Abidjan, Peering, and the Price of Keeping Traffic at Home

Abidjan has a real exchange, but not yet a full market

The core question is not whether Côte d’Ivoire has an Internet exchange point in any formal sense. It does. CIVIX is a functioning national IXP in Abidjan, established in 2013, managed by the telecom regulator ARTCI, visible in major routing and peering directories, and carrying non-trivial live traffic. Public BGP data shows a real exchange LAN with visible members including Orange Côte d’Ivoire, MTN Côte d’Ivoire, Moov/Atlantique Telecom, GVA, MainOne, Akamai, Packet Clearing House, and several smaller local or regional networks. CIVIX’s own public MRTG page shows current aggregate traffic in the double-digit gigabits per second range, not a dormant zero-traffic shell. That matters, because it means CIVIX is not merely a ministry PowerPoint or an obsolete membership list.

But the harder commercial question is whether CIVIX has become economically decisive. On the public evidence, not yet. The exchange is meaningful enough to save money and latency at the margin, especially for smaller domestic networks, public-sector services, and any workloads that can be terminated locally. Yet it is still too small, too regulator-led, too thin in visible content-platform depth, and too inconsistent in its public data quality to conclude that Abidjan has already become a self-sustaining peering gravity well on the model of Lagos, Nairobi, or Johannesburg. Publicly visible membership is only 14 ASNs as of the latest Internet Society Pulse view, aggregate visible port capacity is 78 Gbps, and the visible content roster is still narrow, with Akamai clearly present and Netflix discussed in official materials but not cleanly visible in current public peering directories. That is infrastructure, but not yet deep market thickness.

So the best answer is a split decision. CIVIX materially changes the economics of some domestic traffic already. It likely lowers transit leakage, shortens paths for traffic that can stay on-net, improves technical bargaining leverage for smaller operators, and makes Abidjan more plausible for caches and domestic hosting than it was a decade ago. But it has not yet obviously transformed the whole Ivorian internet economy. Too much still depends on incumbent-hosted points of presence, too much bargaining power still sits with the largest eyeball networks, and too much of the country’s service experience still hinges on submarine-cable resilience and upstream international paths. In commercial terms, CIVIX has crossed the line from symbolic policy object to real infrastructure, but it has not yet crossed the line from useful infrastructure to decisive market institution.

That distinction matters because IXPs create value in stages. The first stage is existence. The second is enough domestic participation to stop obvious tromboning of local traffic via Europe or other regional hubs. The third is enough density and neutrality to attract platforms, caches, and private interconnection. CIVIX is clearly past stage one and meaningfully into stage two. The evidence that it has fully reached stage three is still incomplete.

From failed CI-IXP to regulator-run CIVIX

CIVIX’s present structure only makes sense against Côte d’Ivoire’s earlier failure. A 2007 NSRC report described the implementation of a predecessor CI-IXP project, trained local engineers, deployed equipment in Côte d’Ivoire Telecom’s premises, and noted that five ISPs were connected or expected to connect. The project even obtained independent IP space and an ASN “in order to ensure the total independence of CI-IXP.” In other words, technically, the country had already learned the mechanics of building an exchange nearly two decades ago.

The project nonetheless failed to become a durable national market institution. The clearest public explanation comes from ITU material on IXP policy and broadband migration. It states that the old CI-IXP was housed in the premises of the incumbent operator, which also served as the main node for local and international traffic, and that the exchange failed in execution. The same ITU document says that ARTCI and the Ministry of Digital Economy and Posts then led a renewed effort that produced CIVIX, with two points of presence, one at Orange Côte d’Ivoire Télécom and the other at MTN Côte d’Ivoire, and that seven entities were connected by 2016. Commercially, this history is crucial: Côte d’Ivoire’s IXP problem was never only technical; it was a neutrality and trust problem among competitors.

That historical lesson is visible in CIVIX’s current formal identity. Packet Clearing House lists CIVIX as an active, governmental IXP in Abidjan, managed by ARTCI and established in July 2013. PeeringDB lists the long name as “Cote d’Ivoire Internet Exchange Point (CIVIX)” in Abidjan. AFRINIC-linked registry data visible through APNIC and other WHOIS mirrors shows address resources explicitly identifying the organization as “Cote d’Ivoire Internet Exchange Point,” including a public service network on 196.49.0.0/24 and the exchange LAN on 196.223.4.0/24, associated with AS329666 and with the AFRINIC organization handle ORG-CdIE1-AFRINIC. Those records collectively show that CIVIX is not a loose industry club; it is a regulator-anchored infrastructure entity with dedicated numbering resources and public routing identity.

