Before Jones, Smith, and Dennis, Circuit Judges. Edith H. Jones, Circuit Judge: Steven Lawayne Nelson was convicted of capital murder and sentenced to death for his involvement in the robbery and murder of a pastor. After exhausting his state remedies, Nelson filed a federal habeas petition under 28 U.S.C. § 2254 and sought investigative services under 18 U.S.C. § 3599. The district court rejected his petition for relief, concluded that investigative services were not reasonably necessary, and denied a certificate of appealability (COA). Nelson then petitioned this court for a COA. We granted that petition on a single issue: Whether Nelson’s trial counsel were ineffective for failing to investigate and present at the penalty phase of trial two alleged accomplices’ participation in the robbery and murder. We hold that Nelson’s attempt to reframe his Sixth Amendment counsel ineffectiveness claim in federal court does not save it from the strictures of AEDPA review, 28 U.S.C. § 2254(d). We now AFFIRM. I. On March 3, 2011, while tending to his ecclesiastical duties at Arlington’s NorthPointe Baptist Church, Reverend Clinton Dobson was bound, savagely beaten, and then suffocated with a plastic bag. Nelson v. State, No. AP-76,924, 2015 WL 1757144, at *1 (Tex. Crim. App. Apr. 15, 2015). Dobson’s elderly secretary, Judy Elliott, was also beaten beyond recognition and within an inch of her life. Id. at *1–2. A car, laptop, cellphone, and several credit cards were stolen. Id. Two days later, police arrested Nelson and a grand jury indicted him for capital murder based on, inter alia, physical evidence recovered at the scene, surveillance video showing Nelson using the victims’ credit cards at a mall, and information provided by Nelson’s acquaintances. See id. at *2–3. At the guilt stage of Nelson’s trial, the State presented impressive physical and circumstantial evidence directly linking Nelson to the crime. Nelson’s fingerprints were at the murder scene, and droplets of the victims’ blood were on top of Nelson’s sneakers. Id. at *3. Moreover, distinctive white metal studs from the belt Nelson was wearing when police arrested him were found on and around Dobson’s body. Id. at *2–3. Shortly after the murder, Nelson was seen driving Elliott’s car to a store, where he sold Dobson’s laptop to another customer. Id. at *2. Video surveillance at the local mall showed Nelson using Elliott’s stolen credit cards to make purchases. Id. Further, the day after the murder, Nelson sent a series of incriminating text messages. “One asked to see the recipient because ‘[i]t might be the last time.’ Another said, ‘Say, I might need to come up there to stay. I did some [stuff] the other day, Cuz.’ A third said, ‘I [messed] up bad, Cuz, real bad.’” Id. Nelson even bragged about the murder to a friend. Against his lawyers’ advice, Nelson insisted on testifying. According to Nelson, he waited outside the church to serve as a lookout while two others, Anthony Springs and Claude Jefferson, went inside to rob Dobson and Elliott.[1] Id. at *3. After about twenty-five minutes, Nelson entered the church and saw the victims face down and bleeding out from their heads on the floor, but still alive. Id. Nelson did nothing to aid the victims; instead, he robbed them, taking Dobson’s laptop, Elliott’s keys, and Elliott’s credit cards, and then went back outside. Id. Later, he went back inside and saw that Dobson was dead, but quickly left because he could not stand the smell. Id. Nelson admitted that “he knew people were inside the church and that he agreed to rob them,” he just did not know that his accomplices would kill anyone. Id. Nelson’s story did not square with the State’s extensive evidence. For one, Nelson could not explain how droplets of the victims’ blood got on the top of his shoes or how pieces of his belt broke off at the murder scene. Moreover, Springs and Jefferson each had alibis. Two witnesses and phone records placed Springs over 30 miles away during the time of the murder. A class sign-in sheet and phone records placed Jefferson in his chemistry class. Rather than try to definitively prove Nelson’s story, Nelson’s trial counsel raised suspicion as to Springs’s and Jefferson’s involvement to undermine the State’s theory that Nelson alone committed the murder. For example, Nelson’s counsel challenged Springs’s and Jefferson’s alibis and established that police recovered DNA evidence from the crime scene that did not match the victims, Nelson, or Springs. The trial court gave the jury a law of the parties instruction, meaning that it could return a guilty verdict if it found either that Nelson was (1) directly responsible for Dobson’s murder or (2) a party to the robbery and should have anticipated that a death was likely to occur during the robbery. After deliberating, the jury found Nelson guilty of capital murder without specifying which theory it relied on. Then the court proceeded to the penalty stage. The penalty stage was held before the same jury that convicted Nelson. To sentence Nelson to death, the jury had to first find that Nelson (1) poses a “continuing threat to society” and (2) “actually caused the death of the deceased or did not actually cause the death of the deceased but intended to kill the deceased or another or anticipated that a human life would be taken.” Tex. Code Crim. Proc. art. 37.071, § 2(b)(1)–(2). If the jury answered those questions in the affirmative, then it had to consider whether mitigating circumstances warranted a “sentence of life imprisonment without parole rather than a death sentence.” Tex. Code Crim. Proc. art. 37.071, § 2(e)(1). At the penalty stage, the State continued to press the theme that Nelson alone murdered Dobson. Moreover, the State presented evidence that while awaiting trial Nelson murdered a fellow inmate, Johnathan Holden, vandalized jail property, smuggled weapons into the jail, and repeatedly assaulted jail personnel. See