JUSTICE BUSBY delivered the opinion of the Court. Mark and Birgit Self sued the Texas Department of Transportation (TxDOT), among other defendants, for negligence and inverse condemnation after employees of a TxDOT subcontractor cut down trees on the Selfs’ property outside the boundaries of the State’s right-of-way easement. The court of appeals held that TxDOT’s evidentiary plea to the jurisdiction failed on the negligence cause of action but should have been granted on the cause of action for inverse condemnation. We disagree. As to negligence, the Selfs have not shown either that the subcontractor’s employees were in TxDOT’s paid service or that other TxDOT employees operated or used the motor-driven equipment that cut down the trees, as required to waive immunity under the Tort Claims Act. Regarding inverse condemnation, however, the Selfs have alleged and offered evidence that TxDOT intentionally directed the destruction of the trees as part of clearing the right-of-way for public use. We therefore reverse the court of appeals’ judgment, render judgment dismissing the negligence cause of action, and remand the cause of action for inverse condemnation to the trial court for further proceedings. BACKGROUND The Selfs own a tract of rural land that adjoins a portion of Farm-to-Market Road 677 in Montague County and extends to the centerline of that road. The State has a right-of-way easement that reaches fifty feet from the centerline of the road in each direction and thus burdens part of the Selfs’ property. The Selfs’ predecessors constructed a fence along the edge of the easement, but the Selfs hired a contractor to remove this decaying fence and construct a new fence. The Selfs offered evidence that they instructed the fence contractor to “set the fence two to three feet on [the Selfs'] side of the [right-of-way] easement” to preserve large trees that had grown along the original fence and allow “the trees and fence [to] be maintained.” As a result, a strip of the Selfs’ property two to three feet wide outside the new fence was not burdened by the State’s right-of-way easement. TxDOT started a highway maintenance project and, as part of that project, contracted with T.F.R. Enterprises, Inc. (TFR) to remove brush and trees from the right-of-way. Their contract provided that “[t]rees to be removed shall be marked by the State,” either on the plans or with an X painted on the trunk, “PRIOR TO WORK BEING PERFORMED.” After TxDOT expressed concern about TFR’s ability to complete the project on time, TFR notified TxDOT that it would increase its production by “adding a separate tree removal crew.” TFR subcontracted with Lyellco Inc. to remove the trees. Following TxDOT’s revised instruction to TFR to “clear everything between the fences,” Lyellco workers cut all trees up to the Selfs’ fence line. After the Selfs complained, an email between TxDOT employees acknowledged that a TxDOT inspector “did direct the contractor to cut the trees down, but they were on the state highway side of the fence.” The Selfs sent a letter to TxDOT and attached a survey they had obtained, which showed that twenty-eight oaks and elms with trunk diameters ranging from eighteen to thirty-nine inches were removed near their fence line—thirteen of which were wholly outside the State’s right-of-wayand seven of which were partly outside it. TxDOT indicated in its interrogatory answers that it conducted “no surveys . . . in association with this project” and “TxDOT is not aware of any communications with [the Selfs] prior to clearing or maintaining of trees or vegetation on this project.” The Selfs obtained multiple estimates of the cost to replace the twenty felled trees that had been located wholly or partly outside the right-of-way with trees up to twenty inches in diameter (the largest commercially available), and they sought $251,000 from TxDOT to compensate them for this cost. TxDOT rejected their claim by letter, explaining that the Attorney General’s Office investigated the matter and concluded that TxDOT committed no act of negligence. The Selfs sued TxDOT, contractor TFR, and subcontractor Lyellco for removing trees from the Selfs’ land. They alleged two causes of action against TxDOT: negligence and inverse condemnation. TxDOT filed a plea to the jurisdiction asserting immunity from both causes of action. Regarding negligence, TxDOT argued that the Selfs had not shown a TxDOT employee negligently damaged the trees by operating or using motor-driven equipment. See TEX. CIV. PRAC. & REM. CODE § 101.021(1). The Selfs responded that sovereign immunity was waived because (1) the Lyellco employees who removed the trees were TxDOT “employees” within the meaning of Section 101.001, and (2) other TxDOT employees exercised such control over the motor-driven equipment used to remove the trees that they “operated or used” that equipment within the meaning of Section 101.021. Turning to inverse condemnation, as relevant here, TxDOT conceded that it directed TFR to remove the trees up to the fence line but argued that the Selfs failed to offer evidence that TxDOT intended to have trees removed from property outside its right-of-way. TxDOT pointed to undisputed evidence that the trees were on the state highway side of the fence and that the fence was not located at the edge of the right-of-way.