Competition between OpenAI and Anthropic has spilled into initial public offerings. Both companies are reportedly preparing for IPOs as early as the fourth quarter this year. There are also signs that OpenAI wants to bring forward its IPO timetable with Anthropic in mind.

OpenAI and Anthropic: who will list first?...Targeting an IPO within this year Anthropic to go public before OpenAI?

OpenAI and Anthropic have so far focused on investment while accepting large losses, but an IPO would leave them with little choice but to pay closer attention to revenue and profit. There is also a view that their IPO moves could become an opportunity for the AI industry landscape to be reshaped around profitability.

OpenAI annualised revenue tops $25 billion...Gap with Anthropic narrows

Enterprise software companies are responding to AI threat arguments raised in some quarters with more aggressive AI strategies. According to the industry, major global enterprise software companies have recently rolled out new AI agent products in succession, accelerating a shift to AI-centred businesses. They are also increasingly in an aggressive mode on mergers and acquisitions to strengthen AI capabilities.

Enterprise software industry counters AI threat talk with bold AI agent strategies [Rising Business] How Box broke through the fear of the 'SaaS-pocalypse'

The article also summarises moves by companies in and outside South Korea around AI.

OpenAI again delayed the release of ChatGPT's "adult mode". An OpenAI spokesperson said in an interview with Axios, "We delayed the release of adult mode to focus on work that is more urgently needed by more users first," adding, "Improving AI intelligence, personality and functionality is more pressing." OpenAI also launched an AI-based application security agent, Codex Security. OpenAI is also reported to have revised its AI-based e-commerce strategy. It is said to have focused on enabling payments in certain apps linked to ChatGPT, rather than buying directly within ChatGPT. OpenAI is also said to have started developing an alternative to GitHub, a well-known code repository platform under Microsoft. OpenAI unveiled its new AI model GPT-5.4, stoking competition in AI centred on workplace automation.

OpenAI again delays release of ChatGPT 'adult mode' Following Anthropic, OpenAI also launches a major security push...Unveils Codex Security OpenAI retreats from ambitious ChatGPT commerce strategy..."Supports payments in linked apps" "OpenAI develops a GitHub rival...May also consider deploying it for external business" OpenAI announces GPT-5.4...Expands to documents, coding and work automation

Anthropic launched Claude Marketplace, an e-commerce platform where corporate customers can buy third-party software. Anthropic also introduced a voice mode for its AI coding assistant Claude Code.

Anthropic launches Claude Marketplace...Expands third-party AI software ecosystem Anthropic launches voice mode for Claude Code

Microsoft (MS) released an open-source multimodal AI model, Phi-4-reasoning-vision-15B, with 15 billion parameters. The model combines existing algorithms SigLIP-2 and Phi-4 Reasoning, and is optimised for processing multimodal data such as science and mathematics graphs. Amazon Web Services (AWS) will officially support OpenClaw, an open-source AI agent tool, on its virtual private server (VPS) Lightsail.

Microsoft releases open-source multimodal AI model with 15 billion parameters AWS officially supports OpenClaw on Lightsail..."Lowers barriers to installing personal AI agents"

Google released a command-line interface (CLI) for Workspace. The Google CLI supports AI agent tools such as OpenClaw to access Google's major apps and data. Google is expanding the Canvas feature in AI mode to all users in the United States. Users can draft documents or create customised tools within Google Search, and can enter ideas and convert them into code to produce apps or games. Google also significantly expanded Gemini AI assistant features through a Pixel phone update. With the update, Pixel 10, Pixel 10 Pro and Pixel 10 Pro XL users can use AI to automate various tasks such as grocery shopping and vehicle reservations.

Google releases Workspace CLI...Lowers barriers to linking AI agents Google AI mode opens Gemini Canvas feature across the United States Google strengthens Pixel phone AI...Gemini supports shopping and reservation functions

Meta recruited the team from Atma Sciences, which developed the AI-based coding app "Gizmo". The move is seen as aimed at strengthening its AI strategy and developing next-generation interactive content.

Meta recruits development team behind AI app 'Gizmo' that lets users create mini-games

Cursor, an AI coding tool, unveiled a new tool called "Automation". "Automation" runs agents automatically under certain conditions such as codebase changes, Slack messages and timers. This means engineers do not need to manually manage multiple agents, and operations run with humans stepping in only when needed. Cursor is also reported to have topped $2 billion in annualised revenue.

Cursor unveils new agent coding tool 'Automation' Cursor enters $2 billion annual revenue course...Big-company push pays off

Samsung SDS signed contracts to supply ChatGPT Enterprise to Korea Zinc, iCraft and TmaxSoft, among others. LG CNS unveiled a small data centre, "AI Box", that can house 576 GPUs in a single container, and is moving to target South Korea's AI data centre market. Nota, an AI lightweighting company, developed quantisation technology that cuts memory usage of Upstage's LLM "Solar 100B" by 72.8 percent.

Samsung SDS supplies ChatGPT Enterprise in succession to Korea Zinc and others LG CNS launches small AI data centre 'AI Box'..."Can be built in six months" Nota compresses Upstage 'Solar 100B' memory by 72 percent

South Korean startups that received government graphics processing unit (GPU) support are accelerating AI model development. Companies that benefited from the Ministry of Science and ICT's GPU support programme are developing a diffusion-based transformer large language model (LLM), a Korean-language specialised document-parsing model, and an action-oriented AI agent, among others.

Companies backed by government GPU support accelerate development of next-generation LLMs and action-oriented AI