Advocate General Andrea Biondi on 15 October 2026 advised the Court of Justice of the European Union to dismiss ByteDance's appeal against TikTok's designation as a gatekeeper under the Digital Markets Act, in the first case under the law to reach the EU's highest court.

In Short

The EU named TikTok's owner a "gatekeeper" in 2023, which means it has to follow a strict set of rules for very large platforms, and TikTok's owner has been fighting that label in court. An independent adviser to the EU's top court has now said the fight should fail, although the judges have not decided yet and do not have to agree. If they do, it will become much harder for any big platform to escape these rules by arguing that the numbers don't tell the whole story.

What the opinion proposes

The opinion, delivered in Case C-627/24 P, Bytedance v Commission, concerns a judgment of the General Court of 17 July 2024 in Case T-1077/23. That judgment upheld the European Commission's decision of 5 September 2023, which designated ByteDance as a gatekeeper in respect of TikTok, classified as an online social networking service. ByteDance appealed to the Court of Justice.

Biondi did not examine the whole appeal. According to the opinion, "at the Court's invitation, the present Opinion will focus exclusively on the first and third parts of the first ground of appeal." He also addressed ByteDance's second ground, on the right to be heard, and a request from the Commission to replace part of the General Court's reasoning. The other three parts of ByteDance's first ground, which according to the opinion is "composed of five parts," are not covered.

The conclusion, at point 186, is short. Biondi proposes that the Court "dismiss the first and third parts of the first ground of appeal" and "dismiss the second ground of appeal, on the basis set out in points 167 to 170 and 177 of the present Opinion, which should be substituted for those adopted by the General Court in paragraphs 353 and 370" of the July 2024 judgment.

That last element matters. On the right to be heard, Biondi agrees with the outcome the General Court reached but not with how it got there. The General Court had found that the Commission did breach ByteDance's right to be heard, then excused the breach. Biondi says there was no breach at all.

The opinion does not bind the Court. According to the Court's press release No 151/26, issued in Luxembourg the same day, "the Judges of the Court are now beginning their deliberations in this case. Judgment will be given at a later date." The press release adds that an appeal "on a point or points of law only" may be brought against General Court judgments and, "in principle, the appeal does not have suspensive effect." TikTok's designation has remained in force throughout.

The opinion's original language is English. ByteDance and the Commission presented oral argument at a hearing before the Court of Justice on 12 May 2026, according to the opinion.

How TikTok got here

The procedural history set out in the opinion is compact. On 3 July 2023, ByteDance notified the Commission, arguing "inter alia, that TikTok was a video-sharing platform," that ByteDance "cannot be presumed to be a gatekeeper" and that, in any event, the presumptions "were rebutted on the basis of the arguments and evidence which it had submitted."

The Commission responded by letter on 26 July 2023 with its preliminary views on classifying TikTok as an online social networking service and designating ByteDance. ByteDance replied on 2 August 2023. The Commission also held meetings with ByteDance before notification and on 17 August 2023, according to the Commission's position as recorded in the opinion.

The designation decision followed on 5 September 2023. According to the opinion, the Commission "considered that the arguments presented by the applicant to rebut the presumptions were not sufficiently substantiated so as manifestly to call them into question and rejected its arguments without opening a market investigation."

TikTok launched in the European Union in 2018, according to the opinion. PPC Land listed ByteDance among the six gatekeepers under the Digital Markets Act in March 2024, alongside Alphabet, Amazon, Apple, Meta and Microsoft. That same month, ahead of its obligations taking effect, TikTok rolled out a Data Portability API and upgraded its data download tool while saying it would keep complying during its legal challenge.

ByteDance asked the Court of Justice to set aside the General Court judgment, annul the designation and order the Commission to pay costs in both sets of proceedings. The Commission asked for the appeal to be dismissed with costs.