[1] The Selfs responded that TxDOT acted with sufficient intent by countermanding the contract and ordering its contractor to remove all trees up to the fence, and they pointed to the TxDOT employee’s email acknowledging that a TxDOT inspector “did direct the contractor to cut the trees down.” The trial court denied TxDOT’s plea to the jurisdiction, and the court of appeals affirmed in part and reversed in part. As to the negligence cause of action, the court of appeals ultimately concluded there was a fact issue on whether the Texas Tort Claims Act waived immunity, holding: (1) the trial court erred in identifying a factual dispute regarding whether TxDOT operated or used motor-driven equipment under Section 101.021; but (2) the trial court correctly identified a fact issue about whether the Lyellco employees who did use the equipment were TxDOT “employees” under Section 101.001 rather than employees of an independent contractor. 683 S.W.3d 62, 73-88 (Tex. App.—Fort Worth 2022). Regarding the cause of action for inverse condemnation, the court of appeals reversed the trial court’s judgment and held there was no evidence that TxDOT intentionally destroyed the Selfs’ property. See id. at 88-91. Both TxDOT and the Selfs filed petitions for review, which we granted. ANALYSIS Sovereign immunity protects the State and its agencies and subdivisions from suit and liability, PHI, Inc. v. Tex. Juv. Just. Dep’t, 593 S.W.3d 296, 301 (Tex. 2019), thereby depriving trial courts of subject-matter jurisdiction over suits against them unless the State consents. Tex. Dep’t of Parks & Wildlife v. Miranda, 133 S.W.3d 217, 224 (Tex. 2004). Immunity is properly asserted in a plea to the jurisdiction, id. at 225-26, which “may challenge the pleadings, the existence of jurisdictional facts, or both.” Alamo Heights Indep. Sch. Dist. v. Clark, 544 S.W.3d 755, 770 (Tex. 2018). We review the grant or denial of a plea to the jurisdiction de novo, “determin[ing] whether the plaintiff’s pleadings, construed in favor of the plaintiff, allege sufficient facts affirmatively demonstrating the court’s jurisdiction to hear the case.” Hearts Bluff Game Ranch, Inc. v. State, 381 S.W.3d 468, 476 (Tex. 2012). “If evidence central to the jurisdictional issue is submitted, it should be considered in ruling on the plea to the jurisdiction” as “[e]vidence submitted with the plea may rebut the pleadings and undermine [the] waiver of immunity.” Id. The parties’ petitions present the following issues for our decision. In determining whether the Tort Claims Act waives TxDOT’s immunity from the negligence cause of action, the first issue we must consider is whether the Lyellco employees are TxDOT “employees” as defined by Section 101.001 of the Civil Practice and Remedies Code. The second issue is whetherotherTxDOT employees “operated” or “used” the motor-driven equipment under Section 101.021 by exercising direct and mandatory control over its use. Regarding the cause of action for inverse condemnation, the issue before us is whether there is some evidence that TxDOT acted with the required intent when it destroyed the Selfs’ trees. We address each issue in turn. I. The Tort Claims Act does not waive immunity from the Selfs’ negligence claim. As relevant here, the Tort Claims Act waives a governmental unit’s immunity from suit and makes it liable for “property damage . . . proximately caused by . . . the negligence of an employee acting within his scope of employment” if the damage “arises from the operation or use of . . . motor-driven equipment” and “the employee would be personally liable to the claimant according to Texas law.” TEX. CIV. PRAC. & REM. CODE § 101.021(1). The Selfs contend that their negligence claim falls within this waiver for two reasons. A. “Employee” under Section 101.001 First, the Selfs argue that the employees of subcontractor Lyellco who cut down the trees are TxDOT “employees” as defined by Section 101.001, and thus their use of motor-driven equipment to destroy the trees waives immunity under Section 101.021. Section 101.001(2) of the Texas Civil Practice and Remedies Code provides that “Employee” [1] means a person, including an officer or agent, who is in the paid service of a governmental unit by competent authority, but [2] does not include an independent contractor, an agent or employee of an independent contractor, or a person who performs tasks the details of which the governmental unit does not have the legal right to control. The court of appeals did not address the first part of the definition: whether the Lyellco employees were “in the paid service of a governmental unit by competent authority.” Rather, the court focused on the exclusions listed in the second part of the definition and evaluated whether the Lyellco employees fell into one of the excluded categories, concluding they did not. 683 S.W.3d at 81-88. In so doing, the court conducted the familiar employee-versus-independent-contractor analysis that Texas courts have long used in this and other