The AI programming tool CodeX, developed by OpenAI, was initially aimed at developers but has now been repositioned as a general-purpose intelligent agent product for knowledge workers. OpenAI's Chief Revenue Officer Dennis Drayser stated this Tuesday that the product is experiencing a significant growth surge.

As detailed in a report published on Tuesday, a large number of developers are shifting from the more popular Anthropic Claude coding tool to CodeX. Developers note that OpenAI has been continuously iterating on its models and the CodeX client in recent months, enabling it to handle longer-term and more complex development tasks.

While CodeX has substantially boosted development efficiency for OpenAI's internal engineers, this surge has also created new challenges. According to two informed sources, the sharp increase in employee code submissions has caused frequent outages in the company's massive code repository management systems.

These sources explain that the daily number of code modifications by OpenAI's thousands of engineers has skyrocketed from an average of 2-3 changes to over 10. Each code change requires running thousands of machine-hours of automated tests across multiple servers to verify operational status and check for bugs and security vulnerabilities.

The massive volume of code changes and automated testing loads has overwhelmed OpenAI's continuous integration (CI) operations system and its code repository accounts on the GitHub platform. System overload leads to service interruptions, leaving engineers unable to submit code for minutes or even hours at a time. Such incidents now occur every few weeks.

An official OpenAI spokesperson responded: "CodeX has tangibly improved our R&D efficiency, and we are concurrently scaling up our internal development infrastructure. Some strain on internal deployment facilities during this scaling process is inevitable, but we are fully committed to ensuring the stability and reliability of end-user products. As AI reshapes software development, such operational scaling is becoming an industry norm, and our technical teams are continuously optimizing system capacity and stability."

Currently, this type of system overload issue is concentrated primarily within top-tier tech companies that heavily utilize AI programming tools.

However, this phenomenon signals a future industry trend: as more companies adopt AI-powered automated development tools, similar system strain and failures will become widespread. GitHub executives have previously acknowledged that AI coding tools have caused a massive spike in platform traffic, leading to multiple outages. This column has also reported that OpenAI is developing its own alternative to GitHub, partly to mitigate platform load risks.

Industry Briefing

At an event on Tuesday focused on enterprise adoption of CodeX, OpenAI's Chief Revenue Officer Dennis Drayser disclosed that the company's paying enterprise customers have surpassed 2 million, doubling over the past year. Enterprise business revenue now accounts for 40% of total revenue, with the company aiming to increase this share to 50% by year-end.

In a follow-up interview, OpenAI CEO Sam Altman addressed customer concerns about the company's own products potentially competing with their businesses.

"We have no intention of entering the market to compete with our customers, but widespread market concerns exist, and we need to continuously address them."

However, on the same day, OpenAI launched a new investment banking plugin for CodeX, designed to help banking professionals automatically generate pitch materials, benchmark peer companies and M&A targets, conduct due diligence, and produce investment recommendations. This functionality directly overlaps with the core offerings of financial AI startups within the OpenAI ecosystem, such as Rogo and Hebbia.

Policy Update

President Trump signed a long-anticipated executive order on artificial intelligence this Tuesday, ending weeks of market speculation. While there was a prevailing belief that the administration was resistant to industry regulation, cybersecurity risks posed by advanced large models (like Anthropic's Mythos) ultimately prompted the White House to enact a binding framework, though specific implementation details remain unclear.

Overall, the faction within the White House advocating for stronger AI safety controls prevailed. The final text of the enacted order is largely consistent with a draft from May that was later abandoned. President Trump had previously shelved the earlier version following lobbying by his former AI and crypto industry advisor David Sacks, who expressed concerns that stringent regulation could allow China to catch up in the AI technology race. The only change in the new order is a reduction in the pre-deployment notification period for companies releasing advanced models from a maximum of 90 days to 30 days.

Otherwise, the main provisions of the order follow the previously leaked version: it clarifies that all industry guidelines are voluntary and will not establish a mandatory licensing regime. However, the wording leaves significant ambiguity, with details on implementing voluntary national security notifications and how the government will select "trusted partners" for prioritized access to advanced models remaining unspecified. Sacks finalized this regulatory framework after revising the language. Internal policy dynamics within the White House remain delicate, and a 60-day period for implementing detailed rules will follow the order's enactment. — Leo Schwartz

Investment, Funding, and Product News

Agile Robots: An AI industrial robotics company, is in talks for a new funding round of $800 million, with SoftBank planning to invest over $300 million (Bloomberg).

Suno: An AI music generation application, raised over $400 million in a Series D round, achieving a post-money valuation of $5.4 billion, led by Bond Capital.

Cyera: An enterprise data security AI service provider, is negotiating a $300 million funding round targeting a $12 billion valuation, led by Evolution Equity Partners (TechCrunch).

Coralogix: A Boston-based cloud monitoring startup, raised $200 million in a Series F round, reaching a $1.6 billion valuation, co-led by Apax Partners and CPP Investments.

Town: A personal AI assistant project, raised $55 million in a Series A round led by Andreessen Horowitz.

Adaptive Innovations: A home healthcare AI automation service provider, raised $50 million in a Series A round led by Felicis and Bain Capital Ventures.

Endra: A Stockholm-based mechatronic engineering AI startup, raised $50 million in a Series A round led by a16z.

Subtle Medical: An AI medical imaging enhancement company, raised a total of $33 million, including a Series C round led by Morgan Stanley Expansion Capital.

Findigs: A rental application AI risk assessment company, raised $32 million in a Series C round led by RPM Ventures.

ZeroDrift: An email & call content compliance AI validation company, raised a multi-million dollar seed round from investors including the a16z speed fund and other institutions.