The average Indian carrier runs through six shared metro hub belts

Network map

Schematic: cities sit at their real latitude and longitude, links are straight lines, not road alignments. Hover a line or dot for its name.

National gateways

The shared hub belts. Most big carriers run a mega gateway in each, so congestion here spreads to every carrier at once. These six are the only cities where the model re-sorts parcels on the way through.

Regional and secondary hubs

Who moves India's parcels (FY26)

India moves about 7.4 billion parcels (range 6.1–8.9), or 20 million a day.

Delhivery: FY26 express parcels, including Ecom Express^[33]. Blue Dart: FY26 domestic shipments; includes business documents and B2B parcels^[36]. India Post: FY26 domestic parcels (74.9 million); letters excluded^[38]. Valmo (Meesho): 695 million in Apr–Sep 2025, doubled. Only the orders Valmo's own partner network carries; Meesho orders handed to Delhivery, Shadowfax or Xpressbees are in their counts^[32]. Shadowfax: 726 million FY26 orders less about 140 million same-city (hyperlocal) orders, which never reach a hub^[34]. Xpressbees: About 1.2 million a day in early 2025 (press report), times 365. The company publishes no count^[35]. DTDC: Company figure of about 170 million a year, undated^[37]. Amazon, Ekart and smaller carriers: Not published. What is left of the market after the networks that publish a count; the check against the broker's shares is above^[31][32].

Where the sellers are

Meesho's prospectus is the only public count of sellers by state. Four states hold 54% of its sellers: Uttar Pradesh 15.9%, Gujarat 15.7%, Delhi 13.8% and Rajasthan 8.9%^[41]. Most Uttar Pradesh sellers sit in the NCR districts (Gautam Buddha Nagar, Ghaziabad, Meerut)^[43], and Flipkart names Delhi and Surat as its two biggest sources of new sellers^[42]. We spread the remaining sellers by state GDP. Seller pickups by hub, as a share of all seller pickups: Delhi NCR 26%, Surat 9%, Jaipur 9%, Ahmedabad 5%, Mumbai 4%, Bengaluru 3%.

Estimated parcels sorted at each hub city

The fifteen busiest hub cities, all networks together. Pickups: parcels that start in the hub's area, from sellers or warehouses. Transhipment: parcels from elsewhere, re-sorted here on the way to somewhere else. Deliveries: parcels for people in the area.

B2B goods by state (estimated)

Every consignment of goods worth more than ₹50,000 needs an e-way bill, so the GST network's count is the best public measure of B2B freight. In FY26 there were 524 million inter-state e-way bills, about 1.4 million consignments a day, and 1,023 million within states^[58]. Most parcels are below the threshold, so these barely overlap with the parcel counts above. B2B sources are not where parcel sellers are: Haryana sends the most, while Uttar Pradesh and Delhi lead parcel sellers.

Goods don't flow evenly, and trucks feel it. In FY26 Bihar took in 5.7 consignments from other states for every one it sent, Jammu and Kashmir 5.6, Kerala 3.9 and the Northeast states 8 to 22^[58]. Gujarat and Haryana sent out about twice what they took in. A truck that delivers into a red state has less to bring back, so the trip in likely costs more. The map only counts consignments worth over ₹50,000, so sellers' parcels aren't in it. Rajasthan looks even at 1.05, yet it has 8.9% of Meesho's sellers, the fourth most of any state^[41].

One marketplace's rates lean the same way. On FR8, a 32-foot multi-axle truck costs ₹45,500 from Kolkata to Patna and ₹23,500 back^[59]. Of the 35 routes it lists both ways where one state's e-way bill balance is 1.6 times the other's or more, 28 (80%) are dearer going into the state that takes in more. Delhi, Gurgaon and Noida are one market split across three states, so read them as one.

#Containers by port (rail vs road)

Containers are where exim meets the inland network. Gateway ports handled about 74 thousand containers (TEU) a day from March to June 2026^[60]; Mundra and JNPA alone take 65%. As a check, JNPA's own FY26 total was 8.17 million^[61], close to this rate. Vizhinjam (about 3.8 thousand a day) is left out because its boxes go ship to ship. Only three ports publish how many containers leave by rail; for them the table splits rail from road. Rail containers stay on rail.

