The Market Chose Governance: What Ramp's AI Spend Data Reveals About Brand Trust

Opinions are cheap. Spend data is expensive. When a company allocates budget to an AI provider, that decision passed through procurement, security review, and executive approval. It reflects organizational judgment, not individual preference.

Ramp processes billions in corporate transactions. Their Velocity AI Index, published March 11, 2026, tracks which AI providers businesses are actually paying for. Not downloading. Not trying. Paying.

The numbers tell a story we have been building toward for months.

The Spend Data

As of February 2026, 47.6% of businesses on Ramp’s platform spend on AI tools. That number has been climbing steadily, but the distribution underneath it shifted dramatically.

Anthropic grew from 4% of businesses to 24.4% in twelve months. Their February growth was 4.9 percentage points, the largest single-month gain since Ramp began tracking. When a business purchases an AI provider for the first time and both Anthropic and OpenAI are options, Anthropic wins roughly 70% of the time.

OpenAI recorded its largest monthly decline ever: 1.5 percentage points. Google sits at 4.7%. xAI is below 2%.

Perhaps the most telling detail: Ramp’s economist Ara Kharazian notes that Anthropic is “actively turning away revenue” because of compute constraints. They cannot serve all the demand. And demand keeps growing.

The Uncomfortable Question

Why? Product quality is the obvious answer, and probably the correct one. Claude’s coding performance, its ability to handle long contexts, its consistency on enterprise tasks. These are measurable, and enterprises measure them.

But product quality alone does not explain a 6x growth in twelve months while your primary competitor declines. Products improve incrementally. Market share shifts this dramatic have a second variable.

The timeline matters. OpenAI announced a partnership with the Department of Defense. Public backlash followed. Senator Brian Schatz switched to Claude publicly. Katy Perry did the same. Ramp’s data shows the AI provider choice is becoming an identity signal for businesses, what Kharazian calls a “green bubble/blue bubble” dynamic.

As we explored in The Safety Paradox, Anthropic’s founding story is inseparable from its governance posture. The company exists because its founders believed OpenAI was not taking safety seriously enough. That origin story became a brand asset. Enterprises that care about responsible AI deployment (or need to demonstrate that they care) gravitate toward the provider whose identity is built on that commitment.

Correlation, Not Causation. But Still.

Let me be honest about three things.

First, Ramp’s customer base skews toward tech-forward, VC-backed companies. This is not a random sample of all enterprises. The 47.6% AI adoption rate reflects a population that adopts technology faster than the median Fortune 500.

Second, correlation is not causation. OpenAI’s decline could be driven by pricing decisions, product limitations, or natural market correction after years of dominance. Attributing it entirely to the DoD partnership backlash overstates what the data can prove.

Third, we use Claude. This article was written using Anthropic’s product. Victorino Group’s consulting practice runs on Claude. We have a commercial relationship with the platform that is winning in this data. Take our interpretation with that context.

What the data can support: governance posture correlates with commercial outcomes in the enterprise AI market. Whether it causes those outcomes, amplifies other advantages, or merely signals something deeper about organizational values is a question the data does not answer cleanly.

What the Data Does Answer

In Growth Is Now a Trust Problem, we argued that traditional growth channels are collapsing and trust-based growth is replacing them. Ramp’s data is the clearest empirical evidence for that thesis we have seen.

Enterprises are not evaluating AI providers on a spreadsheet of features. They are evaluating them on a cluster of signals: product quality, safety posture, governance commitments, public behavior, and the identity implications of the choice. The “green bubble/blue bubble” framing from Ramp’s report is not hyperbole. It describes how procurement decisions actually work when the products are close enough in capability that the tiebreaker is something else.

In The $500 Billion Question, we argued that governance is the only durable moat in the AI market. That argument was about the companies deploying AI. Ramp’s data extends it to the companies building AI. Anthropic’s governance posture is functioning as a competitive advantage at the provider level, not just the enterprise level.

The mechanism: enterprises that have invested in governance infrastructure prefer a provider whose values align with that investment. An organization that built compliance controls, safety policies, and responsible use frameworks does not want to explain to their board why they chose the AI provider making headlines for military partnerships and safety rollbacks. The provider choice becomes an extension of the enterprise’s own governance story.

The Identity Signal Thesis

Ramp’s economist describes something we have not seen in enterprise software before. The choice of AI provider is becoming a values statement, similar to how consumer brands function but in a B2B context.

This is new. Enterprise software procurement has historically been a rational, feature-driven process (or at least it has pretended to be). Nobody chose Salesforce over HubSpot because of what it said about their company’s values. They chose it because their enterprise architecture team evaluated integration requirements.

AI is different. The technology is close enough to general intelligence that the philosophical and ethical positions of its creators bleed into the product evaluation. When your CRM vendor has an opinion about consciousness, containment, and military use, that opinion becomes part of the procurement criteria whether you put it on the evaluation rubric or not.

This creates a flywheel. Governance-minded enterprises choose governance-positioned providers. Those providers attract more governance-minded enterprises. The brand association strengthens. The competitor’s brand association weakens. The 70% head-to-head win rate is not just a product metric. It is a trust metric accumulating compound interest.

What This Does Not Mean

It does not mean governance alone wins markets. A governance-positioned provider with an inferior product would not see these numbers. Claude has to be good enough (and by most benchmarks, it is more than good enough) for the governance signal to function as a tiebreaker.

It does not mean OpenAI is finished. They remain the largest single provider in Ramp’s data. Market share shifts of this magnitude are notable precisely because they are rare. The trend could reverse with a strong product release or a strategic pivot.

It does not mean every enterprise choosing Claude is doing so for principled reasons. Some are following peers. Some are responding to internal pressure from developers who prefer the product. Some are simply trying something new. Aggregate data reveals patterns, not motivations.

What It Does Mean

For AI providers: your governance posture is now a revenue variable. Not a PR strategy. Not a regulatory hedge. A measurable driver of commercial outcomes in the enterprise market. Ramp’s data makes that argument with transaction-level evidence.

For enterprises evaluating AI providers: the choice carries signal beyond the product. Your board, your regulators, your employees, and your customers will interpret your AI provider choice as a statement about your organization’s values. That interpretation will be correct.

For the industry: the market is beginning to price governance. Not perfectly. Not completely. But the direction is clear in the spend data. Companies that invest in governance (both AI providers and the enterprises that use them) are being rewarded by the market. Companies that trade governance for speed or military contracts are seeing the cost in their commercial metrics.

We have argued across multiple pieces that governance would become a competitive advantage. Ramp’s data does not prove that argument conclusively. But it is the strongest empirical signal we have seen that the market is starting to agree.

This analysis draws on Ramp’s Velocity AI Index by Ara Kharazian (March 2026). Related Victorino analysis: The Safety Paradox (February 2026), Growth Is Now a Trust Problem (January 2026), and The $500 Billion Question (March 2026).

Victorino Group helps organizations build governance infrastructure that turns AI trust into competitive advantage. Let’s talk.