OpenAI has officially filed confidential paperwork for an IPO, just days after Anthropic made the same move.
After years of racing to build the best models, the biggest AI companies are about to face a very different test: Wall Street.
OpenAI, Anthropic, and SpaceX are all preparing for potential listings at valuations usually reserved for the world’s largest public companies. But the real question isn’t just who gets there first. It’s who can convince investors that the AI economy actually works.
OpenAI has massive consumer adoption with hundreds of millions of ChatGPT users, but it is also burning billions to fund the infrastructure behind its models. Anthropic, meanwhile, has gained momentum with enterprise customers and is pitching investors on a faster path toward profitability.
The AI race is entering a new phase. Building the smartest model was step one. Now these companies need to prove they can become businesses worthy of trillion-dollar expectations.
Supported by Sentry
Don’t do this: an error fires, you dig through Sentry, copy the stack trace, then paste it into your editor to work out what went wrong.
Instead, connect the Sentry MCP server to Cursor, then just ask, in plain English, what’s broken across your projects. No project IDs or exact names needed. Seer reads the stack traces, logs, and your connected codebase to explain why each issue happened, and Cursor applies the suggested fix right in your editor. The same approach works for slow database queries.
Setup is one URL and a login. Beginner friendly, about 10 to 15 minutes.
Apple is finally launching the Siri upgrade it promised two years ago, bringing a more conversational assistant that can understand personal context across messages, emails, photos, and apps.
The problem is that the rest of the AI industry has already shifted its attention somewhere else.
While Apple is catching Siri up to the chatbot era, companies like OpenAI, Anthropic, and Google are racing toward AI agents that can write code, complete workflows, use software, and handle more complex tasks with less human input.
Apple’s bet is different. Instead of moving fastest, it’s focusing on privacy, trust, and deeply integrating AI into the devices people already use every day.
That strategy has worked before. Apple wasn’t first in smartphones, watches, or headphones, but it eventually made those categories mainstream. The question now is whether AI is moving too quickly for the same playbook to work again.
Anthropic says its Mythos AI model can now turn newly discovered software vulnerabilities into working exploits within hours instead of weeks.
That changes one of cybersecurity’s biggest assumptions: companies usually had time between a flaw being discovered and attackers learning how to exploit it. AI could shrink that window dramatically.
In tests, Mythos created its first working exploit for a Windows vulnerability in just 31 minutes and successfully generated multiple attacks across Windows and Firefox security flaws.
The bigger issue is that AI is speeding up both sides of cybersecurity. The same models helping defenders discover and fix weaknesses faster could also make it easier to weaponize those weaknesses before companies have time to respond.
The new security race may not be about who finds vulnerabilities first. It may be about who moves fastest once they appear.
In other developments
Apple is making its AI models free for smaller developers, removing cloud API costs for apps with fewer than 2 million first-time downloads. The move aims to lower the barrier to building AI-powered apps as rising infrastructure costs make experimentation increasingly expensive.
Stock markets slipped as investors grew more cautious about AI valuations and the massive spending needed to build AI infrastructure. The sell-off suggests the AI boom is entering a more selective phase, with investors looking for clearer returns rather than just growth promises.
Microsoft temporarily removed dozens of open-source GitHub projects after hackers reportedly injected password-stealing malware into tools used by AI developers. The breach highlights growing risks around software supply chains as attackers target the infrastructure behind AI coding tools.