Governance remains unusually public-sector-shaped. CIVIX’s charter and ARTCI reporting indicate that a formal operating charter was signed in October 2019, after which a consultative committee, CoCIVIX, was established. Euro-IX’s 2021 spotlight, which appears to have been based on information from CIVIX, says CoCIVIX includes one representative of each connected actor, two representatives of the CIVIX specialized center, and one representative designated by ARTCI’s Director General as chair. The charter material also indicates that the committee oversees functioning and that its deliberations are governed by an internal regulation. This is more collaborative than a purely state-run utility, but still less institutionally independent than many mature member-governed IXPs.

That mixed structure has predictable advantages and disadvantages. The advantage is convening power. Regulators can force the conversation, reduce holdout behavior, and use public legitimacy to bring together competing operators that might otherwise prefer bilateral opportunism over a shared switching fabric. That seems to be precisely what ARTCI has done, not only through CIVIX’s architecture and charter, but through repeated “Peering Days” industry events and official media campaigns linking CIVIX to lower latency, lower internet cost, and national digital sovereignty.

The disadvantage is that regulator-led IXPs often remain partly political objects. PeeringDB shows CIVIX with “Terms: Not Disclosed” and “Service Level: Not Disclosed,” which is a warning sign for any analyst trying to model commercial incentives. It does not mean the exchange lacks rules; it means the public market cannot see them clearly enough to price behavior. A private or member-led IXP that publishes port policies, facilities, service guarantees, and connection procedures more transparently sends a stronger signal to outside platforms and large counterparties. CIVIX’s public visibility still looks more administrative than commercial.

There is also a subtler neutrality issue. CIVIX’s current architecture, according to the charter and official site materials, is built around two points of presence in Abidjan, one at Orange CI and one at MTN CI, with the NOC at ARTCI. That is better than a single-incumbent site, but it is not the same thing as a single carrier-neutral exchange hall. In effect, Côte d’Ivoire solved its original neutrality problem by splitting the exchange across two large operator environments, rather than by immediately anchoring it in a neutral colocation facility. That reduced dependence on one incumbent, but did not eliminate dependence on incumbent-hosted infrastructure altogether.

The commercial significance of that design becomes clearer when read against later developments. In 2025, ARTCI and Raxio signed an MoU to host a third CIVIX point of presence inside Raxio’s Tier III, carrier-neutral facility in Grand-Bassam, in the VITIB technology park. ARTCI described the additional PoP as improving service quality, resilience, and local traffic handling, while Raxio described it as a milestone in national digital infrastructure. Put plainly, the market and the regulator both appear to understand that the original two-PoP structure was useful, but not sufficient. The move toward a neutral facility is not cosmetic. It is an attempt to solve the exact commercial problem that prevented the first IXP era from becoming durable.

What the routing and traffic data actually show

The strongest case that CIVIX is real infrastructure comes from public routing data rather than official rhetoric. Hurricane Electric’s exchange page for CIVIX shows a visible exchange LAN with active members and IP assignments on the peering subnet. The list includes AS20940 Akamai, AS29571 Orange Côte d’Ivoire, AS36924 GVA Côte d’Ivoire, AS36974 MTN Côte d’Ivoire, AS37190 Atlantique Telecom/Moov, AS37282 MainOne, AS37381 VipNet, AS327746 Dataconnect Côte d’Ivoire, AS327773 GUCE CI, AS328809 EDIATTAH, AS328840 ST Digital, AS329666 CIVIX itself, and PCH’s AS42 and AS3856. This is not hypothetical peering: it is visible exchange-LAN population.

The Internet Society Pulse IXP tracker adds useful structure to that picture. It reports 14 members at CIVIX as of May 2026, with a cumulative 78 Gbps of member port capacity. It also reports that nine of the fourteen members peer at the route server, twelve of fourteen members use RPKI, and the member mix includes five Cable/DSL/ISP networks, two educational/research networks, two NSPs, two network-services networks, one enterprise, one content network, and one unknown. It further reports that CIVIX does not participate in MANRS’ IXP program and does not host a RIPE Atlas anchor. Commercially, that paints a picture of an exchange that has enough technical maturity to matter, but not yet the best-in-class operational signalling of a highly institutionalized peering fabric.

