The most important detail in the NSE co-location scandal was never buried in technical jargon. It was simple enough to explain to anyone who has ever waited in a queue. Some brokers, it was alleged, were not standing in the same line as everyone else. They were logging into secondary servers — backup infrastructure — while the rest of the market accessed primary servers -- that is plain and simple NSE co-location scandal.
Backup systems are designed for contingencies. They are meant to activate when something fails. But according to the various investigations and also grievances recorded in the National Company Law Appellate Tribunal (NCLAT) order, certain trading members were repeatedly accessing these secondary server machines even when the primary system was operational. And because those servers were less crowded, data could reach them faster and trading became faster - preferential access. All this is well documented in SEBI's Technical Advisory Committee report and also various other audit reports but the authorities have time and again turned a blind eye.
In algorithmic trading, faster is not cosmetic. It determines queue priority. It determines execution sequence. It determines who trades first.
Last week, the NCLAT dismissed the appeal in Manoj K. Sheth v. CCI, upholding the Competition Commission of India’s (CCI) decision not to pursue a deeper investigation into whether this architecture amounted to abuse of dominance by the National Stock Exchange of India (NSE). With that dismissal, the competition law route into one of India’s most debated market controversies came to an end.
The Tribunal’s conclusion is unequivocal. It states that the CCI “rightly examined the issues relating to co-location facilities” and correctly concluded that no case was made out regarding denial of market access or grant of preferential access to select trading members (para 147, p. 85–86). The allegations concerning the absence of a load balancer and randomiser, it adds, were also not established.
Yet the body of the order tells a story that is more layered than its final paragraph.
The judgment records that one of the core grievances was preferential treatment “through alleged manipulative Co-Location Services as well as access to secondary server to few favourable Trading Member” (para 117, p. 61–62). It notes the allegation that selected brokers had access to secondary servers “which was less crowded than the primary server” (paras 92–93, p. 34–35). It further captures the argument that lower load on secondary servers enabled faster data transmission and that, in the absence of a randomiser or load balancer, equitable access was compromised (para 49, p. 20).
These are not editorial reconstructions. They are part of the judicial record.
The order also recounts findings of the Technical Advisory Committee (TAC), including the “lack of randomizer in Normal TBT feed” and a “failure to monitor frequent secondary server access by certain trading members (TMs), with NSE issuing only emails/advisories despite repeated violations” (para 9, p. 5–6). Crucially, it notes that para 8.1.19(a) of the TAC Report underscored “the significance of microsecond variances in TCP-IP data dissemination for algo-trading by co-located brokers” (para 9, p. 5–6).
Microseconds matter. That is not a rhetorical device; it is a structural feature of high-frequency markets governed by price-time priority.
The Tribunal does not dispute that NSE is a dominant enterprise. It records that “the NSE is undisputedly a dominant player in the relevant market” (para 116, p. 61). Dominance, therefore, is not contested. The question was whether the conduct described — repeated access to less crowded backup servers, absence of a randomiser, limited monitoring — crossed the threshold into abuse under Section 4 of the Competition Act.
On that question, the Tribunal sided with the CCI. It reiterates that at the stage of forming a prima facie opinion under Section 26(1), the Commission is not required to conduct a detailed adjudication. It agrees that no further investigation was warranted (para 138, p. 76). It finds no error in the CCI’s assessment.
The reasoning turns on legal thresholds. The absence of proven denial of market access. The lack of demonstrable anti-competitive harm. The view that co-location services were offered on a first-come-first-served basis and uniformly charged. Without evidence that the exchange itself granted preferential access or structured its systems to advantage specific participants, the Tribunal was not persuaded that competition law had been breached.
And yet, the order leaves a lingering tension.
If secondary servers were repeatedly accessed while carrying lower load, and if monitoring amounted to advisories rather than enforcement, and if microsecond latency differences are acknowledged as significant, then what constitutes sufficient evidence of unfair advantage in a digital marketplace? At what point does architecture translate into liability?
The Tribunal’s answer is narrow: not on the material before it. The competition law mechanism cannot be triggered merely because technical imperfections existed or because some participants may have derived incidental advantage. Abuse of dominance demands more — clearer proof of exclusionary or discriminatory conduct attributable to the dominant enterprise.
For those who followed the co-location controversy over the past decade, that conclusion will read as either principled restraint or institutional reluctance. The order is careful, procedural and grounded in statutory standards. It does not sensationalise. It does not speculate. It confines itself to the record and to the legal test.
But it also means that one of the most debated questions in India’s market history — whether backup infrastructure effectively became a faster lane within a dominant exchange — will not be examined further under competition law.
The servers continue to run. The judgment delivers finality. Whether it delivers closure depends on how one reads the pages that precede its final paragraph. |