Nelson, 2015 WL 1757144 at *6–7. Nelson’s trial counsel challenged the evidence indicating that Nelson murdered Holden. They further argued that Nelson did not deserve the death penalty because others participated in the crime. To show that, Nelson’s DNA expert testified that the items used to restrain both victims contained DNA from unknown contributors. And another expert testified that hair found at the scene did not match the victims, Nelson, or Springs. Finally, Nelson’s trial counsel presented a comprehensive mitigation case by calling numerous witnesses to show that Nelson’s violent tendencies stemmed from mental illness and a difficult upbringing. Notwithstanding these efforts, the jury answered all three questions consistent with the death penalty and the district court sentenced Nelson to death. Nelson next sought state habeas relief. The State appointed John Stickels, an experienced and well-credentialed criminal attorney, to represent Nelson in his state habeas petition. Among other grounds for relief, Nelson alleged that his trial counsel rendered ineffective assistance under the Sixth Amendment at the penalty phase by failing “to adequately investigate and present mitigation evidence.” In particular, he asserted that his “defense team failed to investigate [his] background, history, family, and friends and, as a result, failed to discover relevant and important mitigation evidence.” On the basis of the record, the state habeas trial court recommended denying relief. It noted that Nelson’s trial counsel were “both highly experienced attorneys who were well-qualified to represent [him] at his capital-murder trial,” and that they “became fully versed in and knowledgeable of the information against[him] contained in the State’s file.” Furthermore, “[d]ue to the allegations of the indicted capital-murder case and the subsequent allegations of [Nelson's] severe misconduct while awaiting trial, [they] knew that most of their time would be spent trying to build a strong mitigation case.” Ultimately, the court concluded, Nelson’s trial counsel “made a well-reasoned strategic decision based on a thorough investigation, their professional judgment, the available witness testimony, and their reliance on well-qualified experts about how to best present [Nelson's sentencing] case to the jury.” The Texas Court of Criminal Appeals adopted the state habeas court’s findings and conclusions and also denied relief. See Ex parte Nelson, No. WR-82,814-01, 2015 WL 6689512, at *1 (Tex. Crim. App. Oct. 14, 2015). With new counsel, Nelson then filed the instant § 2254 application. Nelson again raised a single ineffective assistance of counsel claim related to his trial counsel’s alleged deficient performance at sentencing. In addition to the mitigation-related deficiencies identified in his state habeas application, Nelson asserted that his trial counsel deficiently failed to investigate, prepare, and litigate how Nelson’s culpability may be diminished by Springs’s and Jefferson’s participation. Nelson labels this his “participation claim.” Nelson maintained that his trial counsel’s aggregate failure to investigate mitigation, participation, and other sentencing related issues “deprived the jury of powerful information showing that [his] life should be spared.” He likewise argued prejudice—that, but for his trial counsel’s cumulative deficiency in failing to investigate the various sentencing-related issues, there is a reasonable probability that Nelson’s sentence would have been different. Finally, Nelson sought funding to further investigate his ineffective assistance claim. In a thorough and painstaking opinion, the district court rejected Nelson’s ineffective assistance claim. As a threshold matter, the district court held that Nelson did not procedurally default the ineffective assistance claim because he presented the same claim, albeit with fewer alleged instances of trial counsel’s deficient performance, in state court. Even if the mitigation and participation based claims were distinct and the participation based claim was therefore procedurally defaulted, the district court reasoned, Nelson did not overcome that procedural default under Martinez v. Ryan, 566 U.S. 1, 132 S. Ct. 1309 (2012), and Trevino v. Thaler, 569 U.S. 413, 133 S. Ct. 1911 (2013), by showing that his state habeas counsel provided ineffective assistance. In the alternative, the district court rejected Nelson’s participation claim on the merits. The district court also denied Nelson’s request for investigative services under 18 U.S.C. § 3599(f). Finally, the district court refused a COA. Nelson then petitioned this court for a COA. This court granted Nelson’s request in part. Nelson v. Davis, 952 F.3d 651, 670–76 (5th Cir. 2020). Noting that “reasonable jurists could debate whether Nelson’s [participation] allegations ‘fundamentally alter’ his [ineffective assistance] claim,” this court hypothesized that Nelson’s participation based ineffective assistance claim may be distinct from the ineffective assistance claim raised in state court and therefore procedurally defaulted. Id. at 671–72. Next, this court concluded that reasonable jurists could debate whether Nelson’s state habeas counsel was ineffective in failing to raise the participation claim and that, as a result, reasonable jurists could debate whether Nelson could overcome procedural default under Martinez/Trevino. As to the merits of Nelson’s participation claim, this court reasoned that “[b]ecause Nelson’s counsel sought to convince the jury that Springs and Jefferson were involved but arguably failed to take reasonable investigative steps in developing evidence in support of this argument, . . . reasonable jurists could debate that his trial counsel’s performance” was deficient. Id. at 675. This court carried with the development of the participation claim the questions of Strickland prejudice and denial of funding. Id. at 675–76. II. In