Anthropic passed OpenAI in revenue growth for the first time

By Gammatek ISPL, Industrial Systems & Compliance Analyst at Gammatek ISPL

Last updated: August 2026 | 11 min read

Gammatek ISPL advises manufacturing, chemical, and pharmaceutical plants on industrial safety, compliance, and software vendor evaluation at Gammatek ISPL. This analysis draws on Gammatek's direct experience helping plants assess vendor viability during software procurement, alongside publicly reported financial data (current as of August 2026, sourced from Wall Street Journal, CNBC, and Bloomberg reporting). Gammatek has no financial relationship with Anthropic or OpenAI.

When an AI vendor's financial health shifts, the plants depending on their software feel it months later — often without warning

Why This Matters If You're Not Buying AI Chatbots

Two of the world's largest AI labs just posted wildly different quarters. Anthropic reported roughly $11.5 billion in Q2 2026 revenue with its first-ever operating profit. OpenAI posted $6.7 billion, growth that slowed to 18% quarter-over-quarter, and an operating loss that widened to $12.3 billion. If you run a manufacturing, chemical, or pharma plant, this might look like industry gossip that has nothing to do with your operations.

It isn't. If any part of your plant's software stack — predictive maintenance, compliance monitoring, quality control, scheduling — runs on AI models licensed from one of these companies (directly or through a vendor built on top of them), then that vendor's financial trajectory is now part of your own operational risk. A vendor that can't sustain its cost structure doesn't just quietly fail — it changes pricing, deprecates features, gets acquired, or in worst cases, shuts down a product line with a few months' notice. For a plant relying on that software for safety monitoring or compliance recordkeeping, that's not a minor inconvenience.

What Actually Happened, in Plain Terms

	Anthropic	OpenAI
Q2 2026 revenue	~$11.5B (preliminary)	$6.7B
Sequential growth	Roughly doubled from Q1's $4.73B	+18% from Q1's $5.7B
Operating result	First operating profit (~$559M projected, preliminary reports vary)	Operating loss widened to $12.3B, up from $9.3B in Q1
What it signals	Faster path toward sustainable unit economics	Continued heavy compute spend outpacing revenue growth

A caveat worth stating plainly: these are preliminary, investor-disclosed figures reported by outlets including the Wall Street Journal, CNBC, and Bloomberg — not audited public filings, since neither company is yet publicly listed. Numbers could be revised. That uncertainty is itself part of the lesson below.

The Real Takeaway: Vendor Financial Health Is a Procurement Criterion, Not Just a Finance Department's Problem

Plant operations teams are used to evaluating software vendors on features, integration effort, and support quality. Financial stability rarely makes the checklist — it's treated as something IT or finance handles separately, if at all. That's a gap worth closing, for three concrete reasons specific to industrial environments:

1. Compliance systems can't tolerate sudden vendor changes. If a compliance or audit-trail platform your plant depends on is white-labeling or built on top of a foundation AI model, a pricing shift or deprecation from the underlying provider can cascade into your own audit readiness — right before an inspection, if you're unlucky.

2. Predictive maintenance tools depend on continuous model access. A monitoring platform (like FixitX-style predictive maintenance software) that relies on a third-party AI provider for anomaly detection needs that provider to still exist, at a stable price, next year and the year after. A vendor burning $12 billion a quarter with no clear profitability timeline is a different risk profile than one that just posted its first operating profit — regardless of which one currently has the flashier product.

3. Multi-year industrial software contracts outlive typical AI hype cycles. Plants sign 3-5 year software agreements. The AI vendor landscape has shown it can shift dramatically in a single year — this Anthropic/OpenAI divergence itself happened within two quarters. Locking into a platform without checking the underlying vendor's trajectory is a bet most procurement teams wouldn't consciously choose to make if they saw it laid out this way.

A Practical Framework: Questions to Ask Before Signing

Before adopting or renewing any AI-powered software for plant operations, safety monitoring, or compliance, it's worth asking:

• Is this vendor's product built directly by them, or licensed on top of a third-party foundation model (Anthropic, OpenAI, or others)? If the latter, what happens to your product if that underlying provider changes pricing or terms?

• Has the vendor disclosed anything about their own runway, funding stage, or path to profitability — even informally?

• What's the contractual protection if the vendor's underlying AI provider changes access terms mid-contract?

• Is there a fallback or export path for your data and workflows if the vendor's product is discontinued or substantially changed?

• How many other customers in regulated industries (pharma, chemical, food) does this vendor serve — is there a track record specific to your compliance requirements, or are you an early adopter absorbing that risk?

None of these questions require picking a "winner" between Anthropic and OpenAI, or any other lab. The point isn't to bet on which AI company wins — it's to build vendor evaluation habits that account for a genuinely volatile layer of the software stack most plants have never had to think about before.

Where This Leaves Industrial Buyers

The Anthropic/OpenAI divergence is a useful, concrete illustration of something that was already true: the AI vendor layer underneath your plant's software is not static, and its stability now belongs in the same conversation as uptime guarantees, data security, and compliance certifications. Plants that treat this as purely a finance-department curiosity are the ones most likely to be caught off guard when a vendor's trajectory shifts mid-contract.

This is exactly the kind of risk assessment Gammatek ISPL builds into vendor and software evaluations for manufacturing, chemical, and pharma clients — not just "does this software work," but "will it still be reliably compliant and supported in three years."