Anthropic Q4 business spend growth slows to 15-25% as AI market shifts

Anthropic’s Growth Cools as Enterprises Recalibrate AI Investment

Ramp data released on November 3 shows Anthropic’s quarterly business spend growth decelerating sharply, dropping from 47% in the previous quarter to a range of 15% to 25%. The figure signals a broader correction in AI infrastructure spending as enterprises rethink budget allocation ahead of the holiday tech spending season.

The slowdown arrives alongside the release of new reporting tools from Anthropic, designed to give corporate customers greater visibility into their AI expenditure. According to the company, the move responds to customer demand for better spend transparency as CFOs optimize allocation amid decelerating returns.

Anthropic reported an operating profit in the second quarter and maintains 900% net revenue retention, meaning spending from existing customers grew ninefold year-over-year. However, the broader market appears to be adjusting. Jellyfish data, drawn from 276,000 customers, shows Claude Code and other Anthropic-built tools now account for roughly 2% of total engineering costs, up from a baseline earlier this year. While that figure remains modest in absolute terms, the 10x growth in spending on those tools since the beginning of the calendar year represents a significant cost shift that many finance teams did not anticipate.

Nicholas Arcolano, chief scientist at Jellyfish, noted that the rapid ascent of AI coding spend is forcing a reevaluation of budget priorities. “Even if you’re saying 2% is still small, that 10x growth is growth that most CFOs didn’t plan for,” Arcolano told TechCrunch. The data also indicates that while venture-backed firms are the primary drivers of token spend, many Fortune 500 companies are not seeing major operational gains from deploying AI tools at scale.

Anthropic’s new reporting suite allows managers to view a team’s work over a given date range, the associated cost, and where workflows broke down. The company states it will not use this data for model training by default, and the reports include recommendations for improving results. The tools can be characterized as a lighter version of the forward-deployed engineering services that major AI labs have embedded inside customer organizations—a “FDE-lite” offering intended to deepen stickiness without the overhead of full consulting engagement.

Competitive dynamics remain active, though the perception of heightened rivalry on social media does not match the reality on the ground. Nikolas Albarran, AI product lead at Jellyfish, observed that while industry chatter often anoints rivals as superior, Anthropic has maintained its position through strong enterprise workflows and standardized implementations. “Everyone in the Valley is like, ‘Oh, Codex is better,’” Albarran said. “In reality, Anthropic was just there with great enterprise motions, and they had an incumbent weight of their own — this stuff just worked and worked well, and a lot of people just put a lot of effort into standardizing workflows around it.”

The slowdown in Anthropic’s growth, combined with increased scrutiny on AI return on investment, marks a pivotal moment for the sector. As enterprises prepare for Black Friday tech deals and prepare year-end budgets, the focus has shifted from unchecked expansion to measured optimization.

Market Context and What the Numbers Signal

The deceleration in Anthropic’s business spend growth—from 47% quarter-over-quarter to a projected 15% to 25%—comes at a time when the AI infrastructure market is undergoing its first major revaluation since the 2022 surge. Ramp’s estimates, cited by tech industry analyst Ed Zitron, suggest that the era of double-digit percentage gains for frontier labs is giving way to a more mature, efficiency-focused phase.

This shift is not isolated to a single company. Across the sector, revenue deceleration is becoming the baseline expectation as the initial wave of speculative spending gives way to deliberate budgeting. The Jellyfish report underscores this: Claude Code’s cost, now equal to about 2% of total engineering cost, has grown 10x since January. For CFOs, who typically plan on annual cycles, this kind of rapid, unplanned cost growth demands a reexamination of total cost of ownership.

Anthropic’s response—new reporting tools that offer granular visibility into token spend and workflow efficiency—is a direct answer to this market pressure. By giving customers the data to optimize their usage, the company is not only defending its current position but also embedding a layer of dependency. Once a team has instrumented its workflows around a model and absorbed the operational cost of switching, the friction of migration increases, even if the underlying model capabilities are comparable.

The geographic and institutional framing also deserves note. While venture-backed startups and individual developers have been the most visible users of AI coding assistants, the Jellyfish data confirms that the Fortune 500—the primary target for enterprise AI spending—has been slower to realize measurable gains. This discrepancy between hype and results is likely a contributing factor to the growth slowdown. Enterprises are no longer content to experiment; they are demanding proof of value, and Anthropic’s new tools are designed to deliver exactly that.

What was once a market driven by the promise of exponential capability improvements is now being driven by the calculus of cost and consequence. The numbers from Ramp and Jellyfish alike suggest that the AI spending binge is ending, and a period of strategic restraint has begun.

Anthropic’s Strategic Pivot

Anthropic’s launch of customer reporting tools represents a strategic pivot from pure model deployment to value demonstration. The company shared with investors this quarter that it recorded an operating profit in the second quarter, a milestone that underscores its transition from high-growth startup to sustainable business. The 900% net revenue retention figure further validates that existing customers are not just staying—they are expanding their spend at a rate far exceeding acquisition.

The new tools address a gap that has long existed in the AI market: transparency. Large customers have been vocal about their need to understand exactly what their spending is buying. Anthropic’s reports provide that, offering a breakdown of team productivity, cost, and friction points. Critically, the company has built in privacy controls, stating it will not use the data for model training by default. This is a significant consideration for enterprises wary of proprietary data being fed back into model training pipelines.

The introduction of pooled group budgets is another practical feature, allowing teams to manage token spending collectively rather than as isolated cost centers. This moves the company closer to a traditional software-as-a-service pricing and governance model, which may appeal to organizations accustomed to enterprise procurement processes.

Competitive response from the wider industry has been muted relative to the noise on social media. While OpenAI’s Astra model and other rivals have been hyped as potential challengers, the Jellyfish data suggests that Anthropic’s entrenched position—built on years of enterprise workflow integration and standardized implementations—remains durable. Switching costs are real, and Anthropic’s “incumbent weight,” as Jellyfish’s Nikolas Albarran described it, is not easily dislodged.

For the broader AI market, the message is clear: the free-spending phase is over. The growth Anthropic is experiencing, while decelerating, is still growth. The challenge for the company—and for the sector at large—is proving that value can be sustained at lower growth rates, and that the tools enterprises have adopted can deliver ROI in a budget-conscious environment.

What Comes Next

Anthropic’s growth trajectory will be closely watched in the coming quarters as the company navigates this transition from hyper-growth to optimization. The new reporting tools, now available to customers, represent the first major product innovation of the post-expansion era. Their adoption—and the degree to which they move the needle on customer spend efficiency—will be a key indicator of whether the company can maintain its momentum amid a more cautious market.

Jellyfish’s data suggests that the 2% engineering cost benchmark for AI coding tools will likely rise before it stabilizes, as more organizations factor these expenses into their core operating budgets. For enterprises, the immediate takeaway is a need to reevaluate AI spend not as a line item for experimentation, but as a permanent operational cost requiring the same rigor as any other technology investment.

The deceleration in Anthropic’s business spend growth, from 47% to 15-25%, is the latest signal that the AI market is maturing. What was once a race to deploy the most capable models is now a calculated effort to extract maximum value from existing investments. The enterprises that adapt fastest to this new calculus will be the ones best positioned as the next wave of tech spending, including Black Friday deals and year-end budgets, takes shape.