Perigon AI · Yerevan
PerigonAI
07:00 est. passengers this hour
Daily rhythm · Yeritasardakan · typical weekday
Arriving · boarding
Alighting · leaving
Train · 2–3 cars
07:00 system riders this hour
All stations combined · typical weekday
Walking in
Walking out
Trains
Estimated residential population
100 m cells · brighter = more residents
Ajapnyak district
Population density
Residents inside the 15-min walkshed
07:00 station access 
Trip origins89% inside
Trip destinations91% inside
Station-area trips
Trips leaving the Ajapnyak catchment
Origins · inside the walkshed≈35,000 / day
Destinations
Destinations · across the Hrazdan59% · ≈20,700 / DAY
Traffic passing through Halabyan street
Halabyan street
Streets its trips use
less
moretraffic
07:00 trips by hour 
Destinations of trips crossing the Hrazdan
Inside a station walkshed62.4%
Outside the walksheds37.6%
less
moretraffic on street
Bus & minibus routes running along Halabyan
Bus routes · 16
Minibus routes · 3
Halabyan street
Note how these routes tie Halabyan to Arabkir, Kentron and the north - while the south of the city sits out of reach
Halabyan trips crossing the gorge
ending inside a station walkshed49.8%
ending outside the walksheds50.2%
A · Catchment ↔ city
aligned crossings captured
+
B · Halabyan through
with a transfer
+
C · Development
4.4 ha · ≈130,000 m²
+
D · New habits
rerouted trips
=
Boardings · year 3
published forecast ≈11,700
Daily rides by station · with the new station
Flares are an illustrative sample of observed trips.
Scroll

Location Intelligence · Yerevan

Does Yerevan need a metro extension?

Yerevan's metro is modest by international standards: a single line with ten stations, carrying roughly 27 million passengers a year. The city is now planning an eleventh station across the Hrazdan gorge in Ajapnyak, at an estimated cost of $287 million.

We examined whether the underlying demand justifies the extension, using millions of anonymised trips from PerigonAI mobility data, taxi telemetry traces, granular population estimates, and observed traffic across the street network. We looked at how the metro is used today, where trips to and from the proposed catchment begin and end, and what those patterns suggest about the conditions the extension needs to make economic sense.

Yerevan population

1.09 million people · 2025 est.

Estimated residential population on a 100-metre grid, built from PerigonAI mobility data. Each point of light marks a place where people sleep; brighter means more of them.

How the city moves

The typical trip is short.

Estimations from PerigonAI mobility data show that Yerevan makes around 1.3-1.5 million motorised trips a day. The median trip covers 1.9 km of street; paid taxi rides run longer, at a median of 4.9 km, and account for roughly every sixth vehicle trip. Streets are shaded by the traffic they carry, and each flare is a sampled trip moving on its observed route.

1.3-1.5 M
motorised trips / day
1.9 km
median · all
4.9 km
median · taxi
Network trip distance
All motorised1.9 km
Taxi4.9 km
less
moretraffic on street

Trip purpose

The longest travel is to work.

Commutes stretch a median 3.8 km across the network, peak in the morning, and point at the centre. Trips to food, culture and leisure run 1.8 km and build through the evening. Everything else - errands, healthcare trips, friend visits - sits near an average of 2.0 km and spreads across the day.

Dense, repetitive, centre-bound movement is the load a metro carries best.

Median distance by purpose 07:00
less
morecommute traffic

The system · today

13.4 km of metro serving the city.

Yerevan Metro is one north-south line with ten stations and 45 wagons in service. It opened in 1981, and its deepest halls sit 20 to 50 metres down. Ridership fell to 11 million during 2020 and has climbed every year since.

10
stations
≈27 M
rides · 2025 est.
~74 k
rides / day
Annual ridership · millions (PerigonAI estimates)

The system · a working day

Every station breathes.

Passengers walk in and out of every station along the surrounding streets, paced by each station's own daily rhythm, while trains shuttle between them.

Estimated daily rides by station, 2025 · estimated from PerigonAI mobility data

The plan

1.5 km across the gorge.

The extension picks up the twin tunnels bored from Barekamutyun in the late 1980s, crosses the Hrazdan gorge on a purpose-built metro bridge and ends at a station shell abandoned in 1989, on Halabyan street in Ajapnyak.

$287 M
2026 estimate
1.5 km
new line
1
new station

107.3 bn AMD, Yerevan municipality, April 2026. Design: Metrogiprotrans.

A working day · Yeritasardakan

A day at the station.

Riders walk into the station, take the long escalator down, walk towards and gather on the island platform, and board trains on both sides - three cars at the peaks, two off-peak. Alighting passengers stream back out toward the city. Service runs 07:00-00:00 on a weekday.

Yeritasardakan station entrance at night
Yeritasardakan entrance pavilion · photo D. Galanov

The walkshed

15 minutes on foot.

