When ChatGPT is no longer satisfied being a chatbot, and when OpenAI loses $1.22 for every dollar earned while Anthropic stands at the door of profitability, June 2026 marks a fundamental narrative shift in the AI industry: from “whose model is stronger” to “who makes money first.”
ChatGPT’s Biggest Overhaul: From Chatbot to ‘Super App’
According to a June 7 report from CCTV Finance, OpenAI is preparing the largest upgrade to its chatbot ChatGPT since its launch in 2022, with the new version rolling out over the coming weeks.
The core of this upgrade is transforming ChatGPT into a “super app,” integrating coding tools and AI agents, with multiple new products aimed at diversifying revenue streams. The upgrade will heavily prioritize the company’s coding product Codex, increasing resource allocation. This means OpenAI’s products are no longer limited to Q&A chatbots — they are shifting toward agents that can autonomously complete tasks for users.
Codex is OpenAI’s AI code generation training model, built on the GPT architecture and focused on translating natural language instructions into code across multiple programming languages. In August 2021, Codex was released in API beta form. In April 2025, the open-source local tool Codex CLI was launched. Last October, the official Codex version was released, having processed 40 trillion tokens cumulatively, adding Slack integration, Codex SDK, and management tools, supporting enterprise-level environment controls and analytics monitoring — making it one of OpenAI’s fastest-growing products. Today, it is already one of OpenAI’s primary revenue sources.
AI Pulse View: The “super app” concept is not new — WeChat and Alipay have already validated this model in China. But upgrading an AI chatbot into a super app means OpenAI is attempting to redefine the human-AI interaction paradigm: from “asking questions” to “getting things done.” This is essentially a leap from an “information tool” to an “execution tool.” Whoever gets users to habitually “let AI do tasks for me” rather than “chat with AI” first will capture the next generation’s traffic gateway.
OpenAI vs Anthropic: Revenue Leader, Profit Laggard
Alongside the ChatGPT super app upgrade, OpenAI is also accelerating its IPO preparations. According to The Information, OpenAI’s Q1 revenue was approximately $5.7 billion, about $1 billion more than Anthropic for the same period. OpenAI’s growth comes primarily from Codex, enterprise sales, and ChatGPT advertising tests.
However, revenue leadership does not equal commercial health. OpenAI’s Q1 adjusted operating profit margin stood at -122% — meaning, excluding certain large costs, for every dollar earned, the company still loses about $1.22. AI model training, inference compute, product expansion, and user subsidies continue to consume massive amounts of cash.
In stark contrast, Anthropic may achieve approximately $600 million in operating profit in Q2, with a margin of about 5.5%, already approaching the breakeven point.
On user growth, OpenAI is also facing bottlenecks. Q1 average weekly active users were approximately 905 million, down from a February peak of about 920 million. On the paid side, ChatGPT had 55 million consumer subscribers in Q1, up from about 47 million at the end of last year, but growth is decelerating.
AI Pulse View: These figures reveal a harsh reality: in the AI industry, “first-mover advantage” and “scale advantage” don’t automatically translate to commercial advantage. OpenAI has higher revenue, more users, yet sinks deeper into losses. Anthropic, though later to market with lower revenue, demonstrates better cost control and commercial efficiency. This echoes Amazon in the internet era — long-term losses ultimately yielding market dominance. But the question remains: Can OpenAI become the Amazon of the AI age? Or will it be overtaken by more efficient latecomers?
Anthropic’s ‘Financial Offensive’: AI Agents Targeting Wall Street
On the commercialization front, Anthropic has chosen a distinctly different path from OpenAI — deep-diving into high-value industries.
To win Wall Street clients, Anthropic has launched a series of new AI agents designed to handle a broader range of financial services tasks. These AI agents can draft client meeting presentations, review financial statements, and escalate cases for compliance review.
Anthropic stated it will make Claude models work better with third-party software such as Excel, PowerPoint, and Outlook, integrating data from financial industry partners including Dun & Bradstreet and Moody’s. Anthropic is also deepening its footprint in the financial sector through new joint ventures with Blackstone, Hellman & Friedman, and Goldman Sachs. The joint venture company will drive adoption of its software across more enterprises.
