On 15 October 2026, AG Biondi delivered his Opinion in Bytedance v Commission (C-627/24 P). The case has been closely watched as the first DMA designation dispute to reach the CJEU, offering the Court its first opportunity to interpret the regulation’s designation mechanism. AG Biondi’s proposed outcome holds no surprises: he agrees with the General Court (‘GC’) and recommends dismissing the parts of the appeal he examined, which would leave Bytedance as a gatekeeper in respect of TikTok. Yet the Opinion offers important clarifications which, if followed by the CJEU, could shape future designation decisions.
Background
In September 2023, the Commission designated Bytedance as a gatekeeper in respect of TikTok, finding that it met the quantitative thresholds in Article 3(2) of the DMA. Bytedance had sought to rebut the presumptions under Article 3(5) of the DMA, but the Commission rejected its arguments without opening a market investigation, finding that they were not sufficiently substantiated to manifestly call those presumptions into question. In its judgment of 17 July 2024 (T-1077/23), the GC upheld the designation and, importantly, clarified the standard of proof under Article 3(5) required ‘a high degree of plausibility’ (GC, para 71). The GC also found that the Commission had breached Bytedance’s right to be heard on two points, concerning the existence of an ecosystem and the intensity of TikTok’s use, but concluded that Bytedance had not shown that those breaches could have affected the outcome. On appeal, Bytedance raises two grounds: an infringement of Article 3(1) and (5) of the DMA, and a breach of its right to be heard. At the Court’s request, the AG limits his examination to the first and third parts of the first ground of appeal and the second ground (AG Opinion, para 11).
Presumption-based designation mechanism under the DMA and the Commission’s obligation to open a market investigation
To be designated as a gatekeeper under the DMA, an undertaking must meet three cumulative qualitative conditions under Article 3(1). Article 3(2) lays down corresponding quantitative thresholds, meeting which gives rise to rebuttable presumptions that the respective qualitative conditions are satisfied. As AG Biondi reframes, these thresholds serve as ‘proxies for the qualitative requirements’ (AG Opinion, para 14). The DMA also provides a second route to designation under Article 3(8), following a market investigation under Article 17, where an undertaking does not meet all the quantitative thresholds but nevertheless satisfies the qualitative conditions. Under the first route, however, the presumptions are not the end of the story. Article 3(5) allows an undertaking to submit sufficiently substantiated arguments manifestly calling those presumptions into question.
First, AG Biondi considers whether meeting the Article 3(5) threshold requires the Commission to open a market investigation, or whether it has discretion to do so. The GC did not directly address this issue, merely repeating the DMA’s wording, so the AG’s analysis goes beyond the GC’s analysis. The complication partly arises from the text of the DMA. The text of Article 3(5) states that, where an undertaking’s arguments do not manifestly call the presumptions into question, the Commission ‘may’ reject them without opening a market investigation. Conversely, where they do, the Commission ‘may’ open a market investigation under Article 17(3).
AG Biondi reads this wording as identifying the available procedural steps rather than conferring discretion (AG Opinion, para 22). Once the presumptions are manifestly called into question, an investigation is required unless it is already apparent that the qualitative requirements are not met (AG Opinion, para 23). The AG offers four reasons for this reading, each worth considering. First, Article 17(3) requires the Commission to endeavour to conclude the investigation within a specified period. In his view, this necessarily implies an obligation to open one, rather than merely regulating its duration. He also contrasts the wording of Article 17(3) with the language used under Article 17(1) (AG Opinion, paras 25–26). Second, the fourth subparagraph of Article 3(5), which addresses designation following an unsuccessful rebuttal, presupposes that an investigation has taken place (AG Opinion, para 27). Third, this reading is the only way to give practical effect to the rebuttal mechanism, as otherwise the right to challenge a presumption would have little value if successfully calling it into question had no legal consequence (AG Opinion, para 28). Finally, the AG draws support for this interpretation from the Commission’s two-track procedural practice: arguments manifestly calling the presumptions into question lead to a market investigation, while a clear demonstration that the qualitative requirements are not met leads directly to non-designation (AG Opinion, para 29).
In my view, the final two arguments provide the strongest support for the AG’s position. Allowing an undertaking to meet the threshold without any legal consequence would undermine the practical effect of Article 3(5). Read in this light, and against the Commission’s practice, ‘may’ should be read as accommodating the possibility of accepting a rebuttal without a market investigation where the undertaking provides a ‘clear demonstration’ that the requirements are not met. It shouldn’t be read as leaving the Commission free to reject arguments that manifestly call the presumptions into question without further investigation.
