Ramp Is Building the CFO Control Plane

The average Ramp customer has more than three bank accounts. Checking at JPMorgan, a sweep at First Republic's successor, maybe a money market at Schwab. Each one has its own portal, its own balance display, its own cutoff time for wires.

Ramp already sees the spend because it issues the cards. But it doesn't see the cash. It doesn't know what's sitting in the Schwab account or whether Friday's payroll will overdraw the JPMorgan checking account. The CFO knows, because the CFO has four browser tabs open.

Between October 15 and October 20, Ramp made three moves. It raised $1.85 billion at a $60 billion valuation. It bought Vesto, a treasury platform connected to 12,000 financial institutions. And it applied to the OCC for a national trust bank charter covering digital-asset custody and fiduciary investment management.

Each one, reported separately, reads like a funding headline, an acqui-hire, and a regulatory filing. Together they are the architecture of a single product: one screen where the CFO sees every dollar, holds stablecoins without a third-party custodian, and manages AI spend as a line item.

The cash problem

A corporate finance team that uses Ramp for expense management still logs into separate portals to check bank balances, move money between accounts, and forecast whether next week's payroll clears. Ramp sees what goes out on cards, but the banks see what sits in accounts, and nobody sees both in the same place.

Vesto spent three years building connections to those bank portals. Not screen-scraping, which breaks when a bank redesigns a login page. Vesto built direct integrations to enterprise banking platforms: JPMorgan Access, Bank of America CashPro, CitiDirect, and Wells Fargo Vantage. More than a hundred companies used it to pull balances, categorize transactions, and forecast cash across every account.

Ramp's blog post on the acquisition put the number at 12,000 financial institutions connected out of the box. The four enterprise connections matter most. A startup with one bank account doesn't need Vesto. A company with $50 million spread across JPMorgan, BofA, Citi, and Wells Fargo does.

The acquisition folds that visibility into the same dashboard where the finance team already manages cards, approvals, bill pay, and reimbursements. One view of what's going out and what's sitting in reserve, without switching systems.

What Ramp gets from Vesto is a data position, not just a feature. If you can see every bank balance alongside every card charge, you can forecast cash better than the company's own spreadsheet. And if you can forecast cash, you can tell the CFO when to move money before they ask.

The custody problem

Since July 2026, Ramp has offered stablecoin accounts. Customers can hold USDC or USDT balances in Ramp, earn up to 3.25% in rewards, and pay vendors in stablecoins across 140 countries. More than 70% of stablecoin payment volume on Ramp moves outside traditional banking hours, according to the company's blog.

But there's a footnote on Ramp's stablecoin page worth reading. "Stablecoin custody is provided by Bridge Building Inc. and affiliates." Bridge is Stripe's stablecoin subsidiary. Every stablecoin dollar sitting in a Ramp account is actually custodied at Stripe's infrastructure.

Ramp is paying a competitor to hold its customers' money.

On October 16, Ramp applied to the OCC for a national trust bank charter. The filing covers digital-asset custody and fiduciary investment management, with no deposits and no loans.

That distinction determines the regulatory weight. A full bank charter, the kind Mercury is pursuing, means FDIC insurance, CRA obligations, and capital requirements calibrated to a lending book. A trust charter skips all of that and focuses on holding and managing assets under a fiduciary obligation.

Ramp's filing follows the same path as Circle, Ripple, Paxos, BitGo, and Fidelity Digital Assets, all of which received conditional OCC approval in December 2025. Bridge, now part of Stripe, got its conditional approval in February 2026. Modern Treasury and Rain applied in October 2026, the same month as Ramp.

If approved, Ramp National Trust Bank would custody stablecoin balances directly. No Bridge dependency. The stablecoin balance that shows up in the same dashboard as the JPMorgan checking account and the BofA CashPro balance would sit on Ramp's own books, under Ramp's own federal supervision.

The Brainfood newsletter, which first connected these moves, pointed out that Ramp also holds more than $5 billion in customer checking and investment balances, with the investment side custodied at Apex Clearing. A trust charter could bring the investment custody in-house too, but that claim is harder to verify from public filings.

The AI spend problem

AI has become the fastest-growing corporate expense. Ramp's own data, published as the Ramp AI Index, shows corporate AI spend grew 20.7 times between June 2025 and the Router.com launch in August 2026. For most companies, inference costs now change weekly as new models ship.