The route server itself is also visible. Packet Clearing House’s public peering information shows PCH operating at CIVIX with route-server-related addresses 196.223.4.18 and 196.223.4.19, and PCH’s broader peering page lists CIVIX among the IXPs where it provides operational support. This is important because route servers reduce transaction costs. At a small or medium-sized IXP, the difference between having to negotiate and configure eleven bilateral sessions and being able to establish one multilateral session can determine whether a smaller network peers locally at all. CIVIX has that enabling layer.

The routing posture of major members also matters. Orange Côte d’Ivoire publicizes a 10 Gbps operational port at CIVIX, an open peering policy there, and additional interconnection presence at AMS-IX Lagos and IXPN Lagos, as well as an interconnection facility presence at Equinix AB1 in Grand-Bassam. GVA has a 10 Gbps CIVIX port but a much wider African and European peering footprint, with visible peering in Accra, Lagos, Burkina Faso, Brazzaville, Nairobi, Johannesburg, Paris, Lisbon, London and more. MainOne is present at CIVIX, but only at 1 Gbps, while maintaining much larger public peering capacities elsewhere, including 50 Gbps at AMS-IX Lagos and 100 Gbps at LINX London. Moov/Atlantique Telecom presents 10 Gbps at CIVIX and 30 Gbps at AMS-IX Lagos. These details imply that for the larger and more regionally connected players, CIVIX is part of the interconnection portfolio, not the whole game.

Akamai’s presence is the most commercially significant content-side signal visible in public data. Akamai shows a 10 Gbps operational port at CIVIX and participates via route server. Because Akamai is a serious, global edge-content network, its choice to peer locally is evidence that CIVIX has enough domestic demand concentration to justify a real edge presence. Yet it is equally important that public member data shows only one clearly visible large content network in the current list. A single major CDN proves usefulness; it does not prove platform clustering.

Live traffic telemetry confirms that the exchange is carrying meaningful traffic. CIVIX’s public MRTG “aggregate” page on 30 June 2026 showed daily peaks around 14.7 Gbps and weekly peaks around 16.7 Gbps, with monthly averages in the low double-digit Gbps range. Those levels are not African top-tier, but they are large enough to matter economically for a domestic exchange in a mid-sized West African market. Double-digit gigabit steady-state traffic is enough to affect transit bills, user experience, and internal network design decisions.

At the same time, the traffic evidence is noisier than it should be. The live MRTG page reports a “Max Speed” of 40 Gbps, while Internet Society Pulse, drawing from PeeringDB, reports 78 Gbps of cumulative member port capacity. That difference may simply reflect monitoring configuration or full-duplex graphing conventions, but it is still a disclosure-quality issue. More importantly, official CIVIX annual-report snippets publish a “volume total de trafic échangé” series that grows from 435 Mbps in 2013 to 490,992 Mbps in 2020, 933,025 Mbps in 2021, and 1,837,475 Mbps in 2024, while also saying Orange handled 530,500 Mbps and MTN 450,046 Mbps in 2024. Those reported “Mbps” figures are directionally useful as evidence of growth, but they are not readily reconcilable with the live graph and current visible port capacity if read literally as instantaneous peak bitrate. The safest commercial reading is that traffic has clearly expanded by orders of magnitude over time, but CIVIX’s public reporting units and methodologies are not transparent enough for precise traffic economics.

That caveat matters because infrastructure investors and counterparties care less about slogans than about telemetry quality. An exchange that wants to attract hyperscaler caches, serious cloud nodes, or more selective content networks needs public metrics that are coherent across live traffic pages, peering directories, annual reports, and routing registries. CIVIX’s public data now proves existence and growth, but not yet with the degree of statistical hygiene that would make outside operators instantly comfortable. That is one reason the exchange still looks, from a commercial distance, partly like a real network and partly like a policy program.

The economics of domestic traffic and bargaining power

An IXP changes economics through four channels: avoided transit cost, lower latency, route control, and bargaining power. In the classic model, an ISP compares the cost of buying international or upstream transit for traffic that could have been exchanged locally against the local costs of an IXP port, a cross-connect, and the operational burden of peering. If enough of a network’s traffic is domestic or cache-eligible, local peering wins. The logic is standard enough that RIPE, the Internet Society, and ITU all describe IXPs as mechanisms for keeping local traffic local, reducing transmission costs, improving quality of service, and enabling national internet ecosystems.