A further word is needed on these two procedural routes. The GC dismissed the relevance of the Commission’s previous decisions, also noting that it was not bound by its earlier decision-making practice (GC, para 68). While also recognising the non-binding nature of the Commission’s practice, AG Biondi states that ‘it nevertheless constitutes a relevant context a relevant contextual element when assessing how the mechanism established by that provision functions’ (AG Opinion, fn 6). Therefore, in contrast to the GC, the AG identifies two distinct procedural routes: opening a market investigation where the undertaking’s rebuttal arguments manifestly call the presumptions into question, and proceeding directly to non-designation where those arguments clearly demonstrate that the qualitative conditions are not met. Indeed, the latter route is reflected in the Commission’s approach (see for example, Alphabet (Gmail) decision para 150; Microsoft (Outlook) decision para 134; and Apple decision paras 65 and 107).
Standard of Proof for Rebuttal under Article 3(5)
On the matter of the standard of proof, AG Biondi endorses the GC’s ‘high degree of plausibility’ standard for manifestly calling the presumptions into question under Article 3(5) (AG Opinion, para 36). A demanding standard is understandable given the DMA’s reliance on a swift, presumption-based designation procedure. As AG Biondi also explains, if merely casting doubt on the presumptions were enough, the Commission would have to open a market investigation whenever an objection was raised (AG Opinion, para 34). This would undermine the purpose of that procedure and blur the distinction between designation under Article 3(2) and Article 3(8). Going further than the GC, however, the AG sharpens the standard and states that arguments must be ‘genuinely compelling, with no easily identifiable weaknesses’ in their capacity to rebut the presumption (AG Opinion, para 36).
One of Bytedance’s arguments on appeal was that the GC had raised the bar further by treating ‘exceptionally’ as an additional requirement for opening a market investigation, even though the word appears only in the first subparagraph of Article 3(5), not in the second and third. The AG agrees with Bytedance: exceptionality does not concern whether the presumptions are manifestly called into question. It concerns the later finding, with or without a market investigation, that the Article 3(1) requirements are not met despite the quantitative thresholds being satisfied (AG Opinion, para 41). In my view, this reading follows directly from the text. ‘Exceptionally’ qualifies the outcome, arguably discouraging challenges by signalling that rebuttal is intended to remain exceptional. This distinction did not, however, affect the AG’s conclusion. Although AG Biondi describes the GC’s wording as ‘unfortunate’, he concludes that the judgment, read as a whole, applied the correct test (AG Opinion, para 43).
At first glance, the AG’s Opinion does not depart from the GC’s approach: it retains the ‘high degree of plausibility’ standard but adds further wording, perhaps to clarify what that standard requires. Yet this addition, while clarifying that the bar is set high, also raises a question: how much room remains between manifestly calling the presumptions into question through arguments that are ‘genuinely compelling, with no easily identifiable weaknesses’ and the ‘clear demonstration’ needed for direct non-designation? If the first threshold is applied too strictly, that distinction risks losing its practical significance.
TikTok as an important gateway
The heart of Bytedance’s case on the merits was that TikTok is not an important gateway, so the GC erred in upholding the Commission’s finding that the presumption in Article 3(2)(b) has not been manifestly called into question. It raised six complaints against the GC’s reasoning. The AG rejects all of them, but some of his explanations will matter well beyond TikTok.
Bytedance’s first point was that, unlike Meta or Alphabet, it has no wider ecosystem to lean on. On this point, AG Biondi aligns with the GC in accepting that an ecosystem can be a relevant consideration, but not having one is not enough to call the presumption into question (AG Opinion, para 65). Bytedance next argued that the GC substituted its own reasoning for that of the Commission’s on network effects. The AG points to the DMA recitals and adds that network effects are one of the characteristics of a CPS (AG Opinion, para 68). The AG adds that it was for Bytedance to prove the absence of such effects, not for the Commission ‘to prove the contrary and explain why it had come to that conclusion’ (AG Opinion, para 72).
The third complaint was about multi-homing. The GC, relying on recitals 2 and 13, stated that a lack of multi-homing was one factor pointing towards a gateway. The AG emphasised the flip side: multi-homing ‘could, in some circumstances, be a contributing factor’ in challenging the presumption that service is a gateway (AG Opinion, para 77). Yet, the GC held that its mere existence, even to a significant degree, is not enough to rebut the presumption, as online social networks tend to observe a high degree of multi-homing (GC, paras 182-183). Bytedance pointed out this is in conflict with DMA’s language, as lack of multi-homing concerns every core category. The AG sided with the GC and stated that the relevant factors can vary from one category of service to another, and this is apparent in the language of Article 3(8), which is not ‘a mandatory step-by-step checklist’ (AG Opinion, para 79). He further notes that a rebuttal must also be rooted in ‘the circumstances in which the relevant [CPS] operates’, so general statements about a whole category of services will not be enough to rebut the presumption (AG Opinion, para 81). Therefore, this makes it clear that relying on general characteristics common to the relevant category of core platform service will stand little chance of rebutting the presumption, and that the arguments must explain what sets the particular service apart.
On the issue of lock-in, AG Biondi largely follows the GC but adds a definition of his own. The GC had held that lock-in is ‘a factor different from’ multi-homing. The AG goes a step further and defines the concept as denoting ‘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’ (AG Opinion, para 83). AG Biondi accepts that the absence of lock-in ‘could support an argument’ against the gateway presumption, but he doesn’t accept the premise that multi-homing necessarily rules out the existence of lock-in.