an appeal from a district court order denying habeas relief, “this court reviews the district court’s findings of fact for clear error and its conclusions of law de novo, applying the same standards to the state court’s decision as did the district court.” Harrison v. Quarterman, 496 F.3d 419, 423 (5th Cir. 2007) (citing Coble v. Dretke, 444 F.3d 345, 349 (5th Cir. 2006)). The court “may affirm on any ground supported by the record,” Dorsey v. Stephens, 720 F.3d 309, 314 (5th Cir. 2013) (citing Fisher v. Texas, 169 F.3d 295, 299 (5th Cir. 1999)), and is not bound “by the COA opinion’s observations on the merits,” Trevino v. Davis, 861 F.3d 545, 548 n.1 (5th Cir. 2017). We review the denial of funding under 18 U.S.C. § 3599 for an abuse of discretion. Ayestas v. Davis, 138 S. Ct. 1080, 1094 (2018). III. The Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”) governs federal habeas proceedings. Out of respect to “our system of dual sovereignty,” AEDPA greatly restricts the availability of federal habeas relief to those convicted of crimes in state court. Shinn v. Ramirez, 142 S. Ct. 1718, 1730 (2022) (citing Printz v. United States, 521 U.S. 898, 918, 117 S. Ct. 2365, 2376 (1997) and Brown v. Davenport, 142 S. Ct. 1510, 1523–24 (2022)). Chief among AEDPA’s federalism preserving features is the requirement that state prisoners “exhaus[t] the remedies available in the courts of the State” before seeking federal habeas relief. 28 U.S.C. § 2254(b)(1). Generally, state prisoners satisfy “this exhaustion requirement by raising [their] federal claim before the state courts in accordance with state procedures.” Shinn, 142 S. Ct. at 1732 (citing O’Sullivan v. Boerckel, 526 U.S. 838, 848, 119 S. Ct. 1728, 1734 (1999)). A federal court’s review of a claim adjudicated in state court is circumscribed in two ways. First, the federal court may not consider any evidence beyond the state court record. See Cullen v. Pinholster, 563 U.S. 170, 180–81, 131 S. Ct. 1388, 1398 (2011). Second, the state prisoner must show that the state court’s decision “was contrary to, or involved an unreasonable application of,” law clearly established by the Supreme Court, 28 U.S.C. § 2254(d)(1), or that the decision “was based on an unreasonable determination of the facts” in light of the state court record, 28 U.S.C. § 2254(d)(2). In contrast, if a state prisoner fails to present his federal claim in state court for adjudication or comply with state procedures, and thereby procedurally defaults the claim, then a federal court will, in all but the most extraordinary cases, decline to review it. Shinn, 142 S. Ct. at 1732. A prisoner may overcome such procedural default only “if he can show ’cause’ to excuse his failure to comply with the state procedural rule and ‘actual prejudice resulting from the alleged constitutional violation.’” Davila v. Davis, 137 S. Ct. 2058, 2064–65 (2017) (quoting Wainwright v. Sykes, 433 U.S. 72, 84, 97 S. Ct. 2497, 2505 (1977) and Coleman v. Thompson, 501 U.S. 722, 750, 111 S. Ct. 2546, 2565 (1991)). Ordinarily, “[a]ttorney ignorance or inadvertence” does not excuse procedural default “because the attorney is the [prisoner's] agent when acting, or failing to act, in furtherance of the litigation, and the [prisoner] must ‘bear the risk of attorney error.’” Coleman, 501 U.S. at 753, 111 S. Ct. at 2566–67 (quoting Murray v. Carrier, 477 U.S. 478, 488, 106 S. Ct. 2639, 2645 (1986)). A narrow exception to the general rules stated in Shinn exists under Martinez and Trevino. That exception allows—but does not require—a federal habeas court to excuse a state prisoner’s procedural default of a “substantial” ineffective assistance of trial counsel claim where (1) state law forbids raising that claim on direct review or makes it virtually impossible to do so and (2) the prisoner can show his state habeas counsel rendered constitutionally deficient assistance by failing to raise the claim. Trevino, 569 U.S. at 423–24, 133 S. Ct. at 1918. In the rare case where a state prisoner successfully overcomes procedural default, the federal habeas court then considers the claimed ineffectiveness of trial counsel de novo. Hoffman v. Cain, 752 F.3d 430, 437 (5th Cir. 2014) (citing Wright v. Quarterman, 470 F.3d 581, 591 (5th Cir. 2006)). Critically, however, the federal habeas court’s review is limited to the state court record. Shinn, 142 S. Ct. at 1734 (holding that “under § 2254(e)(2), a federal habeas court [reviewing a procedurally defaulted claim] may not conduct an evidentiary hearing or otherwise consider evidence beyond the state-court record based on ineffective assistance of state postconviction counsel”). Under AEDPA’s framework, then, two procedural issues logically precede the merits of Nelson’s participation claim. First, whether that claim was “adjudicated on the merits in State court proceedings” and therefore subject to the strictures of 28 U.S.C. § 2254(d). Second, if not, whether Nelson can overcome the consequent procedural default. We address the first question and hold that Nelson’s participation claim was adjudicated on the merits in state court proceedings. We pretermit the second question and “cut straight to the merits to deny his claim” in the alternative. Murphy v. Davis, 901 F.3d 578, 589 n.4 (5th Cir. 2018).