The presumption machine

The heart of the case is how the Digital Markets Act decides who is a gatekeeper. Regulation (EU) 2022/1925 was adopted on 14 September 2022. Article 3(1) sets three cumulative qualitative tests: a significant impact on the internal market; a core platform service that "serves as an important gateway for business users to reach end users"; and "an entrenched and durable position." Article 3(2) attaches a quantitative threshold to each test. Meeting a threshold triggers a presumption that the matching test is satisfied.

According to the opinion, "the EU legislature must have considered those thresholds to be proxies for the qualitative requirements." An undertaking that meets them must notify the Commission, which must then designate it "without undue delay and at the latest within 45 working days of receiving that information."

The presumptions are not conclusive. Article 3(5) lets an undertaking argue that, "exceptionally," it does not meet the qualitative tests despite meeting the thresholds. If the Commission considers those arguments "not sufficiently substantiated because they do not manifestly call into question the presumptions," it may reject them without a market investigation. If the arguments do manifestly call the presumptions into question, the text says the Commission "may" open a market investigation under Article 17(3).

There is a separate route. Under Article 3(8), the Commission can investigate an undertaking that does not meet the thresholds at all. There, according to the opinion, "the Commission is unable to rely on presumptions and bears the burden of proving that an undertaking meets those requirements." The Commission's cloud computing probes, which it opened into Amazon Web Services and Microsoft Azure in November 2025 even though neither service meets the size and user thresholds, are an example of that second path.

"May" does not mean "may"

Biondi's first move concerns the word "may." According to the opinion, exchanges at the hearing suggested the wording "supports an interpretation according to which that provision confers on the Commission a discretion as to whether to open such an investigation." Biondi disagrees. "In my view, the use of the term 'may' does not, in that context, confer a discretionary power on the Commission, rather it identifies the procedural steps available to the Commission once the relevant conditions laid down in that provision are satisfied."

Under his reading, once an undertaking presents sufficiently substantiated arguments manifestly calling the presumptions into question, "the Commission cannot reject those arguments without opening a market investigation, unless it is already apparent that the requirements under Article 3(1) are not satisfied." The press release puts it this way: in those circumstances "an investigation is required unless it is clear that the undertaking does not satisfy the qualitative requirements."

He offers four reasons. Article 17(3), which sets the deadline for concluding such an investigation, "necessarily implies an obligation on the Commission to open that procedure." The fourth subparagraph of Article 3(5) describes what happens after an investigation and so presupposes one. The scheme needs the rebuttal mechanism to have consequences: "there would be little practical value in providing an undertaking with the possibility of submitting arguments calling into question the presumptions if, notwithstanding their successful demonstration, the legal consequences remained entirely a matter for the Commission's discretion." And, according to the opinion, both parties' positions at the hearing and the Commission's own decision practice support the reading.

That practice includes a set of non-designation decisions cited in the footnotes: a 5 September 2023 decision concerning Microsoft services, a 13 May 2024 decision concerning ByteDance's online advertising services, and a 5 February 2026 decision on Apple's notifications for online intermediation verticals and online advertising services. Biondi cites them for the proposition that where an undertaking "demonstrates clearly and comprehensively" that the requirements are not met, or, "as expressed in recent decisions, provides a 'clear demonstration' to that effect," the Commission "does not initiate the Article 17 procedure and may proceed directly to non-designation."

The point is interpretive, but it is a win for undertakings in principle. It gives a company that clears the bar a right to an investigation, rather than leaving the decision to the Commission's discretion.

How high the bar sits

The win is narrow, because Biondi sets the bar high. According to the opinion, Article 3(5) "requires more than mere speculation or bare assertions" and "is not satisfied by mere prima facie evidence." If casting doubt were enough, "the Commission would be obliged to open a market investigation in each case where an allegation to that effect is raised," which would "compromise the objective of the presumption mechanism, which is intended to ensure a fast designation procedure."

His formulation, at point 36: "the standard of proof required by Article 3(5) amounts to a high degree of plausibility. The arguments have to be genuinely compelling, with no easily identifiable weaknesses in their capacity to rebut the presumption, in order to be considered as 'manifestly' calling it into question." Elsewhere he writes that "a presumption can be manifestly called into question only if the logic of the rebutting argument is clear and highly compelling." A footnote cites academic commentary to the same effect, quoting a 2024 Journal of Competition Law & Economics article by A. Ribera Martinez: "the legal standard for the [Commission] to decide whether to open a market investigation is very high."

