsibly and cooperates with government agencies, it will not forego coverage to which it would be entitled if it forced the agencies to expend their resources in filing suit, particularly where cooperation will probably result in compliance at a lower cost to the insured (and thus to the insurer). Cf. AIU, 274 Cal.Rptr. at 845 (it is unlikely parties to CGL policies intended to cover reimbursement of response costs, but not costs of injunctive relief under which the parties arrive at the same result at a lower total cost); Globe Indem. Co. v. People, 43 Cal.App.3d 745 , 118 Cal.Rptr. 75, 79 (1974) (where insured is covered for damage to property of third parties, insured could reasonably expect coverage would also exist for costs of mitigating such damage). Coverage of cleanup costs should not turn on the fortuity of the form in which the offending company fulfills its cleanup responsibilities. See AIU, 799 P.2d at 1278 (espousing rationale that making insurance coverage hinge on “ ‘mere fortuity’ ” of the government’s choice of enforcement methods would “introduce substantial inefficiency in the cleanup process”). Thus, we hold that the “damages” language is ambiguous as to coverage of this alternative means of CERCLA enforcement as well and, construing the clause, as we must, against the insurer, Hanson v. Prudential Insurance Co., 783 F.2d 762, 764 (9th Cir.1985), we find costs incurred pursuant to a consent decree are covered “damages.”

Public policy supports this holding. If consent decree compliance costs did not constitute “damages,” insureds would be discouraged from entering into consent decrees, and the EPA’s task would be made more time consuming and more costly. Cleanup would be delayed until the government expended its resources to investigate contamination, and perhaps even to conduct cleanup itself, and finally sought relief from the insured. See Stephen Moun-tainspring, Insurance Coverage of CERC-LA Response Costs: The Limits of “damages" in Comprehensive General Liability Policies, 16 Ecology L.Q. 755 , 797-98 (1989) (insurance coverage of response costs would encourage PRP cooperation and thus contribute to conserving government resources). The policy arguments on the other side are not persuasive. Arguably, insureds, confident of the “deep pockets” of the insurer, will have no incentive to minimize their liability by finding the least costly means of decontamination, assuming cleanup obligations only when they are clearly responsible, and preventing contamination in the first place. See Maryland Casualty Co. v. Armco, Inc., 822 F.2d 1348, 1355 (4th Cir.1987) (insured would tend to overuse “free” resource of insurance), cert. denied, 484 U.S. 1008 , 108 S.Ct. 703 , 98 L.Ed.2d 654 (1988). However, certain checks remain in place. The insured will always be concerned that the insurer might deny coverage under an exclusion, and thus will have an incentive to minimize decontamination costs and to prevent pollution. Also, it defies common sense to suggest that a party ever has an incentive voluntarily to assume liability for contamination for which it does not strongly believe it is responsible. Finally, we rely on the good faith of insureds to conduct responsibly their negotiations with government agencies.

Potentially more troubling is the difference between consent decrees and suits for injunctive relief or reimbursement. Such suits have a coercive dimension not present in the context of a consent decree. A consent decree is more akin to a settlement. Generally, insurers have an active role in any settlement by the insured. In the case before us, though, the insured has had no such participation. As we have just noted, however, we believe insureds have adequate incentives to protect their own, and thus their insurers’, interests when they negotiate consent decrees. Insureds are unlikely carelessly to enter into overbroad consent decrees. Moreover, our holding here will give insurers incentives to do what Hartford has failed to do in this case: to follow more closely the insured’s efforts to meet its cleanup obligations.

B. Application of Exclusion (k)

The final issue considered by the district court was the extent to which coverage of cleanup costs was barred by exclusion (k). The district court acknowledged that it remained for the trier of fact to determine which pre-Consent Decree expenses related only to cleanup of Intel’s own property, and which related to property of third parties. However, as to expenses incurred pursuant to the consent decree, the court found “as a matter of law” that “all expenses incurred by Intel pursuant to the Consent Decree are governmentally-man-dated expenses which are fully compensa-ble under the terms of the Hartford comprehensive general liability policy.” Intel, 692 F.Supp. at 1194 . We find this holding to be too broad.

We affirm the district court’s holding insofar as it applies to costs incurred under the consent decree to correct any groundwater contamination that has taken place, and to mitigate any future damage that might occur from the contaminants introduced into the soil and groundwater by Intel, whether or not on Intel’s own property.

Damage to groundwater is not damage to property “owned or occupied by or rented to” Intel. “Release of hazardous waste into groundwater and surface water constitutes actual harm to property in which the state and federal governments have an ownership interest.” AIU, 274 Cal.Rptr. 820 at 836 ; see also Cal.Wat.Code 102 (West 1971) (“All water within the State is the property of the people of the State, but the right to use of water may be acquired by appropriation in the manner provided by law”).