[2] A. The limitations on federal habeas review contained in § 2254(d) apply to any claim “adjudicated on the merits in State court proceedings.” A “claim” for AEDPA purposes is “an asserted federal basis for relief from a state court’s judgment of conviction.” Gonzales v. Crosby, 545 U.S. 524, 530, 125 S. Ct. 2641, 2647 (2005) (defining “claim” as used in § 2244(b)); see also Powerex Corp. v. Reliant Energy Servs., Inc., 551 U.S. 224, 232, 127 S. Ct. 2411, 2417 (2007) (“[I]dentical words and phrases within the same statute should normally be given the same meaning.”). Generally, determining whether the § 2254(d) relitigation bar applies is straightforward. On the one hand, the relitigation bar does not apply where the prisoner fails altogether to present a certain claim in state court. See Trevino, 569 U.S. at 416–17, 133 S. Ct. at 1914–15 (applying procedural default regime rather than § 2254(d) limitations in case where state habeas counsel failed to raise ineffective assistance claim that federal habeas counsel later raised). On the other hand, the§ 2254(d) limitations do apply in cases where a prisoner “fairly presented the substance of his [federal] claim to the state courts.” Nobles v. Johnson, 127 F.3d 409, 420 (5th Cir. 1997) (citing Picard v. Connor, 404 U.S. 270, 275– 76, 92 S. Ct. 509, 512–13 (1971)). In cases where the support for the prisoner’s federal claim evolves across the state and federal proceedings, determining whether § 2254(d)’s relitigation bar applies is more difficult. A court must then consider whether the evolved claim presented in federal court is in fact a new claim altogether, and thus excluded from § 2254(d)’s relitigation bar, or simply the old one already adjudicated in state court, in which case § 2254(d)’s relitigation bar does apply. See Pinholster, 563 U.S. at 186 n.10, 131 S. Ct. at 1401 n.10. To date, the Supreme Court has not identified “where to draw the line between new claims and claims adjudicated on the merits.” Id. Relying on this court’s COA opinion, Nelson posits that when a claim raised in a federal habeas petition fundamentally alters a claim raised in the state habeas petition, it is not “adjudicated on the merits in State court proceedings” and is therefore not subject to § 2254(d)’s restrictions. Nelson, 952 F.3d at 671–72. A claim raised in a federal habeas petition fundamentally alters the related claim raised in a state habeas petition, Nelson opines, where the claim presented to the federal court includes new, material factual allegations that place “the claim in a ‘significantly different legal posture.’” Nelson principally relies on Lewis v. Thaler, 701 F.3d 783 (5th Cir. 2012), and Escamilla v. Stephens, 749 F.3d 380 (5th Cir. 2014), for support. Carefully read, however, neither of those cases supports Nelson’s proposed standard for separating new ineffectiveness claims from those adjudicated on the merits. Lewis addressed whether a federal habeas court could consider expert mitigation evidence offered for the first time in the federal proceedings. 701 F.3d at 789. Reasoning in light of Pinholster, this court eschewed a prior line of cases holding that facts and evidence raised “for the first time on federal habeas review” should be “‘analyzed under the exhaustion rubric of § 2254(b),’ rather than as an issue of ‘factual development’ under § 2254(d) and (e).” Id. at 789 (quoting Dowthitt v. Johnson, 230 F.3d 733, 745 (5th Cir. 2000)); see also Clark v. Thaler, 673 F.3d 410, 416–17 (5th Cir. 2012). Accordingly, the court held that it could not consider the new mitigation evidence. Lewis, 701 F3d. at 791. But the court did not hold that a state prisoner could avoid § 2254(d)’s limitations by presenting new evidence that fundamentally altered a claim already adjudicated in state court proceedings. Escamilla is similarly unhelpful. That case held that where a prisoner’s state habeas counsel raised a particular federal claim in state habeas proceedings, albeit ineffectively under the Sixth Amendment, Pinholster barred the prisoner from presenting new evidence in federal proceedings because the original claim was adjudicated on the merits in state court proceedings. Escamilla, 749 F.3d at 394–95. This court did not outline a loophole around § 2254(d)’s limitations whenever newly offered evidence and legal theories “fundamentally alter” a claim previously presented to the state courts. Indeed, Escamilla cautioned that “once a claim is considered and denied on the merits by the state habeas court, Martinez [v. Ryan] is inapplicable, and may not function as an exception to Pinholster‘s rule that bars a federal habeas court from considering evidence not presented to the state habeas court.” Escamilla, 749 F.3d at 395 (internal citation omitted); see also Broadnax v. Lumpkin, 987 F.3d 400, 408–09 (5th Cir. 2021). Even so, Nelson’s argument, that the new participation aspect of the ineffective assistance claim fundamentally alters the ineffective assistance claim he litigated in the state proceedings, would fail. On this point, a careful comparison of his state and federal habeas applications is useful. In his state habeas petition, Nelson raised a single ineffective assistance of counsel claim, as he contended that his trial counsel “fail[ed] to adequately investigate and present mitigation evidence.” More specifically, Nelson asserted that trial counsel “failed to investigate [his] background, history, family, and friends and, as a result, failed to discover relevant and important mitigation evidence that would have made a difference” at the penalty stage. After reviewing Nelson’s habeas application, the State’s reply, “all of the exhibits and materials filed by each party, and the entire record of the trial and habeas proceedings,” the state habeas court concluded that Nelson’s trial counsel made a “well-reasoned” and informed strategic decision to focus on building a strong mitigation case. Nelson’s federal habeas application likewise raised a single ineffective assistance of counsel claim. He argued that his trial counsel “failed to adequately investigate, prepare, and litigate sentencing.” Trial counsel’s performance was constitutionally deficient, Nelson asserted, because they failed to: (1) investigate, prepare, and litigate how Nelson’s culpability may be diminished by Springs’s and Jefferson’s participation; (2) develop evidence that Holden died of suicide rather than at the hands of Nelson; and (3) investigate and present evidence about Nelson’s background and mental health. As to Strickland‘s prejudice prong, Nelson argued that the cumulative