For Côte d’Ivoire, this logic is economically plausible even without a disclosed CIVIX tariff sheet. The country’s internet market is large enough that there is real same-country demand to exchange. DataReportal estimated 11.23 million internet users at the start of 2024, while an official economic information portal citing sector data reported 29.17 million active mobile-internet subscriptions in 2023. The World Bank has also described mobile coverage as broad, while noting that adoption and fixed-broadband household penetration remain relatively low. In plain English, this is a mobile-heavy, urbanizing, still-under-digitized market where a lot of user demand exists, but not all of it has yet migrated into local data-centre and platform infrastructure. That is exactly the kind of market where a domestic IXP can matter first for cost and QoS, and only later for deep content localization.

The size of the probable economic benefit is very uneven across counterparties. For a smaller access provider, enterprise network, university, or government-service operator, CIVIX can change the economics materially because the alternative may be paying upstream transit just to reach the country’s largest eyeball networks, or to deliver packets to government platforms and domestic service endpoints. The existence of GUCE CI, Dataconnect, and other non-mass-market networks on the exchange supports exactly this interpretation: the exchange is not only for consumer broadband operators; it is also a domestic interconnection fabric for service providers and institutions with national traffic needs. The fact that the primary Point CI server for the .ci country domain has been hosted at CIVIX reinforces the same point.

For the largest eyeball networks, however, the gains are less revolutionary. Orange, MTN, and Moov have enough traffic scale that they can extract value from CIVIX, but they also have enough scale to negotiate bilateral arrangements, buy transit more efficiently, and shape the local interconnection agenda. Orange’s open peering policy and 10 Gbps port at CIVIX help, but Orange also benefits from its broader market power and facility presence. GVA, with a much wider African and European peering footprint than most local operators, has alternatives. MainOne’s relatively small 1 Gbps port at CIVIX compared with much larger external peering capacities suggests that its Abidjan position is still more local-extension than core-hub. In other words, CIVIX improves the economics of local exchange, but it does not abolish hierarchy within the Ivorian telecom market.

The concentration of Côte d’Ivoire’s fixed-broadband market strengthens that reading. In a 2024 official consultation on internet access conditions and pricing, ARTCI stated that Orange very largely dominates fixed internet in general and FTTH in particular, with around 84% market share. In fixed access markets with that kind of concentration, an IXP can reduce dependence for smaller players, but it does not by itself neutralize the bargaining power of the leading network. A dominant fixed operator still has leverage in retail pricing, customer acquisition, on-net destination value, and content negotiations. CIVIX can moderate those asymmetries; it cannot erase them on its own.

This is where the “price of keeping traffic at home” becomes a bargaining-power story as much as a transport-cost story. If local traffic must leave the country to come back, upstream providers and foreign hubs capture value. If domestic operators can exchange traffic in Abidjan, the largest local access networks become must-reach peers on domestic terms, and smaller networks gain an alternative to buying more expensive upstream capacity for obviously local destinations. That is a genuine gain in bargaining position. Yet the scale of the gain depends on who else is present. A thin IXP with one major CDN and a handful of local operators improves negotiating leverage only incrementally. A thick IXP with multiple platforms, caches, and clouds can change the whole structure of negotiations. CIVIX is not yet obviously at that second stage.

The submarine-cable outage of March 2024 is a useful stress test. Cloudflare, Reuters, TeleGeography and the Internet Society all reported serious disruption across West and Central Africa, including Côte d’Ivoire, after subsea cable failures off the West African coast, with MainOne later saying its break was off Côte d’Ivoire. If an IXP had already localized a large share of everyday digital life, one would expect a meaningful layer of domestic traffic and services to remain relatively resilient even amid international disruption. The public evidence does not show such a clear resilience dividend for Côte d’Ivoire. That does not mean CIVIX was useless during the event. It means the broader service mix was still heavily dependent on external routes, foreign-hosted platforms, or upstream capacity. An IXP lowers some fragility; it does not fully offset a market that remains internationally hosted.

This is why the report’s answer must remain sceptical. Yes, CIVIX materially changes domestic-traffic economics in some cases already. But no, the public evidence does not yet support a stronger claim that CIVIX has fully repriced domestic internet economics across Côte d’Ivoire. The exchange has created a domestic option. It has not yet become the unavoidable center of gravity for the market.