Additionally, we hold that exclusion (k) also does not bar coverage of the costs of preventing future harm to ground water or adjacent property that might arise from contamination that has already taken place, whether such contamination has occurred on Intel’s or others’ property. We rely here on AIU and on Aerojet-General Corp. v. Superior Court, 211 Cal.App.3d 216 , 257 Cal.Rptr. 621 (1989) (“Aerojet ”). The AIU court specifically stated it was not addressing the application of exclusions such as exclusion (k). AIU, 274 Cal.Rptr. at 829 n. 7. However it did suggest that where an insured is covered for damage to a third party’s property, that insured would reasonably expect coverage for efforts to mitigate that damage, even when the source of the hazard is on the insured’s own property. Id. at 839 , citing Aerojet, 257 Cal.Rptr. at 627-28 , quoting Globe Indent. Co. v. People, 43 Cal.App.3d 745 , 118 Cal.Rptr. 75, 79 (1974) (“it would seem ‘strangely incongruous’ to the insured ‘that his policy would cover him for damages to tangible property destroyed through his negligence in allowing a fire to escape but not for sums incurred in mitigating such damages by suppressing the fire’ ”). While the AIU court was discussing here the “damages” language, we believe its reasoning applies as well to exclusion (k).

However, we believe that the trier of fact will nonetheless have to face the task of determining what expenses were incurred to remedy existing damage to third-party property or to prevent further damage to that property from contaminants introduced by Intel, and what expenses were incurred only to remedy damage to property Intel itself controlled. The former expenses are covered by the policy, while exclusion (k) bars coverage of the latter. We cannot endorse the district court’s solution of merely adopting the consent decree. The consent decree may be, as the district court puts it, a good measure of costs “ ‘reasonably incurred’ ” as “ ‘necessary costs of response.’ ” However, it does not sort those costs into the two relevant categories: damage to third party property and damage only to Intel’s own property.

The district court’s decision required the trier of fact to examine expenditures incurred prior to the Consent Decree to determine which costs were “consistent with or a foundation for the implementation of the Consent Decree.” Intel, 692 F.Supp. at 1194 . We hold that the trier of fact will have to examine the costs of complying with the decree as well, to determine what expenditures were made solely to remedy damage to Intel’s own property and which were undertaken on account of damage to the property of third parties.

AFFIRMED IN PART, REVERSED IN PART AND REMANDED.

. Fairchild Camera and Instrument Corporation (“Fairchild”) and Raytheon Corporation ("Ray-theon”) were also parties to the consent decree. . The National Priorities List is the EPA’s inventory of the most seriously contaminated hazardous waste sites in the United States. . Hartford cites several cases for the proposition that it is unfair to grant summary judgment against a party that has not had "a full and fair opportunity to litigate the issues.” However, each of these cases involves a district court’s sua sponte grant of summary judgment; in two, summary judgment was granted to the nonmov-ing party. See Horn v. City of Chicago, 860 F.2d 700 , 703 n. 6 (7th Cir.1988) (disapproving of district court’s “surprise’’ conversion of 12(b)(6) motion to dismiss into summary judgment motion against moving party); Williams v. City of St. Louis, 783 F.2d 114, 117 (8th Cir.1986) (district court’s sua sponte award of summary judgment was improper); Cool Fuel, Inc. v. Connett, 685 F.2d 309 (9th Cir.1982) (sua sponte award of summary judgment to nonmoving party was not improper). These cases do not support Hartford’s argument here. . For example: "Did Intel, with all its engineers and experts, know, but not follow, the type of precautions that were necessary to prevent the toxic chemicals from escaping after being dumped into an underground tank? Did Intel keep track of the quantities of chemicals it dumped into and removed from the tank so that Intel had to have known, based on the capacity of the tank, that the tank was leaking?" Hartford does not ask the court to make inferences on the basis of facts in the record, but rather to "infer” new facts. . Indeed, courts often use the two terms interchangeably. 30 Cal. Jur.3d Estoppel and Waiver 2 (1987). See, e.g., Insurance Co. of the West v. Haralambos Beverage Co., 195 Cal.App.3d 1308 , 241 Cal.Rptr. 427, 433 (1987) (elements of “prejudice or detrimental reliance ... are essential to waiver and estoppel"); see also Val's Painting & Drywall v. Allstate Ins. Co., 53 Cal.App.3d 576 , 126 Cal.Rptr. 267, 273 (1976) ("In California, however, the theory that by defending the suit an insurer ‘waives’ its right to claim noncover-age rests upon the doctrine of estoppel.”). . Although there are numerous decisions on CERCLA cleanup costs as “damages," there is a dearth of caselaw on the issue of consent decrees. In Aerojet-General Corp. v. Superior Court, 211 Cal.App.3d 216 , 257 Cal.Rptr. 621 (1989), the insureds entered into a consent decree after the Department of Justice and the State of California filed suit against it for in-junctive relief. However, in that case the insurers sought declaratory relief on the issue of their duty to indemnify the insureds for costs arising out of these suits; the decision focused on whether there was coverage for costs incurred pursuant to an injunction. See abo Aet-na Casualty and Sur. Co. v. Pintlar Corp., 948 F.2d 1507 (9th Cir.1991) (finding coverage for costs of compliance with consent decree entered into by insured after EPA filed administrative claims against it); Boeing Co. v. Aetna Casualty and Sur. Co., 113 Wash.2d 869 , 784 P.2d 507, 516 (1990) (finding coverage for costs of compliance with consent decree entered into by insured after EPA filed complaints against it) ("The costs assessed against the policyholders by the underlying lawsuits are covered by the subject policies to the extent these costs are because of property damage”).