effect of trial counsel’s deficient performance prejudiced him on all three special issues at sentencing. In both the state and federal habeas proceedings, Nelson raised a single ineffective assistance of counsel claim related to trial counsel’s performance at sentencing. Nelson concedes, as he must, that both claims are “similar.” There is no dispute that the “asserted federal basis for relief from [the] state court’s judgment of conviction” is the same. Crosby, 545 U.S. at 530, 125 S. Ct. at 2647. The only difference between the claim adjudicated in the state court and the claim presented in federal court is that Nelson pointed out more instances of trial counsel’s alleged deficient performance at sentencing in the federal court claim. That is not enough to fundamentally alter the ineffective assistance claim adjudicated in the state court to place the claim in a significantly different legal posture. A state prisoner cannot aggregate alleged instances of ineffective assistance of counsel to satisfy the Strickland deficient performance and prejudice requirements and then disaggregate those theories to create new, unadjudicated claims and thereby circumvent § 2254(d)’s limitations. Nelson resists this conclusion by arguing that it will produce absurd results. He hypothesizes that, by the same logic, a Brady claim alleging that the prosecution suppressed exculpatory forensic evidence would be “adjudicated on the merits” if in state court the prisoner raised a Brady claim alleging that the prosecution suppressed favorable eyewitness testimony. But Nelson confounds the distinct natures of Strickland and Brady claims. Conceptually, a Brady claim is specific to particular pieces of material evidence allegedly suppressed by the prosecution whereas a Strickland claim is specific to a particular stage of a proceeding. Compare United States v. Brown, 650 F.3d 581, 588–93 (5th Cir. 2011) (evaluating Brady claims on an item-by-item basis), with Blanton v. Quarterman, 543 F.3d 230, 236–48 (5th Cir. 2008) (evaluating separately state prisoner’s trial counsel and appellate counsel ineffective assistance claims). Thus, this court’s analysis does not produce absurd results, just the results required by § 2254(d). Because we conclude that Nelson’s ineffective assistance claim was “adjudicated on the merits in State court proceedings,” this court’s review is constrained by the limitations articulated in Pinholster and § 2254(d). Nelson does not argue that he can overcome those limitations. On this basis, the state courts’ rejection of Nelson’s ineffectiveness claim did not unreasonably apply Strickland, nor was it an unreasonable application of the law to the facts. Nelson is not entitled to relief. B. Even if Nelson’s participation claim were not subject to § 2254(d)’s relitigation bar and assuming, arguendo, he could overcome procedural default by showing ineffective assistance of state habeas counsel, he would not succeed on the merits of his ineffective assistance claim.[3] In evaluating Nelson’s ineffective assistance claim, this court’s review is limited to the record before the state court. Shinn, 142 S. Ct. at 1734. To prevail on his trial counsel ineffective assistance claim, Nelson must show (1) deficient performance that (2) resulted in prejudice at sentencing. Strickland v. Washington, 466 U.S. 668, 687, 104 S Ct. 2052, 2064 (1984). We address only the prejudice component and conclude that Nelson has not met his burden.[4] To establish prejudice, the “defendant must show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694. Critically, the “likelihood of a different result must be substantial, not just conceivable.” Harrington v. Richter, 562 U.S. 86, 112, 131 S. Ct. 770, 792 (2011) (citing Strickland, 466 U.S. at 693, 104 S. Ct. at 2067–68). Where, as here, the “Strickland claim is based on an allegedly deficient sentencing investigation, the petitioner may establish prejudice by showing that ‘the totality of the available mitigation evidence . . . reweigh[ed] . . . against the evidence in aggravation’ creates ‘a reasonable probability that at least one juror would have struck a different balance’ and recommended a life sentence instead of death.” Busby v. Davis, 925 F.3d 699, 723–24 (5th Cir. 2019) (first quoting Sears v. Upton, 561 U.S. 945, 955–56, 130 S. Ct. 3259, 3266 (2010) (per curiam), and then Wiggins v. Smith, 539 U.S. 510, 537, 123 S. Ct. 2527, 2543 (2003)). Nelson argues that his trial counsel’s failure to investigate and introduce evidence about Springs’s and Jefferson’s potential involvement in Dobson’s murder prejudiced him at sentencing. Specifically, had trial counsel investigated and presented evidence about their involvement, “at least one juror likely would have found that either man (or both) participated in” Dobson’s murder. And the state court record, he contends, is replete with evidence that supports that conclusion. For example, Nelson points to (1) grand jury and trial testimony from one witness that contradicts the timelines that Springs and Jefferson offered as alibis, and (2) testimony from another witness that Springs’s SIM card was in that witness’s phone on the day of the murder. Moreover, the state court record contains extensive evidence showing that Springs and Jefferson retained proceeds of the robbery. With a more fulsome picture of Springs’s and Jefferson’s involvement and his own correspondingly minimal role, Nelson concludes, a juror could have concluded that his “participation or intent fell short of the standards set by the anti-parties issue, that his culpability warranted a favorable answer to the mitigation instruction, or that he would not represent a continuing threat to society.” No doubt proving that Springs or Jefferson also participated in Dobson’s ghastly murder is relevant to the three special questions posed to the jury at sentencing. But that is not enough to show Strickland prejudice. Even if Nelson’s