Carrier neutrality, cache attraction, and platform signals

The decisive hinge variable for Abidjan is no longer whether an exchange exists. It is whether the exchange sits in an interconnection environment that global platforms and neutral counterparties actually want to use. That is why carrier neutrality matters so much. Historically, CIVIX’s two-PoP architecture was lodged in the environments of Orange CI and MTN CI. Functionally that created redundancy and broadened participation, but commercially it still meant that the exchange lived in spaces tied to major competing operators. That arrangement is often good enough to start a market, but not always good enough to attract the full stack of caches, remote peers, and neutral colocation demand.

The emerging neutral-data-centre layer changes the investment case more than any other recent development. Equinix’s AB1 facility in Grand-Bassam is now described by Equinix as a carrier-neutral IBX data centre strategically located for global subsea traffic, serving local and international enterprises as well as content and network service providers. Raxio’s Côte d’Ivoire facility, CIV1, is described by Raxio as the country’s first carrier-neutral Tier III data centre, located in Grand-Bassam inside VITIB and along major fibre routes. In 2025, ARTCI and Raxio then agreed to host a third CIVIX PoP in that environment. Taken together, these developments mean that the physical preconditions for a more commercially credible interconnection market in greater Abidjan now exist in a way they did not a few years ago.

That matters because platforms choose locations differently from regulators. They care about customer density, traffic growth, latency benefit, route predictability, facility neutrality, power reliability, remote-hands competence, and whether enough counterparties can be reached efficiently. Official discourse around CIVIX increasingly recognizes this. ARTCI’s July 2024 communications called the Netflix project “vital” for CIVIX. An official ARTCI source also framed the future of CIVIX in terms of promoting content such as Netflix. A November 2024 ARTCI snippet linked CIVIX to keeping national traffic in-country and explicitly mentioned Google and Netflix in the context of improving internet attractiveness. These are not proofs that Google and Netflix are deeply embedded in Abidjan. They are evidence that the regulator understands platform cache attraction as the next economic frontier.

There is some hard evidence of movement. CIVIX’s 2024 annual report says that two new actors were connected in 2024: Netflix and EDIATTAH. The 2023 annual report had already mentioned a survey carried out in view of Netflix’s connection. That is significant as an official claim, because Netflix is exactly the kind of large-volume streaming workload that can transform an IXP’s economics. A Netflix handoff or cache presence can lower upstream cost, improve buffering performance, and increase the value of local interconnection for every eyeball network that connects to it.

But here the evidence becomes commercially interesting precisely because it is inconsistent. Public routing and peering directories in mid-2026 do not obviously show Netflix’s AS2906 among CIVIX’s visible members. The Internet Society Pulse member list for May 2026 shows 14 members and includes Akamai but not Netflix. Hurricane Electric’s CIVIX participant page likewise shows Akamai and the local operators, but not Netflix. This discrepancy could mean several things: Netflix may be connected in a way not reflected in public peering databases; it may be using non-public arrangements; a cache may be present without public peering-list disclosure; or the official annual report may be using “connected” in an administrative sense not identical to “publicly visible peering member.” The commercial point is not to accuse one side of error. It is to note that a market trying to attract platforms still has metadata ambiguity around one of the world’s most important cache operators.

Akamai’s clearly visible 10 Gbps presence, by contrast, is settled evidence. That proves CIVIX is already attractive enough to support at least one major global content-delivery footprint. The same cannot yet be said, in public data, for a broad cluster of hyperscalers or OTT platforms. There is no comparable public evidence in the current member roster of Cloudflare, Meta, Google LLC, or Microsoft peering directly at CIVIX. That absence does not prove no edge infrastructure exists in-country. It does mean that the public peering surface still looks shallow by the standards that usually precede large-scale cache clustering.

Even CIVIX’s own membership rules hint at the problem. Public CIVIX materials say that any physical or legal person seeking membership must first have a public ASN and public IP address, and that external actors must justify a peering agreement with at least one local CIVIX member before becoming members. For a domestic ISP, that may be normal enough. For a global platform or foreign content network considering market entry, it is a subtle friction. The rule effectively says: you may need local peering relationships before the exchange will fully admit you. That is not fatal, but it is less frictionless than the message many content networks prefer, which is: here is a neutral hall, here are the port prices, here is a route server, here is the route-server policy, here are the local facilities, and here are the eyeball networks already present.