The 15-minute isochrone - the area reachable in 15 minutes from the station on foot - houses roughly 26,000 residents and 45,800 jobs. Yerevan's metro has no park-and-ride, so nearly everyone arrives and leaves on foot. Our data shows it directly: 89% of the station's trip origins and 91% of its destinations fall inside this shape.

≈9 in 10

passengers using the station start or end their trip within the 15-minute walkshed.

The network

Ten stations, ten walksheds.

Overlaying the ten walksheds on the population grid and removing everything beyond a 15-minute walk leaves the city the metro actually serves.

250 k
residents ≤15 min
217 k
jobs ≤15 min
23%
of the city's people

The other side of the gorge

33,600 people, 15 minutes.

The proposed station's own walkshed holds 33,600 residents at about 19,700 per square kilometre, plus 14,500 jobs. Tumo, Haybusak University and the Republican Hospital all sit nearby. Put simply, one in three residents of Ajapnyak district could walk to this station in under 15 minutes. That is a bigger walk-in population than half of the existing stations can claim, and today these people have no metro at all.

33,610
residents
19,700
per km²
14,500
jobs

Demand · 01

Where trips begin.

Each magenta point on the map to the right is the start of an observed trip inside the 15-minute catchment of the proposed station on an average day. About 30-35,000 motorised trips leave from and arrive at the catchment daily. Together they make a total of about 65,000 trips a day to or from this catchment.

Demand · 02

Where trips end.

More than half of the ~35,000 outbound trips - about 59% or 20,700 trips - cross the Hrazdan toward the city, over one of its bridges. Their destinations concentrate along the northern arterials, the centre and the station squares, and thin out gradually across the rest of the city.

Demand · 03

The demand is already on the map.

The metro's walksheds split those destinations in two.

62.4%≈12,900 trips a day

of the roughly 20,700 daily trips that leave the catchment and cross the Hrazdan end within a 15-minute walk of an existing station - about 12,900 boardings a day already going where the metro goes, before a single rail is laid.

Origin-destination alignment with station catchments is the primary siting signal in transit forecasting, and the major appraisal standards are built on it. The US federal STOPS model forecasts new-station ridership from existing journey-to-work flows between station areas, and Germany's Standardisierte Bewertung, mandatory for federal co-funding of rail projects, rests on the same comparison of travel demand with and without the line. Californian studies found residents near rail stations commuting by transit up to five times as often as their city average, and workers in offices near stations riding at nearly three times the rate of workers away from them. Dutch research adds the distance decay - rail use is about a fifth lower for homes 500 to 1,000 metres from a station, and up to half lower beyond 3.5 km.

The corridor

Halabyan is the north-west's main artery.

One street ties Ajapnyak and the districts behind it to the rest of Yerevan. About 52,300 vehicle trips a day pass through Halabyan, with every second trip between Ajapnyak district and the rest of the city passing through it. In people, that is roughly 68,000 person-trips a day - about 34,000 in each direction.

52.3 k
veh / day
51%
of district crossings
2,070
veh/h at 18:00

The corridor · load

Evening peak: at 80–95% of capacity.

The evening peak runs mainly away from the city towards Ajapnyak. At 18:00, about 2,070 vehicles an hour cross toward Ajapnyak, right at the corridor's effective capacity of roughly 2,200 for three lanes with curb parking. A single stalled car breaks the corridor, and there is no parallel high-capacity route.

08:00 Ajapnyak-bound trips / hour
capacity ≈ 2,200
Trip ends near an existing station
Trip ends elsewhere

The corridor · destinations

Half the crossing traffic ends at the metro's walksheds.

About 32,600 person-trips head along the Halabyan corridor toward the city each day. Four in ten of them start inside the Ajapnyak walkshed and are already counted in the catchment itself. The other 61%, some 19,900 trips, begin beyond it and are only passing through. Of those, 55.9% cross the Hrazdan, and 49.8% of the crossings, roughly 5,500 trips, end within a 15-minute walk of an existing station - every green flare on the map. Their destinations accumulate as they land: green inside the walksheds, magenta outside.

These travellers already go where the metro goes. A station on the Ajapnyak side would let them make the whole journey on one train, without the bridge.

Part II

The demand exists.
What about the economics?

From flows to riders

What the new station can capture.