legal contexts,[2] focusing on whether there was evidence that TxDOT exercised sufficient control over Lyellco’s employees performing the tree removal that they should be considered “employees” of TxDOT. See id. In a footnote, the court observed that TxDOT had argued on rehearing that Lyellco employees were paid by TFR and thus were not in the “paid service of a governmental unit.” The court concluded that TxDOT waived this argument by failing to raise it previously and thus declined to “address TxDOT’s eleventh-hour issue.” Id. at 86 n.7. This last conclusion was error for two reasons. First, TxDOT’s “paid service” argument was not waived and should have been considered. As subject-matter jurisdiction is never presumed and cannot be waived, see Tex. Ass’n of Bus. v. Tex. Air Control Bd., 852 S.W.2d 440, 443-44 (Tex. 1993), the issue can “‘be raised for the first time on appeal by the parties or by the court,’ [and] a court is obliged to ascertain that subject matter jurisdiction exists regardless of whether the parties have questioned it.” Univ. of Tex. Sw. Med. Ctr. at Dallas v. Loutzenhiser, 140 S.W.3d 351, 358-59 (Tex. 2004) (citations omitted). Thus, an appellate court’s review of a plea to the jurisdiction is not limited to the grounds set forth in the governmental unit’s plea in the trial court. Dallas Metrocare Servs. v. Juarez, 420 S.W.3d 39, 41 (Tex. 2013) (“[A]n appellate court must consider all of a defendant’s immunity arguments, whether the governmental entity raised other jurisdictional arguments in the trial court or none at all.”); see also Rusk State Hosp. v. Black, 392 S.W.3d 88, 95-96 (Tex. 2012). Second, the first part of Section 101.001(2) defines “employee” as a person in the “paid service of a governmental unit by competent authority,” and the second part then narrows that category with various exclusions. Under this definition, if the Lyellco employees were not in TxDOT’s “paid service,” they cannot qualify as TxDOT employees even if they do not fall within any of the exclusions. Accordingly, the court of appeals should have addressed the parties’ dispute regarding “paid service.” See Harris County v. Dillard, 883 S.W.2d 166, 167 (Tex. 1994) (holding because there was “no dispute that Skeen was not in the paid service of Harris County,” he was “therefore not an ‘employee’, within the meaning of the Tort Claims Act,” and not reaching the exclusions); Marino v. Lenoir, 526 S.W.3d 403, 406 (Tex. 2017) (concluding first that Lenoir was “in the paid service” of a governmental unit and then considering whether she was nonetheless excluded from the “employee” definition by one of the statute’s exceptions).[3] Turning to whether Lyellco’s employees were in TxDOT’s “paid service,” TxDOT argues in its briefing for a strict “paycheck test”: a requirement that the person receive payment directly from the governmental unit. At oral argument, however, TxDOT conceded that this test is too narrow. We agree and reject such a test. One problem with a paycheck test is that the statute does not include the phrase “paid by a governmental unit.” Instead, it uses the broader phrase “in the paid service of a governmental unit by competent authority.” That choice of words must be given meaning. Additionally, the definition specifically excludes independent contractors, agents or employees of independent contractors, and persons who perform tasks the details of which the governmental unit does not have the legal right to control. Many of these excluded persons and entities—for example, employees of independent contractors—would not receive a paycheck directly from the governmental unit. But excluding employees of independent contractors in the second part of the definition indicates that they could fall under the “paid service” umbrella of the first part. Thus, a strict “paycheck test” would render this exclusion superfluous. Relatedly, the phrase “by competent authority” indicates that some person or entity other than the governmental unit itself can be made competent—perhaps by law or contract—to place others in the unit’s paid service. This phrase must likewise be given meaning. A paycheck test would also be subject to easy manipulation in a manner inconsistent with the concepts that the Legislature concluded should be the focus of whether immunity is waived, including control and negligent operation of motor-driven equipment. See TEX.CIV.PRAC. &REM.CODE §§ 101.001(2), 101.021(1). As long as a governmental unit took steps to ensure its employees’ paychecks did not come directly from the unit, its immunity would not be waived under a paycheck test regardless of how much control it exerted over their negligent operation of motor-driven equipment. Moreover, the cases cited by TxDOT (1) stand for the limited proposition that “paid service” excludes unpaid volunteers, which we do not have in this case, and (2) themselves indicate that “paid service” has a broader meaning than the paycheck test. For example, in Harris County v. Dillard, we held that “a volunteer reserve deputy subject to being called into service” but who had not been called into service “was not in the paid service of Harris County at the time of the accident.” 