The press release frames it in plainer terms. "Merely casting doubt on a presumption is insufficient," it says, but the undertaking "does not, however, have to prove conclusively, at this preliminary stage, that it is not a gatekeeper."

The word "exceptionally"

ByteDance's first complaint was that the General Court read an "exceptionality" requirement into the standard of proof. At paragraph 70, the General Court wrote that a market investigation "may be opened where the arguments of the undertaking concerned are sufficiently substantiated so as, exceptionally, manifestly to call into question the presumptions." ByteDance pointed out that "exceptionally" appears only in the first subparagraph of Article 3(5).

Biondi partly concedes the point on wording. "While the use of the term 'exceptionally' in paragraphs 70 and 71 of the judgment under appeal may be unfortunate, a comprehensive reading of the judgment does not disclose the error of law alleged by Bytedance." The exceptionality, he explains, concerns the later finding that the qualitative tests are not met, not the initial question of whether the presumptions are manifestly called into question.

In his reading, the General Court's actual test was in the second part of paragraph 71: arguments "must be capable of showing, with a high degree of plausibility, that the presumptions laid down in Article 3(2) of the DMA are called into question." He also notes paragraph 66, where the General Court said the standard cannot mean "that the applicant must adduce evidence which rebuts that presumption once and for all."

ByteDance also flagged five passages where the General Court used "demonstrate," "prove" or "disprove." According to the opinion, those words reflect the fact that the undertaking carries the burden, and "are otherwise neutral in regard to the standard of proof required." The General Court "did not require Bytedance to disprove the presumed circumstances," but "to demonstrate or prove that the inference of that circumstance from the quantitative threshold is manifestly questionable."

Why multi-homing did not save TikTok

The third part of ByteDance's first ground, aimed at paragraphs 132 to 285 of the General Court judgment, is where the commercial facts sit. It contains six complaints, all of which Biondi would reject. They all turn on the Article 3(2)(b) presumption that TikTok is an important gateway between businesses and users.

No ecosystem

ByteDance argued that, unlike Meta or Alphabet, it lacks a wider ecosystem it could draw on. The Commission disagreed. In recital 132 of the designation decision, according to the opinion, it found that ByteDance operates "its own ecosystem consisting of highly popular video-editing services, enterprise software, advertising, news and healthcare applications."

The Commission also said that even without an ecosystem, ByteDance had not displaced the presumption. Biondi treats this as decisive: an important gateway position "may equally result from the cumulative effect of other factors, such as significant network effects or lock-in effects of the platform." He adds that "Bytedance's unwillingness to accept that line of reasoning remains a recurring theme of its appeal."

The General Court had not held that a lack of an ecosystem is irrelevant, he writes, only that it is "insufficient to challenge the applicable presumption." TikTok had grown its user base "to a scale that was not negligible" compared with rivals that do run ecosystems.

Network effects without a social graph

ByteDance argued that TikTok lacks the network effects associated with the "social graph" of "real" social networks such as Facebook or Instagram. According to the opinion, ByteDance "had confined itself to arguing that the alleged lack of an ecosystem leads to the absence of significant network effects" and "had not made any independent argument suggesting that TikTok lacks such effects."

Biondi also rejects the idea that TikTok's gateway status rested on its classification as a social network. "TikTok's gateway position was presumed because the quantitative threshold laid down in Article 3(2)(b) of the DMA was met by the CPS at issue, and not because it was classified as an OSN."

Multi-homing

The core of ByteDance's argument was that TikTok users also use other services. The General Court, at paragraph 184, found that "all the online social networks to which [Bytedance] refers have high degrees of multi-homing, in so far as more than 65% of the end users of those networks use two other platforms in parallel." ByteDance said that distorted its evidence, because the only OSN to which Meta users multi-home above 65% is another Meta platform.