AI Pulse View: Anthropic’s commercial strategy reflects a “precision strike” mentality — rather than competing with OpenAI on consumer user scale, it’s building moats in high-value B2B scenarios. Financial sector clients have strong willingness to pay, high per-customer value, and strong stickiness. Once trust is established, it’s hard to displace. This mirrors how Salesforce entered through enterprise CRM — Anthropic is attempting to become the “AI infrastructure” for the financial industry.
Talent Exodus Accelerates: Sora Core Member Gabriel Leaves OpenAI
Against the backdrop of this commercial race, OpenAI’s talent drain is also accelerating.
Sora core member and high school dropout genius Gabriel Petersson has officially announced his departure from OpenAI. Earlier this year, Gabriel had already stepped away from the Sora project and formed a new team within OpenAI to酝酿 (incubate) a “big plan.” According to his own statements, he is betting everything on building the “last product before AGI.”
This follows Clive Chan (OpenAI’s self-developed chip team’s “employee #2” joining Anthropic) and Andrej Karpathy (OpenAI co-founding member joining Anthropic’s pre-training team) — marking yet another key talent departure from OpenAI.
AI Pulse View: Talent loss is one of the most telling leading indicators for technology companies. When core engineers choose to “go out and build” rather than “incubate internally,” it signals disagreement with the company’s direction and pace. Gabriel’s departure isn’t because he’s bearish on AI — it’s because he has more radical ideas. This also indirectly reflects that OpenAI, under commercialization pressure, may need to slow down some forward-looking research.
Three AI Labs Hire Economists in Unison: AGI Economics Arrives
A notable trend: Google DeepMind, OpenAI, and Anthropic are all hiring economists simultaneously.
Google DeepMind has even created a new “Director of AGI Economics” position, filled by University of Chicago professor Alex Imas. He and Epoch AI’s Head of Economics, Phil Trammell, recently completed a 70-minute deep conversation attempting to answer a core question: When AI can do everything, what becomes scarce?
The U.S. labor income share dropped to 53.8% in Q3 2025, the lowest level since records began in 1947. DeepMind co-founder Demis Hassabis said at Stanford last week that AGI is expected around 2030, with impacts 10x that of the Industrial Revolution, arriving 10x faster.
AI Pulse View: When AI companies start hiring economists, it means they’re no longer just asking “can we build it?” but seriously considering “what will the world look like after we do.” This isn’t academic curiosity — it’s commercial necessity. If your product will profoundly reshape labor markets, income distribution, and economic structures, you need to understand these changes in advance to make the right calls on regulation, pricing, and strategy. “AGI Economics” may well become the most important emerging interdisciplinary field of 2026.
The Bigger Picture: The AI Industry Is Undergoing a ‘Value Reset’
Synthesizing these events, we can clearly see the AI industry undergoing a profound “value reset”:
From growth to profit: Capital markets no longer reward “user growth” and “token volume” — they demand a real path to profitability. OpenAI’s -122% margin is a wake-up call: no matter how high the revenue, without positive cash flow, it’s all a castle in the sky.
From general to vertical: Anthropic’s deep dive into finance shows that AI’s value is shifting from “general capabilities” to “vertical applications.” Saving an investment banker 100 hours with AI is more commercially valuable than getting a million users to chat 10 more sentences.
From technology to economics: The simultaneous hiring of economists by all three AI labs signals that the industry’s focus is expanding from pure technical competition to systematic thinking about economic and social impact.
From product to platform: ChatGPT’s “super app” transformation shows that AI companies are attempting to evolve from single products into platforms, building moats by integrating multiple tools and services.
AI Pulse View: The AI industry in 2026 is undergoing a transition from “adolescence” to “adulthood.” Technical fervor is giving way to commercial rationality, and the model arms race is yielding to the profitability race. This isn’t a decline of the AI industry — it’s its maturation. For companies and talent that truly understand how to create economic value in the AI era, the best is yet to come.
This article synthesizes public reporting from CCTV Finance, The Information, 36Kr, The Verge, and other media sources. All data and events are from verifiable public sources.