CellPoint Gets $34M for Zenith; Payment Margins Are the Test

CellPoint announced a $34 million investment from Toscafund and the launch of Zenith in London on September 22, 2026. Managed through Toscafund’s private equity affiliate Penta Capital, the investment backs an AI decision layer for airline and hotel payments. Zenith is designed to sit above CellPoint’s own payment orchestrator or one supplied by another company, leaving the merchant’s existing payment stack in place.

PhocusWire’s September 22 report independently confirms the investment and Zenith launch. The commercial aim is to increase approvals, lower payment costs and recover revenue from failed payments. Those are product aims, not quantified customer outcomes: the case for better margins depends on whether any additional completed sales outweigh the full cost of obtaining them.

The investment funds Zenith’s build-out

FinTech Futures’ account of the financing identifies Kevin Murphy as group chief executive and Patrick Uckermark as chief product and technology officer, and reports plans for 50 new roles across product, data, AI, sales and partnerships. The planned hiring spans Europe, the Americas and Asia. It indicates how the investment is intended to support development and sales, rather than how many merchants have deployed Zenith.

The leadership and staffing changes matter to the investment case because Zenith’s proposed value depends on both payment data and integration with systems merchants already use. Building a decision layer is one part of that work; getting comparable results across different merchants and payment providers is another. The disclosed financing does not establish that either task has produced a measured improvement in customer margins.

Zenith sits above the existing orchestrator

The proposed position in the stack is straightforward: airline or hotel checkout → Zenith decision layer → existing payment orchestrator → configured payment providers. The orchestrator remains connected to the routes available to the merchant, while the added layer is intended to turn payment data into automated commercial decisions. This is a conceptual map of the announced deployment model, not a published specification of every integration or decision Zenith makes.

That position could spare a merchant from replacing established payment connections before assessing the new product. It also widens the potential market beyond users of CellPoint’s own orchestrator. The public description does not identify a live integration with a third-party orchestrator, however, or specify what data, controls and permissions such an installation would require.

The distinction matters when attributing results. If a merchant changes acquirers, fraud settings or retry rules while adding Zenith, improved approvals might come from any of those changes. A different mix of countries, payment methods or customers could have the same effect. The layer’s contribution can only be separated from those factors when the comparison records what else changed.

Better margins require three linked measures

Approval rate is the share of eligible payment attempts that receive authorization. For Zenith, a meaningful comparison would distinguish initial attempts from retries and compare similar transactions by market, payment method and other relevant characteristics. Otherwise, a higher overall rate could reflect more payments from easier-to-approve segments rather than better decisions for comparable bookings.

Payment cost per completed sale captures a different part of the claim. It should include provider and processing charges, the cost of unsuccessful attempts and retries, and any charge for the added decision layer. Looking only at the fee on an approved transaction could miss a more expensive sequence of attempts that led to it. The same cost definition must be used for the existing policy and the Zenith-guided policy.

Recovered revenue requires more than counting payments approved after an initial failure. The relevant outcome is an additional sale that is paid and fulfilled, with cancellations, refunds and duplicate attempts accounted for. A transaction that the customer would have completed through another route without Zenith is not wholly incremental. This makes a comparable group continuing under the existing policy valuable for estimating what would have happened anyway.

These measures need to be read together. A higher approval rate can justify a costlier route if the added fulfilled sales contribute more than the extra payment expense; a cheaper route can damage margins if it loses valuable sales. A controlled comparison would keep the available providers and fraud policy stable where possible, document unavoidable changes, and report approval, total payment expense and fulfilled sales on the same eligible population. The resulting margin calculation would include Zenith’s own cost rather than treating its claimed benefits as free.

Customer results remain the missing evidence

No named Zenith deployment, measured approval lift, payment-cost reduction or recovered-revenue figure appears in the cited announcement or independent accounts. They also do not establish whether the announced launch means general production availability, a limited rollout or another deployment stage. Those gaps prevent a numerical estimate of the product’s effect on airline or hotel payment margins.

The confirmed event is an investment and a product launch announcement, supported by a plan to expand the team. The unresolved commercial question is whether a decision layer placed above existing orchestration produces incremental, fulfilled revenue after payment costs and the layer’s own fees are counted. Comparable customer results would be needed to answer it.