Portfolio update: Portals Is Routing +$1B and just announced Base Launch

Every so often a portfolio company does something that only makes sense in hindsight.

Portals.fi spent this year rebuilding infrastructure nobody was asking them to rebuild. In September it turned into a day one launch partnership with Base. This is a short note on what happened and why we think it mattered.

What Portals.fi does

DeFi has a discovery and execution problem. There are millions of yield-bearing positions across dozens of chains. Finding the good ones is hard. Getting into them is harder, because it usually means several transactions across several interfaces, each one a chance to lose money or give up.

Portals collapses that into one action. Hold any token, on any supported chain, and enter any position in a single transaction. They call it a zap.

Underneath sits aggregation infrastructure they built themselves rather than licensed: 480 distinct protocols across 14 networks, 30M+ discoverable assets, and a proprietary historical dataset now past 16 billion rows.

That infrastructure runs two ways. There is a consumer product, the Portals Explorer, where people find and enter positions directly. And there is an API, used by more than 1,500 builders, serving 60M+ calls a month. Same engine, two front doors.

They have routed over $1B in cumulative volume.

The thing that changed: a smart order router of their own

For most of their life, Portals did one thing it did not control. Swaps routed through third party aggregators.

That is the normal arrangement, and it works until it does not. Quotes come back that revert on chain, and your partners ask you why. Coverage has gaps. And some asset classes are gated entirely by the compliance posture of whoever sits underneath you, which means you cannot execute them no matter how good your own stack is.

So Portals rebuilt their DEX aggregation as a genuine smart order router. It assembles its own routes and splits orders across venues, rather than asking someone else and forwarding the answer.

This is worth being precise about, because the aggregator landscape is less competitive than it appears. Several well known meta aggregators do not assemble routes at all. They wrap the same small set of underlying sources. Query five of them and you are frequently querying one source five times through different intermediaries, which is why their quotes cluster within a few basis points of each other.

Portals is now one of a small number of teams doing independent routing, and in their own testing they are competitive with the leading swap aggregators on pure swaps, with zaps and yield routing on top.

They rebuilt cross-chain in the same stretch, on Relay, with no new contracts. Roughly 70 networks of coverage. The flow now runs end to end: exit a position on one chain, swap, bridge, enter a position on another, as one intent.

There is no longer any part of the execution path that Portals rents.

Why the timing mattered

In September, Base launched tokenised equities. Portals was a day one, week one launch partner, with a dedicated campaign page live from the start.

Those two facts connect. The third party router Portals used to depend on would not permit tokenised stock execution. Owning the router is what made the Base partnership technically possible. Infrastructure work that looked like housekeeping in June turned into a distribution deal in September.

That is usually how it goes, and it is usually invisible from outside.

Nine people

Portals is a team of nine.

Nine people running their own data infrastructure across 14 networks, their own pricing engine, their own transaction simulator, their own router, and a consumer product with a 16.1% weekly retention rate against a fintech average of 11% to 12%. Average session time is 14 minutes. Bounce rate is 10%.

We pay attention to output per head, and this is among the best we have seen in the portfolio. It comes from a deliberately senior team, heavy internal tooling, and a refusal to add headcount to problems that can be automated instead.

Protocol onboarding is the clearest example. New vaults and yield sources appear on Portals faster than teams many times the size manage, because that pipeline runs without a developer in the loop.

A distribution channel that did not exist last year

One more thing worth noting, because we did not expect it.

Portals has collapsed their API so that a single call returns approval status, allowance, the approval transaction, a fully simulated transaction, and all the call data required to execute. Three calls became one. No API key is required. Their transaction simulator, Foresight, ships with x402 support and an interface designed to be read by software rather than a person.

None of that was built as an AI play. It was built because it is a better API.

The side effect is that Portals is unusually cheap for an autonomous agent to call, and the referral logs have started to reflect it. Traffic is arriving from Manus and ChatGPT without anyone at Portals having pursued it.

Most on-chain infrastructure was designed for a human clicking a button and retrofitted for programmatic access afterwards. Portals arrived at the opposite position by simply making the developer experience good. If more on-chain activity ends up being initiated by software acting for users, that is a useful place to be standing.

What comes next for Portals.fi

Nine months of work on a new consumer app is close to release. We have seen parts of it. It is the reason the execution layer was rebuilt first, and it is the thing to watch over the next quarter.

Portals spent three years building infrastructure they own outright, through a market that did not reward it. That phase is finished. The interesting question now is distribution, which is a different problem and in some ways a harder one.

We are glad to be along for it.

Portals.fi: explorer.portals.fi · foresight.portals.fi · blog.portals.fi