Biondi rejects the distortion claim. The 65% figure came from ByteDance's own data, which included YouTube at ByteDance's prompting. According to the opinion, "there is no indication as to the importance of the circumstance that Facebook and Instagram are operated by the same undertaking. The remark to that effect appeared only in the appeal."

The deeper problem for ByteDance is logical. If a whole category of service tends toward multi-homing, then showing that one service in that category is multi-homed proves little. "Showing that the CPS at issue is characterised by a significant degree of multi-homing is not sufficient manifestly to call into question the presumption if, as illustrated by Bytedance itself in the present case, a given category of CPS generally tends to demonstrate a significant degree of multi-homing."

Nor does multi-homing exclude lock-in. Biondi defines lock-in as "a situation where users are de facto constrained to remain within a gatekeeper's service because of high switching costs, behavioural biases or other factors that make switching or multi-homing unlikely." The DMA, he notes, "states that lock-in effects may hinder multi-homing, but it does not indicate that the prevalence of multi-homing implies a lack of lock-in effects." Lock-in may also limit what the Commission called "migration of focus" between platforms.

Asymmetry and the age of rivals

ByteDance's data showed asymmetric multi-homing: a high share of TikTok users also use Facebook and Instagram, while a lower share of those platforms' users also use TikTok. ByteDance read this as evidence that TikTok lacks network and lock-in effects.

The Commission offered a different explanation, which the General Court accepted and Biondi endorses: Facebook and Instagram "have been in operation for a longer period in the European Union and that they had a large user base in the European Union before TikTok was launched." Statistically, a large number of their users could be expected also to use TikTok. Other older platforms cited by ByteDance, Biondi notes, do not appear to have had a large EU user base.

Engagement and young users

The Commission also observed that ByteDance's multi-homing figures did not reflect intensity of use. In recital 141 of the decision, quoted in the opinion, it said "it has been repeatedly reported that TikTok has a significantly higher engagement rate than other online social networking services, with end users spending significantly more time on TikTok as compared to other online social networking services [in particular,] younger end users."

ByteDance called this "unsubstantiated speculation" drawn from a misread online article. Biondi checked the article. According to a footnote, it states: "The percentage of U.S.-based TikTok users by age: 10-19 - 32.5%, 20-29 - 29.5%, 30-39 - 16.4%, 40-49 - 13.9%, 50+ - 7.1%," citing Comscore. The Commission read that as 62% of users aged 29 or younger. Biondi adds that the article "also states elsewhere that 60% of TikTok users are 'Gen Zers' and that 67% of teens use TikTok."

He concedes that "the article does not appear to state explicitly that the engagement rate of young people is higher than the engagement rate of older users." But it does show higher engagement on TikTok among a user base in which young people form the majority, which in his view is enough.

The figures are United States data cited in a European designation. The opinion does not say whether equivalent EU figures were before the Commission.

ByteDance also objected to the use of time spent as the measure of intensity, arguing interactions would be better. According to the opinion, the General Court "did not hold that time spent is the only metric, but that it is an appropriate one." And on whether a platform can be a gateway for a subset of users, such as young people, Biondi says yes: the presumption rests on total active users, "while the requirement under Article 3(1)(b) does not necessitate the CPS in question to be such a gateway for all those users."

An important channel, not the only one

On business dependency, the General Court had said that a service need not be "the only channel" through which businesses reach users; "it is sufficient for [the CPS] to be an important channel for that purpose." ByteDance argued that dependency does not arise where other effective channels exist.

Biondi rejects that. "The DMA does not require extreme business dependency, which could exist, as the Commission points out, only where a CPS holds a monopoly." It is sufficient that the service "constitutes an important, but not the only, gateway for business users to reach end users." The press release summarises the point: "the fact that users also use other social networks does not, in itself, show that TikTok is not an important gateway."

For advertisers and brands, who are among the business users the DMA has in mind, this is the passage with the most practical weight. A platform can be one of several places where a brand reaches consumers and still be an important gateway under the law.