A · The catchment itself
Daily trips departing the catchment - the station logs boardings at entry≈35,000
… of which cross the Hrazdan toward the city - about 59%≈20,700
… of which end within an existing station walkshed - 62.4%≈12,900
Capture: 17-28% of these aligned trips, 5-7% of the other crossings2,585 - 4,160
B · Halabyan through-traffic · city-bound
Person-trips through Halabyan toward the city that do not start inside the walkshed - 61% of the city-bound flow, so none of it counted in A≈19,900
… of which cross the Hrazdan - 55.9%≈11,100
… of which end within an existing station walkshed - 49.8%≈5,500
Capture with a transfer: 9-17% of these aligned trips, 3-6% of the other crossings665 - 1,275
C · New development on site
4.4 ha at the shell - a preliminary figure≈130,000 m²
≈1,000 homes (≈2,800 residents) plus ≈2,500 daily workers - each boards here about once a day≈5,300 people
Boardings at the 25-40% metro share typical of station-top housing1,300 - 2,100
D · Feeders and new habits
The wider district beyond walking distance - 110,000 residents, feeder buses on the free transfer, drop-offs, Davtashen reroute≈7,500 candidate departures
Feeder-dependent capture (3-8%)225 - 600
Boardings a day by year three
A + B + C + D4,775 - 8,135
Published forecasts sit around 11,700 daily riders; our estimate confirms solid demand while staying deliberately more conservative

Capture rates are calibrated on PerigonAI station-catchment work in cities of comparable size and transit mix.

The competition

The first riders might come off the buses.

The capture estimate we provided is a demand ceiling, not a traffic forecast. 16 bus routes and 3 minibus routes run along Halabyan street, visualized in the map to the right.

Of the roughly 12,900 outbound trips a day that connect the catchment with existing station walksheds - the trips that would board at the new station - around four in ten, some 4,500-5,800, already ride these services. Those are the first boardings the metro would capture, which means part of its ridership will come from cannibalising the buses rather than emptying the road. If most new riders shift from transit rather than from cars, the station reaches its ridership range while congestion barely moves.

This analysis is not a substitute for a traffic forecast model or a traffic impact assessment. We used PerigonAI mobility data, traffic data, our modelling engines and our experience with such projects to lay out a simplified, grounded case for the extension's feasibility and the reasoning behind it.

Economics

$287 million against $1.3-2.2 million a year.

The public cost estimate has climbed from about $35 million in 2020 to $287 million in 2026. As expected - this is true of nearly every metro in the world - fares cannot repay it: 4,775-8,135 new daily boardings at a fare of roughly $0.75 gross $1.3-2.2 million a year, before operating costs such as continuous groundwater pumping. On fares alone, repayment would take more than a century. Even if fares double or triple at some point, farebox recovery stays unviable - at three times today's fare the station grosses $3.9-6.6 million a year against the $287 million build.

A standalone farebox case is commercially unviable. That does not kill the extension - it moves the weight of the decision onto how the project is owned and financed.

Economics · the balance sheet

The station must own its skyline.

The $287 million buys the tunnel, the bridge and the station - nothing above ground. Kept alone on the city's books, it is a permanent subsidy. The 4.4-hectare site around the station shell changes that calculus: it can carry roughly 130,000 m² of homes, retail and offices - about $110 million to build, around $200 million of value at post-metro prices, and 1,300-2,100 daily boardings of its own.

Ajapnyak housing trades at about half of Kentron's level. Rail narrows that gap, and the walkshed's existing stock alone stands to gain $40-80 million. Hong Kong's MTR, Copenhagen's Ørestad and many other successful Transit-Oriented Development projects financed their lines this way: rail, station, land and development on one balance sheet. Each piece feeds the others - the line delivers riders and footfall to the shops and offices, the development delivers passengers and land value to the line, and the rising land value pays down the tunnel. Where these components have sat on separate balance sheets, history shows the owners drifting into conflicting business and political interests - service cut where footfall was needed, land sold before the uplift arrived - and decisions that break exactly this feedback loop. Kept together, they are one investable project.

Verdict

The demand is there. What decides the outcome is the financing model.

The case for the station rests on things we can already observe. A third of Ajapnyak lives within a 15-minute walk from the proposed station. The district's one road link to the city runs at its limit every evening. And half the trips that cross the gorge end in places the metro already reaches.

Our estimate is 4,775-8,135 new daily boardings by year three, a 6-11% lift for the whole system. That is a good result, though well short of the announced +25% headline. The demand is clearly there. Therefore, whether the economics of the project work depends less on the riders than on the ownership structure: built as one land-and-rail project, the station will pay for a meaningful share of itself. Built as a tunnel and station alone, the city will pay for it twice - once in concrete, and once in land value handed to others.

PerigonAI

Infrastructure decisions require street-level evidence

The how and the why

PerigonAI combines anonymised mobility data, credit card spend, satellite imagery, property records and field observation into one picture of how a place actually works - who moves through it, what they do there, and what that is worth.

Mobility datawhere people actually go, hour by hour
Satellite & aerial imagerythe built fabric, roof by roof
Property & land datawhat space actually trades for
Traffic & infrastructure datahow the network carries its load

This story is one corridor in one city, and it carries a general lesson: transit, land and development decisions are the same decision. We run this analysis - trade areas, footfall, site selection, demand forecasting, cannibalisation, value capture - for governments, developers, retailers and operators deciding what to build, and where.

Reach out to learn more →