The recent “Peering Days 2026” conversation is best read as an intelligence signal rather than settled proof. Local industry publication Digital Mag reported that participants expected CIVIX to “reinvent” itself infrastructurally, lower latency, lower internet cost, and onboard more public, private, national, international and even satellite-sector actors. ARTCI’s own LinkedIn and Facebook posts framed the event around more performant, resilient, and sovereign connectivity. Commercially, this tells us three things. First, the exchange is active enough to justify live industry mobilization. Second, its sponsors still believe significant value remains unrealized. Third, the market itself does not speak as though the cache-and-platform problem has already been solved. If it had been solved, the discussion would be about scaling success, not reinventing the infrastructure.

In that sense, the Raxio PoP is the single most important forward-looking development. If the third PoP becomes a real neutral meeting point, if public terms become clearer, and if more high-volume content players become visibly active, then CIVIX can move from being primarily an access-network interconnection tool to becoming a broader market-shaping digital-infrastructure node. Without that progression, the economics of cache attraction remain positive in theory but unproven in scale.

Competitive position, risks, and what would change the view

Côte d’Ivoire’s competitive position is better than it was, but weaker than it could be. The country has major submarine-cable landing diversity in Abidjan and Grand-Bassam, with TeleGeography listing 2Africa, ACE, MainOne, Maroc Telecom West Africa, SAT-3/WASC and WACS at Abidjan. The World Bank had already described five international cables by 2022, with a sixth expected. The market also now has at least two meaningful carrier-neutral data-centre propositions in the Grand-Bassam/VITIB orbit, through Equinix AB1 and Raxio CIV1. These are real strategic assets. They mean you can now tell a more credible story about Abidjan as a regional edge location than you could in the “one IXP, no carrier-neutral data centre” picture described in a 2021 higher-education connectivity study.

Yet the market still has clear structural risks. The first is concentration. Orange’s dominance in fixed access, the scale of MTN in mobile, and the continuing importance of the incumbent and large mobile operators mean that local peering cannot be analyzed as a neutral bazaar of equals. It is a market in which the biggest eyeball networks remain powerful gatekeepers. The second risk is disclosure quality. CIVIX’s public materials show inconsistent units, incomplete publication of terms, and directory gaps. The third is platform depth. Akamai is there; Netflix is an official strategic focus, but its public visibility is ambiguous; the wider roster of globally important cache or cloud networks is still thin. The fourth is resilience. The March 2024 cable-cut event showed that despite the presence of an IXP, Côte d’Ivoire’s internet experience can still be heavily disrupted by offshore failures.

Another risk is that CIVIX remains disproportionately legible as a regulator project rather than as a commercially optimized exchange environment. That can be enough to convene local operators, but global content networks usually prefer markets where the neutral-colocation and interconnection proposition is highly standardized. Equinix and Raxio can help solve that. But until the third PoP is visibly operational in the neutral environment, until more facilities and service terms are disclosed, and until the platform roster deepens, a sceptical investor or infrastructure strategist will still treat CIVIX as promising but not fully de-risked.

The biggest positive interpretation of today’s evidence is that CIVIX has already cleared the threshold where local peering is economically rational for a meaningful slice of the market. The biggest negative interpretation is that CIVIX is still caught between two eras: too real to dismiss, but not yet dense enough to dominate. On balance, the positive interpretation is stronger. Live traffic, visible BGP members, major-operator participation, a clear Akamai presence, a formal charter, an active consultative committee, and a new neutral-PoP strategy are too much evidence to write off as a mere membership-list artifact. But the negative interpretation also survives because the exchange’s public telemetry, facility neutrality, and platform mix have not yet reached the point where the economics become self-evident to outside capital or counterparties.

What would change the view decisively? First, a clearer operating footprint at neutral facilities: a visibly live third PoP at Raxio, public facility listing in PeeringDB, and better disclosed cross-connect terms. Second, a deeper and publicly verifiable platform roster: not just Akamai and a partly opaque Netflix project, but visible presence from additional major CDNs, cloud networks, or OTT platforms. Third, better telemetry: coherent public reporting of peak traffic, average traffic, member counts, and growth series using units that match the live graphs. Fourth, more evidence that domestic high-value applications are really being served locally: public-sector services, fintech, media, gaming, education, and enterprise SaaS endpoints that can be shown to stay in-country or at least in-Abidjan. Fifth, observable resilience during future international-fault events. If local services continue to work when submarine routes wobble, that is when bargaining power truly migrates back home.