Evidence that came too late

ByteDance had also argued TikTok was small relative to other platforms. Before the General Court it produced comparisons against Pinterest, LinkedIn, Reddit, DailyMotion and Vimeo, and figures on the overall scale of social network activity. The General Court found those arguments and figures new and inadmissible.

Biondi agrees. During the administrative procedure, "it was originally Bytedance's choice to compare TikTok user numbers to those of Facebook and Instagram." He goes further: allowing new evidence before the General Court "is contrary to the scheme and objectives of the DMA," because Article 3 and Implementing Regulation (EU) 2023/814 require all rebuttal arguments to be filed with the notification. That rule raises the stakes of the 45-working-day notification phase for every future designation.

Two further claims fell on the same footing. Evidence on the engagement of registered business users covered "only a small portion of all business users" and so "revealed an easily identifiable weakness in its capacity to rebut the presumption." Evidence of TikTok's interoperability tools was, according to the opinion, not dismissed but found insufficient.

The right to be heard

The second ground of appeal raises the procedural question in the case. Under the presumption route, must the Commission share its provisional assessment of an undertaking's rebuttal arguments, and invite comment, before it designates? "I consider that it is not," Biondi writes.

The General Court had found that the Commission breached ByteDance's right to be heard on two points: its finding in recital 132 that ByteDance operates an ecosystem, and its reliance in recital 141 on reports about TikTok's engagement. It then excused the breach because ByteDance had not shown that the decision would otherwise have been different. ByteDance argued the proper test is whether being heard would have helped it defend itself. The Commission asked the Court to substitute grounds and hold that there was no breach at all.

Why the presumption route is different

Article 34(1) to (3) of the DMA requires the Commission to communicate preliminary findings, and base its decision only on findings the undertaking could comment on, when it designates after a market investigation. According to the opinion, those provisions do not apply to a designation under Article 3(4) and (5) where no investigation is opened: "That presumption-based mechanism for a designation decision is simply not referred to in Article 34(1) of the DMA."

The 45-working-day deadline reinforces the point. "By design, there may be insufficient time to communicate any provisional findings to the undertaking, still less to give it an opportunity to comment on them."

The general right to be heard under Article 41(2)(a) of the Charter of Fundamental Rights still applies, Biondi says, because a designation "gives rise to significant legal and operational consequences for the undertaking." But in this setting the facts come from the undertaking itself and the legal framework is fixed by the regulation. The right requires that the undertaking can challenge the presumptions and that the Commission examine its arguments "carefully and impartially." It "does not, however, require the authority to disclose its intended decision or its provisional assessment to the affected party."

There is one carve-out. If the Commission has used its investigative powers before opening a market investigation, the analysis changes. Biondi also acknowledges that the line between assessing an undertaking's evidence and making new findings "may be difficult to draw," and says the Commission "should therefore exercise caution and, to the extent possible, provide the undertaking concerned with information on its preliminary assessments."

Applying it to TikTok

On the ecosystem, the substituted reasons at points 167 to 170 run as follows. The Commission's other grounds proceeded on the assumption most favourable to ByteDance, that it had no ecosystem, and "would suffice" on their own. The ecosystem finding was based on ByteDance's own evidence; the Commission "simply did not agree." The right to be heard does not require a further chance to comment on that evidence. And the Commission acted with caution: its letter of 26 July 2023 had already said the absence of an ecosystem was not enough to rebut the presumption. If ByteDance disagreed with the outcome, it could challenge it in court, "That is exactly what Bytedance has done."

On engagement, the substituted reason at point 177 is that the Commission's references to online articles "cannot be regarded as introducing new facts requiring Bytedance to be heard again." They "were intended only to illustrate why multi-homing data, considered in isolation, did not necessarily demonstrate the absence of an important gateway position." Recital 141 is "an assessment of the probative value of the material submitted by Bytedance, not an independent finding."

Why the General Court's approach would backfire

Biondi argues that the General Court's position leads to "one of two equally undesirable outcomes." If the Commission had to share every element of its assessment, it "would find it difficult to comply with the time limit of 45 working days" and "would most likely be forced to open a market investigation in each and every case."