trial counsel had further investigated Springs’s and Jefferson’s alibis and presented evidence about their involvement, the State’s case for death on each special question would have remained unassailable. We consider each special question in turn. First, the anti-parties question. In answering this question, the jury had to consider whether Nelson “actually caused” Dobson’s death or “anticipated that a human life would be taken.” Tex. Code Crim. Proc. art. 37.071, § 2(b)(2). At trial, the State adduced a mountain of uncontroverted evidence that strongly suggested Nelson’s direct participation in Dobson’s murder. Nelson’s fingerprints were found on the wrist rest of Dobson’s desk. Distinctive studs broke off Nelson’s belt at the crime scene, indicating a struggle. Drops of the victims’ blood were found on top of Nelson’s shoes, and those shoes matched a bloody print left at the scene. Nelson alone used Elliott’s credit card in the ensuing days to make purchases, and he alone sold Dobson’s laptop. By contrast, no physical evidence linked Springs or Jefferson with Dobson’s murder. In light of this evidence, it is unlikely that evidence of Springs’s and Jefferson’s involvement would have made any difference in how the jury answered the anti-parties question. More fundamentally, Nelson’s own testimony severely compromised any chance for trial counsel to persuade the jury to spare Nelson’s life on the anti-parties front. Nelson claimed that he acted as a lookout for Springs and Jefferson. When he entered the church, he saw Elliott and Dobson bleeding out on the floor—but still alive—and did nothing to assist them. Instead, he stole Dobson’s computer, Elliott’s credit cards, and her car keys and went back outside, leaving the victims defenseless with his alleged accomplices. Proving that Nelson was an accomplice, and not the primary perpetrator, of the capital murder would do nothing to falsify that he “anticipated that a human life would be taken.” Tex. Code Crim. Proc. art. 37.071, § 2(b)(2). That the same jury had just convicted Nelson of capital murder means that the jury either concluded that Nelson was directly responsible for Dobson’s murder or else that he was an accomplice to the robbery and that he should have anticipated that a death was likely to occur during the course of the robbery. Thus, even if the jury accepted Nelson’s testimony at face value, there is little reason to think any juror would have answered the anti-parties question differently, much less a substantial likelihood that any juror would have done so. Next, the future dangerousness and mitigation questions required the jury to assess whether Nelson poses a “continuing threat to society” and whether other mitigating circumstances warrant a “sentence of life imprisonment without parole rather than a death sentence.” Tex. Code Crim. Proc. art. 37.071, §§ 2(b)(1), (e)(1). According to Nelson’s own recounting of the events, he participated in the aggravated robbery of a church during which that church’s ecclesiastical leader was brutally and senselessly murdered. While in custody and awaiting trial for Dobson’s murder, Nelson murdered a fellow inmate, engaged in several altercations with jail officers, repeatedly vandalized jail property, and smuggled weapons into jail. Nelson, 2015 WL 1757144, at *6–7. And after murdering his fellow inmate, who suffered from intellectual disabilities, Nelson “did a ‘celebration dance’ in the style of Chuck Berry, ‘where he hops on one foot and plays the guitar.’” Id. at *6. Further, Nelson’s own forensic psychologist “agreed that characteristics of antisocial personality disorder describe him” and that he “has many characteristics of a psychopath.” Id. at *8. Even if Nelson’s trial counsel could definitively establish Springs’s or Jefferson’s involvement, they had little hope of showing that Nelson did not pose a continuing threat to society or that other mitigating evidence warranted life imprisonment rather than a death sentence. Accordingly, Nelson cannot show a substantial likelihood that a juror would have answered the future-dangerousness or mitigation questions differently had his trial counsel investigated and presented evidence of Nelson’s lessened participation. For all these reasons, Nelson cannot demonstrate a reasonable probability that at least one juror would have recommended a life sentence had his trial counsel investigated Springs’s and Jefferson’s involvement and presented evidence about the same at sentencing. He was not prejudiced, and his ineffective assistance of counsel claim would fail even if it were not assessed under the rigorous standards of AEDPA § 2254(d). AFFIRMED. James L. Dennis, Circuit Judge, dissenting: I respectfully dissent from the majority’s opinion affirming the district court’s denial of Nelson’s federal habeas petition. We previously granted Nelson a COA on his unexhausted claim that his trial counsel was ineffective for failing to investigate whether two of Nelson’s friends, Anthony Springs and Claude Jefferson, committed the murder for which Nelson was convicted (the “IATC-Participation claim”). Instead of resolving the merits of Nelson’s petition, we should reverse the district court’s order and remand with instructions to grant Nelson’s request for investigative funding to further develop his IATC-Participation claim and for a stay so that Nelson may exhaust this claim in state court. I. Investigative Funding Section 3599 “authorizes federal courts to provide funding to a party who is facing the prospect of a death sentence and is ‘financially unable to obtain adequate representation or investigative, expert, or other reasonably necessary services.’” Ayestas v. Davis, 138 S. Ct. 1080, 1092 (2018) (citing 18 U.S.C. § 3599). In evaluating funding requests, courts consider whether funding is “reasonably necessary” in light of the potential merit of the applicant’s claims, the likelihood that the services would render useful evidence, and the prospect that the applicant could overcome any procedural hurdles. Id. at 1093–94 (proposed services must be “reasonably necessary” for the applicant’s representation.”). Courts of appeals review district court funding decisions for abuse of discretion. Ayestas v. Davis (Ayestas II), 933 F.3d 384, 388 (5th Cir. 2019). Nelson sought funding under Section 3599(f) to pursue evidence supporting his theory that Springs and Jefferson were primarily responsible for the murder. The district court denied Nelson’s request, concluding— under the “substantial need” standard later rejected by the Supreme Court in Ayestas—that based on the evidence presented to the jury, Nelson committed the crime alone so no evidence of another’s participation exists. 