The port volumes come from terminal counts; the rail shares are what each port or CONCOR reported, for the period shown^[62][63][63]. Mundra's and Pipavav's are older than JNPA's.

Cargo (Mn tonnes) by port, September 2026 (Ministry of Ports)

Everything each port handled in the month, containers and bulk together, in million tonnes, against September 2025^[66][67]. State ports are reported by state, so Mundra, Hazira and Sikka are one Gujarat row. Updated monthly (about four weeks after the month ends).

Air cargo by airport, August 2026 (AAI)

Freight each airport handled in the month, international and domestic, in tonnes, against August 2025^[68]. All airports together: 369,878 tonnes, +11% on last year. Updated monthly (AAI posts about five weeks after the month ends). Airports under 1,000 tonnes a month are left out.

Truck traffic at toll plazas, August 2026 (IHMCL FASTag)

B2B inter-state lanes (estimated)

Empty running. A truck counted at a plaza may be empty. Where goods flow mostly one way, trucks come back light: if a link carries 97% of its goods one way and trucks return the way they came, at least 48% of its passages run empty (Guwahati – Silchar). That is arithmetic on the estimated direction, not a measurement; nothing public measures empty trucks.

Not shown on the map: links with no NHAI toll plaza, such as the Mumbai–Pune expressway, which is run by the state. B2B is not yet part of the choke points below.

Load by month: B2B and parcels together

Parcels a day: estimated, the long-distance parcels routed over the leg on an average day, all networks, both directions (see "Choke points for parcels"). Heavy trucks a day: measured, the busiest plaza on the leg. The two are not added: a truck is a truck, a parcel is one item in it, and the plazas already count the trucks that carry parcels. Parcels by hub follow the same quarters: no public source gives a hub's volume by month.

Choke points for parcels

Every long-distance parcel flow in the volume model above is routed along the shortest road on the map. Each link is as wide as the share of parcel flow crossing it. Ringed hubs are choke points; squares are ports. Then each hub and link is closed in turn to see what is cut off and how far the rest must detour. B2B goods and port containers are not included, because no public source says which roads they use (see above). Confidence: Low to Medium.

Parcel flow: share of long-distance parcel flow that crosses the hub or link (modelled). Heavy trucks a day: measured at the busiest toll plaza on the link, a floor for its truck traffic; blank for hubs, and "no plaza" where none on the link's highway could be placed. Most exposed zone: share of that zone's long-distance parcel flow (in and out) that depends on it. A long way round means the next route on this schematic map is more than 300 km longer. The map leaves out smaller parallel roads, so this shows how much the network leans on a corridor, not the route trucks would actually take.

Trunk lanes

Northeast, seasonal and rail links

What this cannot tell you

• No dated public source names facility locations for Xpressbees, so they add nothing to the scores.
• India Post reports 188 parcel hubs but publishes no city list we could verify. Only its Thane sorting hub is counted.
• Coverage is uneven. Amazon and Flipkart announce openings; others rarely do. Cities such as Indore and Vijayawada are probably under-scored.
• Some sources are old (Blue Dart's Bhiwandi superhub dates from 2003). A facility may have moved or closed.
• B2B lanes are estimated. Plaza counts check how much moves on a link, not which states it comes from or which way it goes. E-way bills also cover goods sent by rail, and one bill can be a carton or a full truck. Short trips near the plazas add trucks the model cannot see.
• Hub volumes are modelled, not measured. Seller locations rest on one marketplace (Meesho); Amazon and Flipkart sellers may sit elsewhere. Mumbai's warehouse pickups are probably too high, because its warehouse stock includes much that is not parcels.
• Transhipment assumes road routing through national gateways only. Direct trucks between big seller towns and distant gateways, air freight, and carriers' own regional hubs (for example at Nagpur or Lucknow) would move some of it elsewhere.
• Blue Dart, DTDC and India Post carry business parcels as well as e-commerce, so they sit slightly outside the e-commerce market total. That makes the Amazon and Ekart remainder slightly low.