A map of the categories we think work from here, the ones we avoid, and the reasoning behind both. We would rather be specific and wrong than broad and unfalsifiable.
The stack
Four layers, different economics.
Most local capital concentrates on the consumer layer. We think the durable companies are further down.
Where value accrues
Layer
Example
Our view
Infrastructure and tooling
Developer tools, APIs, data infrastructure
Strongest fit — global buyer from day one
Vertical software
Workflow tools for a named foreign industry
Strong fit — if the buyer is reachable remotely
Applied AI
A model wrapped around one costly workflow
Selective — the workflow must be specific
Local consumer
Apps monetised by local advertising
Poor fit — rupee revenue, capital intensive
Infrastructure
Vertical
Applied AI
Consumer
Fundable pockets
Six places we expect to write cheques.
Each one shares the same shape: engineering-heavy, remotely sellable, and priced in a currency that holds its value.
Developer tooling
Testing, observability, build and deploy infrastructure. The buyer is an engineer who evaluates the product rather than the postcode.
Back-office automation
The unglamorous processes inside foreign mid-market companies that are still run on spreadsheets and email.
B2B fintech infrastructure
Reconciliation, compliance workflow and payment operations software — not lending, not deposits, not a licence we do not hold.
Vertical SaaS for export industries
Textiles, surgical instruments, sports goods, freight. Sectors where Pakistani teams have genuine domain knowledge and foreign counterparties.
Applied AI on a named workflow
Document handling, quality inspection, support deflection. The value has to be measurable in hours or errors removed.
Data and integration services
Pipelines and connectors sold as a product, built by the same engineers who spent a decade building them as a service.
White space
The gap is the first cheque.
Pakistan has accelerators that give small amounts and mentorship, and it has regional funds that arrive at Series A with an international metrics bar. Between the two sits the institutional pre-seed round, and it is thinly served.
That gap is why strong teams here often raise from an angel syndicate on inconsistent terms, end up with a cap table no institutional investor will touch, and discover the problem eighteen months later when it is expensive to fix.
Friends, family, angels Served
Accelerator cheque, mentorship Served
First institutional round Thin
Structuring for foreign capital Thin
Series A and beyond Served, from outside
Sizing note
An honest word about the numbers.
We do not publish a total addressable market for Pakistani software, because any figure we produced would be a projection dressed as a fact. Reported IT export revenue is a real and useful number, but it measures services, which is precisely the thing our thesis argues against.
The number that actually determines whether this fund works is much smaller and much harder to observe: how many teams here start a genuinely product-shaped company each year, with a foreign buyer and the will to sell to them. We believe it is enough for a concentrated first fund and not enough for a large one. That belief is the single most important assumption we hold, and it is the one we are most willing to be wrong about in public.
Building in one of these pockets?
Or in one we have missed. If you think the map is wrong, that is a conversation we want to have.
Infrastructure
Vertical
Applied AI
Consumer