138 S. Ct. at 1092 (adopting “reasonably necessary” standard for funding requests brought under § 3599). But in relying on the existing evidence on the record, the district court failed to consider “the potential merit of the claims” and “the likelihood that the services will generate useful and admissible evidence.” Id. at 1094. Nelson sought to conduct the requested investigation precisely to locate evidence that he alleges exists and could have been uncovered to support Nelson’s principal theory of defense and convince the jury to spare Nelson’s life at the sentencing phase. The district court thus abused its discretion in denying Nelson’s request for funding because Nelson has demonstrated that further investigation is likely to reveal evidence that supports his substantial IATC-Participation claim. Nelson’s IATC-Participation claim is likely meritorious. To prevail on this claim, Nelson must demonstrate both deficient performance by his trial counsel and prejudice to the outcome of his case. Strickland v. Washington, 466 U.S. 668, 687 (1984). Nelson can probably show that his trial counsel was deficient in failing to reasonably develop the principal defense theory in Nelson’s case—that Springs and Jefferson carried out the murder while Nelson served as a lookout for what he believed to be a robbery. Despite the importance of evidence suggesting Springs’ and Jefferson’s involvement in the murder, trial counsel did not even attempt to contact either Springs or Jefferson, let alone otherwise independently verify their alibis. At sentencing, trial counsel only used the fact that DNA from an unknown person was at the scene of the crime to support this defense theory. Moreover, on appeal Nelson identified several “red flags” that would have prompted a reasonable attorney to conduct further investigation to gather evidence of Springs’ and Jefferson’s involvement in the murder. Rompilla v. Beard, 545 U.S. 374, 391 (2005) (Counsel “could not reasonably have ignored mitigation evidence or red flags simply because they were unexpected.”). As to Springs, for example, Nelson points out that trial counsel should have investigated leads suggesting that Springs obtained Dobson’s property directly from the scene of the crime, Springs’ alibi witnesses had motive to protect him from jail, someone else had Springs’ SIM card in their phone on the day of the murder and bruising on Springs’ knuckles at the time Springs was arrested was consistent with the struggle with the victims. Nelson also highlighted the weaknesses in Jefferson’s alibi given that while Jefferson testified that he was taking a quiz in class at the time of the murder, the teacher of that class stated that there was no quiz that day, Jefferson often skipped class, and a classmate could have signed in for Jefferson that day. These “red flags” indicate that trial counsel’s investigation was likely deficient, and that further investigation will generate useful and admissible[5] evidence in support of Nelson’s IATC-Participation claim. Nelson can also likely show, with the aid of further factual development, that he was prejudiced by trial counsel’s failure to investigate because there is a reasonable probability that the jury would not have sentenced Nelson to death if trial counsel had gathered and presented the jury with more evidence of Springs’ and Jefferson’s participation in the murder. In sentencing Nelson to death, the jury necessarily concluded that he “actually caused death or anticipated that death would occur,” see TEX. CODE CRIM PROC. ART. 37.071 § 2(b)(2), and further evidence that Springs and/or Jefferson committed the murder would have cast doubt on whether Nelson’s culpability for the murder warranted the death penalty. Yet, as we noted in granting Nelson’s COA, in the absence of the undiscovered evidence, the court finds itself in “something of a Catch-22″ because “[w]e cannot determine whether Nelson was prejudiced without knowing what evidence could have been uncovered” in the absence of further investigation and therefore “should not make this [prejudice] determination based solely on the record before us when he may be entitled to investigative funding to support this claim.” Nelson, 952 F.3d at 675. Based on the deficiencies in trial counsel’s performance, and the various avenues for investigation identified by Nelson, “[t]here is[] good reason to believe that, were [Nelson's] § 3599(f) motion granted, he could establish prejudice under Strickland.” Ayestas, 138 S. Ct. at 1100 (Sotomayor, J., concurring). Further factual development is likely to lead to useful and admissible evidence to support Nelson’s substantial IATC-Participation, and as such, is reasonably necessary for Nelson to be adequately represented by his present counsel. We should reverse the district court’s denial of Nelson’s petition and remand with instructions to grant Nelson’s request for investigative funding under Section 3599(f). II. Rhines Stay An order staying a federal habeas proceeding and holding it in abeyance pending a return to state court is appropriate when a petitioner brings an unexhausted claim in federal court and: “(1) the district court