883 S.W.2d at 167 (emphasis added). And in Murk v. Scheele, we similarly held that a University of Texas medical student was not in the paid service of the university because he received no monetary benefit other than medical professional liability insurance coverage. 120 S.W.3d 865, 867 (Tex. 2003). Murk indicates that the question is instead whether “part of” the person’s compensation “was ultimately paid by” the governmental unit, id. (emphasis added), which suggests that direct payment from the unit is not necessary to paid service. As discussed below, the facts of this case do not require us to define the precise contours of what constitutes “paid service.” But the parties identify some cases from our courts of appeals that have addressed whether particular people meet this statutory requirement when they do not receive payment directly from a governmental unit, as well as cases that have addressed similar issues in other legal contexts. For example, two courts have held that workers are in the paid service of a governmental unit when the unit pays them through a staffing company based on timesheets they submit or when a contractor bills the unit for reimbursement of the workers’ wages.[4] TxDOT also cites a case addressing when an entity is an employer under the Labor Code, arguing that this can be a helpful analogy in defining paid service under the Tort Claims Act. See Risk Mgmt. Strategies, Inc. v. Tex. Workforce Comm’n, 464 S.W.3d 864, 871-73 (Tex. App.—Austin 2015, pet. dism’d) (holding trusts employed caregivers when management company paid their wages and processed payroll but trusts reimbursed company for their services). In City of Bellaire v. Johnson, for instance, we concluded that a worker was “a person in the service of a political subdivision who has been employed as provided by law”[5] for purposes of workers’ compensation coverage because he “was paid by the City through [a staffing company], and on the basis of the hours he reported to the City.” 400 S.W.3d 922, 923 (Tex. 2013). We do not endorse using any of these approaches to define “paid service” in Section 101.001(2) today, as doing so is not necessary to decide this case. Instead, we simply note that these authorities may provide a useful starting point as courts and litigants attempt to develop a more comprehensive test for “paid service” in future cases. Here, the relationship between TxDOT and Lyellco’s employees is far removed from the kinds of payment relationships that courts have concluded amounts to paid service. TxDOTdid not have any contractual relationship with subcontractor Lyellco, which was retained by TxDOT’s contractor TFR. And the TFR–Lyellco subcontract specifies that “the payment provisions of the Contract between [TxDOT] and [TFR] are not a part of this Subcontract and specifically are not incorporated by reference.” Indeed, the record indicates that TxDOT paid TFR for tree removal services on a per-tree basis, while TFR paid Lyellco on a per-day basis.[6] There is no indication that these prices were substantially similar. Nor is there any indication that TxDOT made payments that “flowed through” TFR and Lyellco to Lyellco’s employees to compensate them for their work, that TxDOT had any role in determining how much Lyellco’s employees were paid, that any payment by TxDOT was a prerequisite to TFR’s obligation to pay Lyellco or Lyellco’s obligation to pay its employees, or that TxDOT had any reimbursement obligation regarding subcontracted labor. For these reasons, we hold that Lyellco’s employees were not in TxDOT’s paid service as required by Section 101.001(2), and therefore they are not TxDOT employees whose operation or use of motor-driven equipment could provide the basis for a waiver of immunity under Section 101.021 of the Tort Claims Act. Because we hold that Lyellco’s employees were not in TxDOT’s paid service, we do not address whether those employees fall within one of Section 101.001(2)’s exclusions. B. “Operation or use” under Section 101.021 The Selfs next argue that even if Lyellco’s employees who cut down the Selfs’ trees do not qualify as employees in TxDOT’s paid service, immunity is waived under Section 101.021 because other TxDOT employees “operated” or “used” the motor-driven equipment that cut down the trees by exercising direct and mandatory control over its use. In reviewing this argument, we apply our conclusion in LeLeaux v. Hamshire-Fannett Independent School District that a governmental unit’s employee must negligently operate or use the motor-driven equipment to satisfy this immunity waiver. 