Why we built Mela

Two scenes from the same evening, in the same person’s day.

7 a.m., DC. A nurse on her way to a shift sends $400 to her mother in Awassa from a Remitly account. The rate looks fine. The fee feels invisible until you do the math.

7 p.m., DC. The same nurse, now off shift, on U Street, ordering injera and tibs at her favorite Habesha restaurant. She pays with her debit card. She doesn’t know — and the owner doesn’t always say — that roughly three cents of every dollar she just spent went to Visa and the processor before it ever hit the till.

Same person. Same evening. Two transactions. Both leak money out of the community.

That is the problem we built Mela to fix.

The diaspora financial loop

There is a closed loop most people don’t talk about as one thing.

The same A1 — the dollar-paid Habesha-American living in DC, Atlanta, LA, Seattle, Minneapolis — sends remittance home, and eats at the Habesha restaurant on Friday, and shops at the Habesha grocery on Saturday, and gets her hair done at the Habesha salon every other week. The same person is the sender on the remittance side and the customer on the merchant side. The community already runs as an economy. We just don’t run it on a rail we own.

That is the central observation behind everything Mela does.

Right now, every step of that loop pays a tax to a network the community has no equity in. The remittance side pays it as an FX spread. The spending side pays it as card-network MDR. Different mechanism, same consequence: dollars leak out of the corridor, out of the strip-mall lease, out of the salon chair, into card-issuer revenue, network assessments, and processor margins.

If you only fix one side of the loop, you only stop one of the leaks. You still pay the other one twelve times a month.

The remittance leak

Start with the side most people already think about.

A typical sender moves $200–800 a month home. The serious senders move more around holidays and family events. The corridor as a whole is roughly $5–7B/year, recovering since 2023. Take the median sender at $500 a month, and assume an effective spread (rate + fee + card-funding cost) of around 5%. That’s $300 a year, per sender, that doesn’t reach Awassa, Bahir Dar, or Hawassa.

Multiply by the addressable US Ethiopian-American base — somewhere between 250,000 and 500,000 senders — and the cumulative number is large enough to be its own business model. It already is somebody’s business model. Just not ours.

The good news is the levers to close this leak are well understood. ACH-pulled USD into the wallet (no card swipe spread). Plaid for instant US-bank linking. A corridor-specific FX book priced for Ethiopia rather than as a line item in a generalist’s pricing matrix. Three things we shipped in Q1 2026, and three things that change the math from “5% leak you can’t avoid” to “compete on the actual rate.”

But — and this is the part most diaspora-fintech pitches stop at — fixing remittance only solves one corner of the loop.

The merchant leak

Here is what most diaspora customers don’t know.

The Habesha restaurant where she ate on U Street pays its processor every time she swipes a card. The bill that landed at the table for $58.40 — by the time the kitchen sees it the next morning, it’s $56.74. The networks took $1.66 of that meal.

Now scale it.

The MDR — merchant discount rate, the all-in card-acceptance cost — for US small-business card volume runs about 2.5–3%. On a busy Habesha restaurant in DC doing $200,000 in monthly card volume, that is roughly $5,000 per month, or about $60,000 a year, paid to networks the owner-operator does not own equity in. The interchange piece (the largest slice, ~1.5–2%) goes to the customer’s card-issuing bank — a bank that has nothing to do with the diaspora. The assessment goes to Visa or Mastercard. The processor margin goes to Square, Toast, Stripe, or whoever sold them the terminal.

For many Habesha owner-operators in DC, Atlanta, and LA, that monthly card-fee number is bigger than the rent on the building they’re standing in. We have heard owners read off both lines on the same statement and pause at how close the numbers are.

That is not a hypothetical. Walk into a Habesha restaurant on U Street, a Habesha grocery in Silver Spring, a Habesha café in Atlanta, a Habesha salon in Seattle. Ask the owner what they pay in card fees. Most know the number to the dollar. They have to.

This is the leak the diaspora customer rarely sees. And it doesn’t get smaller because the corridor is friendly. It gets smaller — at all — only when there’s an alternative rail.

Why a feature on top of someone else’s app couldn’t do this

Once you see the loop as one thing, the strategic conclusion follows.

You cannot plug both leaks with a “Habesha mode” inside Remitly, or a partnership tab inside Western Union, or a clone of Ria. Those products own a transaction surface. They don’t own the rail underneath. To fix the merchant leak, you need a wallet on the consumer side, a USD-settling QR rail on the merchant side, and a corridor that connects to the Ethiopian birr leg without re-importing card-network costs.

That’s not a feature. That’s the product.

Three things had to be ours, and built end-to-end, for any of this to work:

1. The bank account. USD wallet, ACH-pulled, identity-verified, in our own ledger.
2. The card. Spendable anywhere Visa is accepted today, with the long-run goal of spending Mela-to-Mela without re-entering card rails.
3. The merchant terminal. A QR-payment rail (mela-merchant) where the customer’s Mela wallet settles direct to the merchant’s USD wallet — no interchange, no assessment, no processor margin.

Three different products. One ledger. One company. Owned by us.

What “owned by us” actually means

Mela is US-incorporated, headquartered in Dover, Delaware. We run our own double-entry ledger. Our corridor relationships in Ethiopia are direct, not licensed. Our team is Ethiopian-led. We are not white-labeling someone else’s bank. We are building the rail.

We are also opening our first round of funding to the Ethiopian community. The intent is to make community ownership a literal claim, not a rhetorical one. The round structure is being finalized with counsel; until that’s settled, the page exists for interest registration only.

The “first Ethiopian-owned banking” line on the home page is a careful claim, not a marketing claim. We say banking, not bank — Mela is not a chartered bank in any jurisdiction, and we are not interested in the regulatory or rhetorical risk of pretending otherwise. Banking-experience-wise, in the United States, end-to-end, owned by Ethiopians, for Ethiopians worldwide — to our knowledge, we are the first.

What we are deliberately not yet

The credibility of any “we are building X” claim is in what you say no to as much as what you say yes to.

• We are not a chartered bank. We say banking and banking experience throughout, on purpose.
• We are not making FDIC claims directly. Pass-through deposit insurance lives at our partner bank, has conditions, and must be cited correctly — a sentence we will write once that’s done with the partner-bank’s compliance team.
• We are not yet open to consumers in Ethiopia. The Ethiopia consumer + Ethiopian merchant side of the network is a 2027+ scope. The 2026 scope is US-only. The Ethiopian user base we acquired during the 2025 freelancer phase is real and we are still serving it; we are not leading new marketing with that audience.
• We are not pretending Ekub has shipped. The home page surfaces it as coming soon. Rotating savings circles are on the roadmap, not in the app.

If any of those change, this page will change with them.

The same evening, on a rail we own

Back to the nurse, on the same evening, in the same DC.

7 a.m. The same $400 to her mother. On Mela, more birr arrives in Awassa for the same dollar — illustrative numbers as of today are roughly +ETB 1,350 vs Remitly on a $500 send, but the operative quote is whatever the live comparator shows in the app at send-time.

7 p.m. The same dinner on U Street. On a Mela QR, the kitchen sees the full $58.40 instead of $56.74, because there is no interchange, no assessment, no processor margin between her wallet and the till.

Same person. Same evening. Same dollars. The difference is whose rail those dollars run on.

“Your grandmother knew how to bank long before banks did.”

We are building the software version of that, with the legal status and the rails that make it work in 2026. የእኛ banking።