determines there was good cause for the petitioner’s failure to exhaust his claims in state court; (2) the claim is not plainly meritless; and (3) there is no indication that the petitioner is engaging in abusive litigation tactics or intentional delay.” Young v. Stephens, 795 F.3d 484, 495 (5th Cir. 2015) (citing Rhines v. Weber, 544 U.S. 269, 277–78 (2005)). The district court did not address that standard and instead summarily denied Nelson’s motion in light of its denial of Nelson’s habeas petition on the merits. However, Nelson meets the standard for a stay under Rhines, and the district court abused its discretion in failing to issue a stay to allow Nelson to exhaust his IATC-Participation claim in state court. Rhines, 544 U.S. at 270 (“[I]t likely would be an abuse of discretion for a district court to deny a stay . . . if the petitioner had good cause for his failure to exhaust, his unexhausted claims are potentially meritorious, and there is no indication that he engaged in intentionally dilatory litigation tactics.”). As an initial matter, I disagree with the majority’s conclusion that the IATC-Participation claim was exhausted in state court. In granting Nelson’s COA on this claim, we found that Nelson’s state habeas IATC claim “did not touch on Nelson’s allegations in this IATC-Participation claim that undiscovered evidence indicating that he played a minimal role in the capital murder itself could have been presented to the jury.” Nelson, 952 F.3d at 671– 72. Nelson’s IATC-Participation claim thus “fundamentally alters” his state court IATC claim, which only challenged whether his trial counsel sufficiently investigated his “background, history family, and friends.” See also 2 Randy Hertz & James S. Liebman, Federal Habeas Corpus Practice & Procedure § 23.3c, at 982 (4th ed. 2001) (“The controlling standard seems to be that the petitioner exhausts the factual basis of the claim as long as she did not [] ‘fundamentally alter the legal claim already considered by the state courts[.]‘”). The majority faults Nelson’s IATC-Participation claim for also being an IATC claim yet fails to meaningfully grapple with the case-specific differences in these two claims and the simple fact that his state habeas counsel did not raise and the state court did not adjudicate any claim based on the allegation that trial counsel failed to gather evidence of Nelson’s diminished culpability for the murder. Moore v. Quarterman, 533 F.3d 338, 341 (5th Cir. 2008) (“Determining whether a petitioner exhausted his claim in state court is a case- and fact-specific inquiry.”). Nelson’s IATC-Participation claim is thus unexhausted because it was not adjudicated in state court. Nelson meets the requirements for a Rhines stay to allow him to exhaust his IATC-Participation claim in state court. The first requirement— “good cause”—is satisfied when state habeas counsel is deficient in failing to raise a claim in state habeas proceedings. Ruiz v. Quarterman, 504 F.3d 523, 529 n.17 (5th Cir. 2007) (recognizing that the “failures” of Texas’ state habeas system in affording competent state habeas representation establishes equitable good cause for a Rhines stay). Here, Nelson’s state habeas counsel’s representation was deficient in failing to raise the IATC-Participation claim during state habeas proceedings. The decision to sentence Nelson to death was predicated in part on whether Nelson intended to cause death or anticipated loss of life, see TEX.CODE CRIM PROC.ART. 37.071 § 2(b)(2), and as such, any reasonably competent habeas attorney would have appreciated the importance[6] of raising the IATC-Participation claim during the state habeas proceeding. Trevino v. Davis, 829 F.3d 328, 348–49 (5th Cir. 2016) (state post-conviction counsel’s failure to investigate an IATC claim is deficient performance where the “[t]he deficiency in [trial counsel's] investigation would have been evident to any reasonably competent habeas attorney.”). Yet not only did state habeas counsel do nothing to investigate the IATC-Participation claim, but he also spent only 4.5 hours reviewing trial counsel’s records. Here, Nelson’s state habeas counsel was thus deficient and such deficiency prejudiced Nelson since his underlying IATC-Participation claim is substantial. The second Rhines requirement—that the underlying claim presented is not “plainly meritless”—is also satisfied. As explained above, Nelson’s IATC-Participation likely has merit, especially in light of the potential evidence he might uncover if allowed to conduct further investigation into Springs’ and Jefferson’s role in the murder. Finally, there is no sign of “intentionally dilatory litigation tactics” on Nelson’s part that might justify a district court’s denial of a Rhines stay. See Rhines, 544 U.S. at 277–78. Nelson discovered the underlying bases for his IATC-Participation Claim during federal habeas counsel’s investigation, and he filed his federal petition shortly thereafter. Because Nelson’s IATC-Participation claim is unexhausted and Nelson has met the requirements set forth in Rhines, a stay is appropriate. We should reverse the district court’s denial of Nelson’s request for a Rhines stay and remand with instructions that the district court stay this proceeding to allow Nelson to exhaust his IATC-Participation claim in state court. III. Because the district court should have granted Nelson funding to further develop his IATC-Participation claim and a Rhines stay to exhaust the claim in state court, I respectfully dissent from the majority’s decision to affirm the denial of Nelson’s habeas petition based on the incomplete record before it. We should instead reverse the district court’s denial of Nelson’s requests for funding and a stay, and remand with instructions that the district court grant Nelson’s request for investigative services and stay this proceeding while Nelson returns to state court to exhaust the IATC-Participation claim.