Until then, the right commercial conclusion is disciplined rather than romantic. CIVIX already matters. It probably saves real money for some networks and measurably improves path efficiency for some domestic traffic. It gives ARTCI a real infrastructure lever rather than only a policy slogan. It creates a platform on which Abidjan could become a more serious cache and edge location. But it is still in transition from “a nationally important IXP” to “a market-making interconnection hub.” That transition is plausible. It is not fully proven.

Evidence ledger

Source	URL	Source type	What it supports	What it does not prove	Why it matters economically
CIVIX MRTG aggregate traffic	http://mrtg.civix.ci/aggregate.html	Official live traffic page	CIVIX is carrying live traffic in the double-digit Gbps range; the exchange is operational, not dormant.	It does not reveal traffic composition, bilateral splits, or retail cost savings.	Live traffic is the minimum proof that local exchange is real and potentially cost-reducing.
PCH CIVIX directory entry	https://www.pch.net/ixp/details/1801	Authoritative IXP directory	CIVIX is active, governmental, managed by ARTCI, established July 2013.	It does not prove neutrality quality, pricing, or actual traffic.	Confirms formal institutional identity and public-sector governance.
Hurricane Electric CIVIX member list	https://bgp.he.net/exchange/CIVIX	Public BGP/routing directory	Visible exchange-LAN participants include Orange, MTN, GVA, Moov, MainOne, Akamai, PCH and others.	It does not show private peering, hidden members, or traffic volumes by participant.	Demonstrates real interconnection and who is publicly visible at the exchange.
Internet Society Pulse IXP tracker for CIVIX	https://pulse.internetsociety.org/en/ixp-tracker/ixp/338/	IXP analytics based on PeeringDB and Internet Society tooling	Reports 14 members, 78 Gbps cumulative capacity, route-server usage, RPKI adoption, no MANRS IXP participation, no RIPE Atlas anchor.	It does not prove actual throughput or commercial terms.	Good for benchmarking CIVIX’s maturity and operational signalling against peers.
PeeringDB entry for CIVIX	https://www.peeringdb.com/ix/810	Self-reported peering database	Shows official website, contact info, traffic-stats URL, prefixes, and lack of disclosed terms/service level.	Self-reported data can lag or omit facilities and bilateral realities.	The absence of published terms is itself informative for market transparency and cache attraction.
NSRC 2007 CI-IXP report	https://nsrc.org/AFRICA/CI/Report-IXP-CI-2007.pdf	Technical project report	Shows Côte d’Ivoire tried to build an exchange in 2007, with training, equipment and multiple ISPs.	It does not explain the full political economy of later failure.	Proves the country’s IXP story is older than CIVIX and that technical know-how was not the only constraint.
ITU broadband/IXP policy report	https://www.itu.int/dms_pub/itu-d/opb/stg/D-STG-SG01.01.1-2017-PDF-E.pdf	Multilateral policy report	States the earlier CI-IXP failed, links that failure to incumbent-site issues, and describes the creation of CIVIX with two PoPs.	It does not quantify CIVIX’s present-day performance.	Gives the cleanest public statement of the neutrality/trust problem that shaped CIVIX’s design.
CIVIX charter and governance snippets	https://www.civix.ci/images/CIVIX/CHARTE.pdf	Official charter and secondary industry summary	Supports two-PoP architecture, CoCIVIX governance, and consultative management model.	It does not show whether this governance is commercially efficient in practice.	Governance affects trust, holdout behavior, and the ability to attract outside counterparties.
CIVIX annual reports 2023 and 2024	https://www.civix.ci/images/CIVIX/Rapport_annuel_CIVIX_2023.pdf and https://www.civix.ci/images/CIVIX/Rapport_dactivit_annuel_CIVIX_2024.pdf	Official self-reporting	Supports claims of traffic growth, the Netflix project, EDIATTAH connection, and Orange/MTN traffic dominance.	Unit labeling is inconsistent with live graphs; the figures do not allow clean traffic modeling.	Useful directionally for growth and strategic focus, but weak for precise economics.
Raxio–ARTCI third-PoP announcement	https://www.businesswire.com/news/home/20250626829454/en/...	Official company announcement and regulator-linked media	Supports the 2025 move to host a third CIVIX PoP in a carrier-neutral Tier III data centre in Grand-Bassam.	It does not prove the PoP is already carrying material traffic.	The move toward neutrality is the key structural change that could deepen cache attraction and market confidence.
Equinix AB1 and Raxio CIV1 facility pages	https://www.equinix.com/data-centers/europe-colocation/cote-divoire-colocation/abidjan-data-centers/ab1 and https://www.raxiogroup.com/data-centres/cote-divoire/	Official facility pages	Support the existence of carrier-neutral colocation in Grand-Bassam/VITIB, aligned to subsea routes and fibre corridors.	They do not prove CIVIX has fully migrated there or that platforms are already anchored there.	Neutral facility choice is decisive for platform-cache and counterparty attraction.
TeleGeography Abidjan landing-point record	https://www.submarinecablemap.com/landing-point/abidjan-cte-divoire	Industry infrastructure map	Supports the breadth of Abidjan’s submarine-cable connectivity, including 2Africa, ACE, MainOne, MTWA, SAT-3/WASC and WACS.	It does not prove retail resilience or peering depth.	Cable diversity improves the logic for making Abidjan an edge and interconnection hub.
Cloudflare, Reuters and Internet Society on March 2024 cable outage	https://blog.cloudflare.com/undersea-cable-failures-cause-internet-disruptions-across-africa-march-14-2024/ and related reporting	Network-observability and news reporting	Supports the claim that Côte d’Ivoire remained vulnerable to international cable failures in 2024.	It does not prove which local services remained up via CIVIX.	A real test of how much “keeping traffic at home” has actually been achieved.
ARTCI consultation on fixed-internet market structure	https://www.artci.ci/images/stories/VF_CON2.DOC.pdf and related consultation material	Official regulatory consultation	Supports Orange’s strong fixed/FTTH market dominance.	It does not map every bilateral peering relationship or mobile market share nuance.	Market concentration affects whether IXP economics redistribute leverage or merely soften edge-case inefficiencies.

