Shopify Bets on Outsiders

If You Only Read One Thing

The shopper looks like a customer to Shopify and a competitor to Amazon. Their split over Meta’s Muse shopping agent reveals which part of a transaction each company needs to own. The EU–Philippines trade breakthrough approaches commerce from another direction: make alternative customers easier to reach. Both developments change the bargaining power that comes from having somewhere else to sell.

Shopify Can Profit Without the Visit

Shopify can profit from an AI shopper without owning discovery. That distinction matters after Meta’s 11% Monday rally: enthusiasm for Muse does not establish who earns money on its purchases. Shopify sells transaction processing; Amazon also monetizes the shopping visit.

Amazon has blocked Meta’s Muse from shopping on its site. In a statement reproduced by GeekWire, Amazon says the agent did not identify itself and appeared to store customer credentials. On September 21, Shopify CEO Tobi Lütke announced a partnership to enable Muse checkout through Shop Pay across Shopify stores. That is an integration commitment; it does not establish that every merchant can accept those orders today.

The distinction is customer acquisition versus transaction processing. Think of a shop that pays someone to bring a buyer through the door, then pays a separate company to handle the card payment. An agent can replace the first interaction while preserving the second. Shopify’s checkout documentation describes precisely that separation: merchants can add Shop Pay without moving their existing commerce platform to Shopify.

Amazon has more to protect when discovery moves elsewhere. An outside assistant can compare sellers before the buyer reaches Amazon, weakening the retailer’s influence over product selection and sponsored placement. Shopify can gain transactions across its merchants without first persuading shoppers to browse a Shopify destination. The same referral therefore carries different opportunity costs.

Security is a substantial counterargument, not an excuse to dismiss. A merchant must know who submitted an order and how payment authority was obtained. Meta’s launch materials describe protected credentials and approval before purchases; Amazon disputes the acceptability of the access arrangement. A buyer’s consent does not resolve every operational disagreement between the services handling the order.

The investment read favors Shopify’s payment infrastructure over an assumption that a standalone shopping assistant captures all the margin. Over the next one to two years, I have medium confidence that negotiated agent channels can expand Shopify’s addressable transactions. The read fails if agent orders mostly replace existing Shop Pay purchases, or if the cost of handling disputes absorbs the incremental profit.

September 9’s Muse launch introduced the subscription relationship. Today adds a merchant-side bargain: distribution earns money only where transactions can complete. The concrete test is whether Shopify’s next merchant documentation specifies live Muse eligibility, transaction fees and responsibility for disputed orders.

Europe Gives Manila Another Customer

The EU–Philippines trade breakthrough makes diversification more credible without delivering an immediate tariff windfall. Its prospective value is a more dependable route to a second large customer base.

Reuters reports that Commission President Ursula von der Leyen said on September 22 that she and President Ferdinand Marcos Jr. had agreed on a free-trade deal. That announcement does not establish an effective date or the tariff treatment of individual products. Those details determine when political agreement becomes commercial access. Treating the headline as duty-free trade already in operation would turn a political milestone into imaginary earnings.

The starting point is meaningful trade with existing preferences. In its June WTO statement, the EU put bilateral trade at €17.6 billion in 2025, up 4.5%. It described the eventual agreement as replacing GSP+, the arrangement under which the Philippines already receives preferential access. The commercial case must therefore measure improvement over current treatment, rather than assume every shipment currently pays the full tariff.

A credible alternative customer changes bargaining before it changes sales. Imagine a manufacturer dependent on one large buyer: another qualified buyer makes a threatened price cut less decisive. A trade agreement can help create that option by making market access more predictable. The useful concept is an outside option, a believable place to take the next order.

That is where today’s two stories meet. Shopify wants merchants to receive orders through another discovery channel; Manila wants exporters to have another durable commercial relationship. Neither gains much from a nominal alternative that cannot handle actual transactions.

The strongest objection comes from Europe’s own assessment. The WTO statement identifies regulatory complexity, logistics constraints and restrictions on foreign participation. Tariff preferences cannot repair a late shipment or make an unqualified product meet a buyer’s standards. Established exporters may consequently benefit before new factories do.

Over a two-to-three-year horizon, my medium-confidence investment read favors Philippine exporters already meeting European requirements and the logistics businesses serving them. Announced market access should not justify a blanket premium for domestic assets. The read weakens if final exclusions preserve today’s costs or implementation fails to reduce border friction.