835 S.W.2d 49, 51 (Tex. 1992). As noted above, Section 101.021(1) requires that the property damage be caused by the employee’s negligence and that it arise from the operation or use of a motor-driven vehicle or piece of equipment. We explained in LeLeaux that the statutory phrase “arises from”, requires a nexus between the injury negligently caused by a governmental employee and the operation or use of a motor-driven vehicle or piece of equipment. While the statute does not specify whose operation or use is necessary—the employee’s, the person who suffers injury, or some third party—we think the more plausible reading is that the required operation or use is that of the employee. This requirement is consistent with the clear intent of the Act that the waiver of sovereign immunity be limited. Id. But as the court of appeals explained in detail, there is a split in authority on what it means for a government employee to “operat[e] or use” a motor-driven vehicle or equipment as required by LeLeaux. Must the employee be physically manning the controls of the vehicle or equipment to “operate” or “use” it, or is it sufficient if the employee exercises direct and mandatory control over the person at the controls? The courts of appeals in County of Galveston v. Morgan, 882 S.W.2d 485 (Tex. App.—Houston [14th Dist.] 1994, writ denied), and City of El Campo v. Rubio, 980 S.W.2d 943 (Tex. App.—Corpus Christi–Edinburg 1998, pet. dism’d w.o.j.), held that an employee of a governmental unit can exercise such direct and mandatory control over motor-driven vehicles or equipment that even though a private third party is physically at the controls, the employee is operating or using the equipment for purposes of Section 101.021. Morgan involved a dump truck that contacted an electrical line. The truck’s driver was an employee of a third-party contractor, but county employees acted as spotters who told the driver when to move forward and when to stop. Morgan reasoned: There is no requirement [in Section 101.021] that the vehicle in question be a county vehicle, only that a county employee “used” or “operated” the vehicle. [LeLeaux, 835 S.W.2d at 51.] “Operation” refers to “a doing or performing of a practical work,” and “use” means “to put or bring into action or service; to employ for or apply to a given purpose.” [Id.] The spotters in question were county employees. They were a necessary part of the job. The spotters told the truck driver when to move forward, how far to move, when to raise his bed, how far to raise it, when to lower his bed, and when to stop. The movement of the truck and the laying of the gypsum was within the spotters’ sole discretion. If a driver moved his truck contrary to the spotters’ direction, he could be fired. Although the spotters were not the drivers of the trucks, the spotters “used or operated” the trucks by exercising complete control over their “use or operation.” 882 S.W.2d at 490. Rubio involved a police officer who arrested a driver for driving with a suspended license. After arresting the driver, the officer ordered the driver’s wife, who was not a licensed driver, to follow him in the family’s vehicle to the police station and allegedly showed the wife how to operate the vehicle by demonstrating the use of the gas and brake pedals. When the officer pulled onto the highway, the wife followed him as ordered and was struck by another car. Noting that LeLeaux defined “use” broadly to include “bring into action” or “apply to a given purpose,” 980 S.W.2d at 946, the court concluded that the officer used the family’s vehicle. Id. at 947. Other courts of appeals have declined to follow Morgan and Rubio, requiring instead that the government employee physically use the motor-driven equipment to trigger Section 101.021′s waiver. See 683 S.W.3d at 75-80 (collecting cases). Because it is undisputed that employees of Lyellco—not TxDOT—physically used the motor-driven equipment to remove the trees, TxDOT would retain its immunity under this line of cases. This split may be a manifestation of the tension the Selfs identify between LeLeaux‘s reading of the statute and its plain text, which as LeLeaux acknowledges does not expressly require the employee to use or operate the motor-driven equipment. In PHI, Inc. v. Texas Juvenile Justice Department, we recently explained that it was error for the court of appeals to require that a government employee be actively operating the vehicle at the time of the incident, as “no court has the authority, under the guise of interpreting a statute, to engraft extra-statutory requirements not found in a statute’s text.” 593 S.W.3d at 305. As with the “paid service” issue, however, the facts of this case do not require us to resolve the split. As the court of appeals observed, the actions of TxDOT’s employees did not rise to the level of control that cases like Morgan and Rubio determined was sufficient to create a fact question on whether they were operating or using the equipment. See 683 S.W.3d at 80-81. The record indicates that TxDOT gave instructions to TFR to clear trees beside the highway within certain parameters. TFR then contracted with Lyellco to have Lyellco employees physically remove the trees. These facts are not analogous to Morgan and Rubio, which involved very precise direction by a government employee in close physical proximity to the equipment being operated. For these reasons, we hold that the Tort Claims Act does not waive immunity for the Selfs’ negligence claim. Accordingly, the court of appeals erred in affirming the trial court’s order denying TxDOT’s plea to the jurisdiction regarding that cause of action, which must be dismissed. II. Evidence that TxDOT destruction supports condemnation.