Unresolved intelligence questions

The public record is now strong enough to say that CIVIX matters, but not strong enough to answer several investment-grade questions cleanly. These are the issues that would most change the view if resolved.

• Is Netflix actually live at CIVIX in a traffic-bearing sense, and if so in what form? Official CIVIX reporting says Netflix was connected in 2024, but current public participant directories do not clearly show AS2906. The answer could distinguish between administrative connection, route-server peering, private handoff, or cache deployment. That difference is commercially profound.

• How much of CIVIX traffic is truly domestic eyeball-to-content traffic versus domestic infrastructure or operator backhaul flows? The live aggregate graph proves traffic, but not its composition. A 15–17 Gbps exchange carrying mostly high-value local consumer traffic has very different economics from one carrying narrower enterprise or service-network flows.

• What are the actual port, cross-connect, remote-hands, and colocation costs faced by members at each PoP? PeeringDB does not disclose CIVIX terms or service level. Without those numbers, one can infer avoided-transit economics in general but not model them precisely for Côte d’Ivoire.

• How operational is the third PoP in the Raxio facility, and who has migrated or dual-homed there? The 2025 MoU is strategically important, but the public evidence does not yet show whether the neutral-PoP strategy has already changed member behavior, cache deployment, or traffic distribution.

• Which domestic digital services are verifiably hosted in-country and reachable locally through CIVIX? The primary .ci infrastructure and some enterprise/public-service signals are visible, but the larger prize is localisation of government platforms, fintech, media, and enterprise applications. That would determine whether CIVIX is merely an operator tool or a broader digital-economy infrastructure layer.

• How would CIVIX behave in the next major international fault event? The March 2024 outage demonstrated systemic external dependence. A future event offering clearer evidence that domestic platforms and local caches remain available would materially strengthen the thesis that Abidjan is beginning to keep meaningful traffic at home.

On the evidence available today, CIVIX is no longer just a policy artifact. But the commercial question has shifted. The debate is no longer whether Côte d’Ivoire has an IXP. It is whether Abidjan can turn that IXP, plus new carrier-neutral data-centre capacity, into a denser market for local interconnection, visible caches, and in-country digital services. That is the threshold that would convert CIVIX from useful network infrastructure into a real pricing and bargaining institution for the Ivorian internet economy.