Ramp launched Router.com on August 19, the same day reports surfaced of Stripe's $7 billion acquisition of OpenRouter. A model router sends each API request to the lowest-cost model that meets a quality threshold. Ramp's version connects to models from OpenAI, Anthropic, and SpaceXAI, with open models from Nvidia, DeepSeek, GLM, and Qwen served through providers like Fireworks AI.

Ramp said customers using Router reduced inference costs by 40% on average. The company built Router for its own engineering team three years ago and cut its internal inference costs by about 30%.

But the routing matters less than the spend data it generates. Ramp already sees what companies pay for AI through their corporate cards. Router adds the volume data: which models, how many tokens, what each request costs. Ramp can now tell a CFO not just how much the company spends on OpenAI per month, but whether they're overpaying per token relative to a cheaper model that would produce the same output quality.

Stripe bought OpenRouter to close the same loop from the revenue side, pairing what AI companies charge their customers with what those companies pay their model providers. Ramp approaches from the cost side, pairing what companies spend on Stripe with what they spend in tokens. Both are building toward full visibility over the AI P&L, one from the merchant's perspective and one from the buyer's.

The control plane

Each of these moves solves a distinct problem, but the three together assemble something that didn't exist before: a single layer between the CFO and every financial system the company uses.

Ramp already owns the spend layer, where cards, bill pay, and reimbursements flow through one approval engine. Vesto adds the cash layer, where bank balances, forecasts, and account activity from 12,000 institutions show up in the same dashboard. The trust charter adds the custody layer, where stablecoins and investment balances sit under Ramp's own fiduciary obligation. Router adds the AI cost layer, where model selection and inference spend become as visible as any other vendor payment.

A finance team using all four layers would see their JPMorgan checking balance next to their Ramp stablecoin balance next to their AI spend by model, all in the same interface, all governed by the same approval policies.

Ramp processes more than $275 billion in annualized purchase volume, up 175% year over year according to Reuters. It generates more than $2 billion in annualized revenue across 85,000 customers. Those are big numbers, but they describe the card business. The question is whether the card business is the foundation or the ceiling.

What it costs

The trust charter isn't free. Running a federally chartered entity means OCC examinations, capital requirements, BSA/AML compliance infrastructure, and a board composition that meets regulatory expectations. Based on precedent from other trust charter applicants, conditional approval is where the work starts. The timeline from conditional approval to full operation has typically run six to eighteen months for companies like Anchorage Digital and Paxos.

And the charter is being contested. The Independent Community Bankers of America sued the OCC in October 2026, arguing the agency exceeded its authority by issuing trust bank charters for non-fiduciary activities like crypto custody. If a court sides with the ICBA and narrows the OCC's charter authority, every pending application, including Ramp's, is affected. The legal risk is already in court.

Vesto's 12,000 bank connections also carry integration risk. Direct connections to enterprise banking platforms are not API calls that never break. CashPro and CitiDirect change their interfaces. Banks add authentication requirements. The hundred-plus companies that used Vesto before the acquisition represent real production experience with these connections, but maintaining them at Ramp's scale of 85,000 customers is a different problem.

Router competes against Stripe's OpenRouter, which has a two-year head start, $140 million in revenue, and developer relationships across 88% of the Forbes AI 50. Ramp's edge is that it already sees corporate card spend and can tie token costs to the same CFO dashboard, while Stripe sees the revenue side and can tie model costs to billing. The winner likely depends on whether the buyer is the CFO or the engineering team.

What the company told us

Ramp's public statements about the trust charter are minimal. The company's posts on X and LinkedIn said it would offer "fiduciary investment management and digital-asset custody" and that the proposed trust bank "would bring these services under a consistent federal regulatory framework, helping us build great products alongside our banking partners."

The phrase "alongside our banking partners" is deliberate. Ramp isn't positioning the charter as a replacement for its banking relationships. First Internet Bank of Indiana still provides the checking account deposits. The charter is for assets the banks don't hold: stablecoins and, potentially, tokenized money market funds.

Ramp CEO Eric Glyman hasn't publicly detailed how the three moves connect. The closest external analysis came from Simon Taylor's Brainfood newsletter, which framed Ramp as "a stablecoin fiduciary, not a bank" and compared it to Circle and Bridge rather than Mercury.

That framing fits the filing. Mercury wants to take deposits, so it applied for a full bank charter. Ramp wants to hold assets, so it applied for a trust charter. Different products for different customers, both reaching for the same outcome: not renting the infrastructure anymore.