Alternatively, the General Court "set the bar unduly high, but paradoxically did so in a manner that provides a justification for any subsequent failure to attain it." Because the Commission typically rests its rejection on several grounds, an undertaking could never show that one unheard ground changed the result. "It is inconceivable, however, that the EU legislature could have intended the DMA to establish a mechanism that tolerates repeated breaches of the right to be heard without affording the party concerned any effective means of challenging those breaches."

The conclusion is that the General Court "erred in law in finding that the Commission was required to give Bytedance the opportunity to submit observations" on recitals 132 and 141, but was entitled to reject ByteDance's claim.

Why it matters beyond TikTok

The case is the first test at the Court of Justice of a regulation that now shapes how the largest advertising platforms in Europe operate. If the Court follows the opinion, three points become settled law across every designation: a high plausibility bar for rebutting the presumptions, a duty to investigate once that bar is cleared, and a limited right to be heard when the Commission does not investigate.

Other gatekeeper litigation is running in parallel. On 3 June 2026, the General Court annulled Facebook Marketplace's designation but upheld Messenger's in Case T-1078/23. PPC Land's report on that ruling noted that the Commission had cited the July 2024 ByteDance judgment in the Meta proceedings, and that Meta had relied on a 73% multi-homing figure that the court found covered only Germany in 2021. On 8 July 2026, Apple lost its challenge to the App Store and iOS designations in joined cases T-1079/23, T-1080/23 and T-214/24. Any appeal in those cases would be decided against whatever the Court of Justice says in ByteDance.

Enforcement has also moved from designation to penalties. The Commission fined Apple 500 million euros and Meta 200 million euros on 23 April 2025, and Meta filed an appeal against its decision in July 2025. On 23 July 2026, the Commission fined Google 890 million euros across two decisions, 460 million euros for self-preferencing in search and 430 million euros for Google Play anti-steering. A week earlier it had specified how Alphabet must share search data with rivals under Article 6(11). Each of those obligations depends on a valid designation.

For TikTok's advertising customers, the opinion does not change the rules TikTok currently operates under; the designation has applied since 2023 and the appeal has no suspensive effect. What it addresses is whether the label sticks. The Commission's 13 May 2024 decision on ByteDance's online advertising services, cited in the opinion as one of the decisions where the requirements were found not to be met, shows that the label applies service by service. TikTok's social networking service is designated; the opinion does not concern any other ByteDance service.

There is also a point about evidence. Biondi's reasoning repeatedly turns on gaps in ByteDance's own data: multi-homing figures that included YouTube, registered business-user figures that covered a small subset, comparisons with Facebook and Instagram that the company itself chose. Under the presumption route, the undertaking's notification file is the case. Comparisons not made by the 45th working day cannot be made later.

The press release gives no date for the judgment.

Timeline

Summary

Who: Advocate General Andrea Biondi of the Court of Justice of the European Union, ByteDance Ltd (operator of TikTok) and the European Commission.

What: Biondi proposed that the Court dismiss the examined parts of ByteDance's appeal against the General Court judgment upholding TikTok's gatekeeper designation under the Digital Markets Act. He held that the Commission must open a market investigation when rebuttal arguments manifestly call the presumptions into question, set that bar at a "high degree of plausibility," found no error on the important gateway question, and proposed substituting reasons on the right to be heard, finding that the Commission did not breach it.

When: The opinion was delivered on 15 October 2026, following a hearing on 12 May 2026. The judgment will follow at a later date.

Where: The Court of Justice in Luxembourg, in Case C-627/24 P, on appeal from General Court Case T-1077/23.

Why: It is the first Digital Markets Act case to reach the Court of Justice. The final judgment will fix how much evidence a platform needs to escape designation despite meeting the thresholds, when the Commission must investigate further, and what procedural rights a company has within the 45-working-day designation window. Those questions bear on parallel litigation by Meta and Apple and on every obligation and fine that depends on a designation.