This extends the standing pattern of policy shaping supply chains, with a useful reversal: governments can create value by widening eligible markets as well as excluding rivals. The decisive next document is the final tariff-and-origin schedule, showing which products qualify and how much local production they require.

The Contrarian Take

Everyone says: Amazon is resisting the future while Shopify embraces AI shopping. Openness will decide the winner.

Here’s why that’s incomplete: The valuable distinction is which part of the transaction each company can afford to give away. Shopify can accept an outside recommendation engine if its merchants and payment service retain the sale. Amazon has stronger reasons to negotiate because an outside agent can substitute for its shopping interface. Neither position requires a different view of AI’s eventual capability.

A negotiated integration can also be more open commercially than unrestricted browser access. It can tell merchants how to recognize an order, how payment was authorized and who handles a complaint. The cost of that agreement is real, but so is the cost of an order that cannot be fulfilled or defended. The winner could be the service that standardizes those obligations across many assistants, even if consumers never learn its name.

Under the Radar

Muse’s local flaw puts a price on connected access. Patrick Wardle’s demonstration shows how local code could redirect dictation and capture authentication material. Meta says it issued a hotfix; this was not a remote break-in. The incident extends September 18’s connected-account risk to a consumer assistant: one compromised app can expose additional services. It establishes a vulnerability, not evidence that customer accounts were abused.
Asia’s diversification still includes China. ASEAN’s new ministerial statement puts 2025 investment inflows from China, Japan and South Korea at $44.8 billion, or 18.2% of ASEAN’s total. That regional investment base matters alongside Europe’s Philippine agreement: another trade relationship can supplement existing production networks. It does not establish that factories, financing or intermediate inputs have stopped depending on them.

Quick Takes

OpenAI wants common standards without a global licensing gate

OpenAI’s September 21 proposal calls for international measures of automated research and incident severity, while explicitly excluding mandatory prerelease approval from the standards themselves.

Beyond yesterday’s incident-warning dialogue, the new move is an institutional design: common measurements, with governments deciding legal force. Labs able to shape those measurements gain influence, but no international enforcement mechanism has been agreed. (Source)

Oil relief has a physical component

Brent fell 3.4% on Monday to around $100 as Saudi exports through Hormuz increased and hopes for US–Iran talks improved, Bloomberg reported.

That adds physical supply to September 12’s pipeline disruption. My medium-confidence near-term read favors energy buyers as scarcity eases. It fails if renewed disruption removes those barrels: shifting exports back through Hormuz restores volume without creating a safer route. (Source)

The press pool withdraws a shared service

The five television networks suspended presidential pool coverage after CNN’s exclusion. AP reports that the networks communicated the suspension to subscribers on Monday.

This materially advances September 20’s press-ban story: the response now includes collective withdrawal of production capacity, alongside litigation. Sharing cameras makes coverage economical; withholding that shared service raises the administration’s cost of replacing independent footage. (Source)

The summit tests the tariff truce

Trump and Xi are expected to meet Thursday, with trade and AI on the agenda. The new test after yesterday’s incident-channel proposal is whether leaders turn dialogue into a more durable commercial arrangement. A warning channel could reduce accident risk while leaving tariffs intact. Exporters gain planning certainty only from commitments that change costs or extend access, not from a successful meeting alone. (Source)

The Thread

An alternative changes bargaining power only when the party using it can complete the business. For a merchant, that means an agent can find the product, authorize payment and leave a workable order. For an exporter, it means a European customer can receive qualifying goods at a predictable landed cost. This extends September 15’s demand-conversion problem: the new issue is whether a second route makes the first intermediary less indispensable. The distinction matters for value capture. A new interface can attract attention while the established payment provider still earns the fee; a new trade pact can expand theoretical access while qualified exporters collect the initial gains. Competition becomes commercially consequential when switching is practical, rather than merely imaginable.

Predictions

I predict: The EU and Philippines will announce formal conclusion of their FTA negotiations by December 31, 2026, following today’s leaders’ announcement. Signature, ratification and entry into force are separate milestones and do not determine this forecast. (Confidence: medium; Check by: 2026-12-31)

September 22, 2026 · 03:25 AM ET