On business dependency, the AG simply agrees with the GC and does not go further than what the DMA text reveals. Bytedance argued that business users cannot be said to depend on TikTok if they have other effective ways of reaching the same end users. The GC rejected that argument, stating that a CPS does not need to be ‘the only channel’ through which business users reach end users; ‘it is sufficient for [the CPS] to be an important channel for that purpose, which those business users can access only if they have an account on that CPS’ (GC, para 210). The second criterion is met by almost every social platform, so it just leaves us back with ‘important’. The AG takes the same view. The DMA ‘does not require extreme business dependency, which could exist … only where a CPS holds a monopoly’(AG Opinion, para 102). That reading fits the DMA’s broader approach, which aims to tame gatekeepers that do not even have to be dominant in the competition law sense.
The fourth complaint alleged distortion of the evidence and a failure to state reasons concerning multi-homing, intensity of use and interoperability, all of which should be rejected in the AG’s view (AG Opinion, paras 104–122). The fifth complaint concerned the assessment of TikTok’s size compared with other platforms. One criticism was that before the GC, Bytedance had tried to compare TikTok with other platforms. The GC refused to consider this because it was a new argument raised. The AG endorsed this approach, adding that allowing undertakings to introduce new rebuttal arguments and evidence before the GC would be contrary ‘to the scheme and objectives of the DMA’ (AG Opinion, paras 127-131). The GC’s decision preserved an exception for new arguments responding to points raised in the decision on which the undertaking had been unable to comment (GC, para 234), while the AG does not seem to mention that exception.
Bytedance’s final complaint concerned the alleged application of the wrong legal test for an important gateway. In that regard, Bytedance made two arguments, one of which is rejected by the AG by relying on the GC’s ruling. On the first point, the AG pointed out that Bytedance misread the GC decision and that asymmetric multi-homing alone says nothing about whether network and lock-in effects exist, and it was up to Bytedance to show they did not (AG Opinion, para 137). On the second point, Bytedance argued that its registered business users, who should be the most engaged on the platform, showed limited engagement. The AG finds that this data covered only a small portion of all business users and was therefore not representative. The argument ‘revealed an easily identifiable weakness in its capacity to rebut the presumption’ (AG Opinion, para 140). This is actually the point where the AG applies the test he set out earlier, that rebuttal arguments must be ‘genuinely compelling, with no easily identifiable weaknesses’ (AG Opinion, para 36).
The right to be heard: where the AG parts ways with the GC
On the right to be heard, AG Biondi parts ways with the GC on the reasoning while reaching the same conclusion and rejecting Bytedance’s complaint. Bytedance argued that in the designation procedure under Article 3(4) of the DMA, the Commission was required under Article 41(2)(a) of the Charter to disclose its assessment. The GC found that the Commission had breached Bytedance’s right to be heard on two points: the existence of an ecosystem (GC, para 353) and the intensity of use of TikTok (GC, paras 358–370). The GC nevertheless held that those breaches did not warrant annulment, because Bytedance had not shown that the procedure could have led to a different outcome.
AG Biondi sides with the Commission in this matter. In his view, the procedural safeguards in Article 34 (1)-(3) DMA are not triggered where no market investigation is opened, and the Commission does not conduct an independent investigation (AG Opinion, para 147). The AG reasons that the strict 45-day deadline, by design, leaves insufficient time to formulate provisional findings, let alone give the undertaking an opportunity to respond, and that the Commission’s task under Article 3(5) is limited (AG Opinion, para 147). While this is in line with the text of the DMA, the distinction drawn for procedural safeguards between the Commission’s reliance on presumptions and its own reasoning may reveal a blurry line in practice, as the AG himself also acknowledges (AG Opinion, para 160). On the other hand, AG Biondi clarifies that designation under Article 3(4) has significant legal and operational consequences and can adversely affect the undertaking; thus, Article 41(2)(a) of the Charter applies to the designation decision under that provision. But drawing on case law, the AG says the right to be heard does not require the authority to disclose its provisional assessment, but it does require an opportunity to challenge the presumptions, a careful and impartial look at the arguments, and a detailed and reasoned decision (AG Opinion, para 158). AG Biondi therefore proposes that the Court dismiss this ground too, but for different reasons (AG Opinion, paras 167–170 and 177).
What to watch for next?
The AG Opinion builds on the GC’s judgment, offering further interpretative insights into the procedure under Article 3(5). In my view, three questions arising from the AG’s Opinion will merit particular attention: whether the Court agrees that ‘may’ effectively means ‘must’ where the rebuttal threshold has been met; whether it accepts the AG’s reasoning rather than the GC’s on the right to be heard; and whether it will endorse the two-track procedure reflected in the Commission’s practice and recognised by the AG. Finally, it should be borne in mind that the Opinion covers only part of the appeal, and it remains to be seen how the Court will address the remaining issues.
