OpenAI Doubles Revenue While Seeking New $30 Billion Investment

OpenAI has roughly doubled its revenue in the first seven months of the year, hitting $12 billion in annualized sales, The Information reports. The publication says ChatGPT now attracts about 700 million weekly active users across consumer and business accounts.

The Microsoft-backed firm has lifted its projected 2025 cash burn to about $8 billion, up $1 billion from an earlier estimate. Sequoia Capital, Tiger Global and other investors are close to adding $7.5 billion toward a second $30 billion funding tranche, the report adds.

Rapid revenue gains paired with a rising burn rate highlight the capital intensity behind keeping ChatGPT’s momentum. Persistent multi-billion-dollar backing from SoftBank, Sequoia and Tiger Global shows marquee investors remain engaged despite the spending ramp.

Microsoft researchers analyzed 200,000 anonymized Bing Copilot chats to score how exposed each occupation is to chatbot-driven automation. Their non-peer-reviewed paper finds roles built around providing information, writing, teaching, and advising are the easiest targets for replacement.

Translators, historians, sales reps, writers, authors, and customer service reps receive the highest “AI applicability” scores, while heavy machinery operators, housekeepers, roofers, massage therapists, and dishwashers sit at the bottom. The ranking stems from observed chatbot usage frequency and success, not speculative model capabilities.

The data invert the usual automation story, suggesting low-wage manual labor is safer than well-paid knowledge work. Even so, the authors note AI still cannot perform every task of any single job and warn that downstream employment effects remain uncertain.

Reddit posted Q2 earnings with advertising revenue hitting $465 million, representing 93% of its total haul. The jump underscores the company’s accelerated bet on AI-powered ad tools.

Last month it rolled out Reddit Insights and Conversation Summary Add-ons, mining user posts so brands can spot trends and embed live discussions in ads. Data-licensing revenue, fueled by new content deals with Google and OpenAI, rose 24% year-over-year to $35 million.

The AI-driven Reddit Answers feature now serves 6 million weekly users, up from 1 million the previous quarter. CEO Steve Huffman says the team is integrating Answers deeper into search to make search a core service across the platform.

Columnist Michael Taylor outlines three situations where generative AI gives misguided business advice and shares fixes for each weakness. He frames the guidance as a test for when leaders should accept algorithmic recommendations or override them with human judgment.

A core failure is models operating on outdated training data, meaning ChatGPT’s knowledge stops at October 2023 and Claude’s at January 2025. Taylor notes Claude even flagged reports of U.S. strikes on Iranian nuclear sites as misinformation because the event post-dated its cutoff.

University of California San Diego researchers show that injecting fresh news into prompts can bring models back in line with reality. Taylor builds the same technique into his Ask Rally tool, allowing users to add “memories” so AI personas reflect current events accurately.

London-based record producer Ben Porter, known as benners, is using industry keynotes and panels to push for AI systems that support rather than replace musicians. He argues that music’s future must honor the values that make the art form meaningful.

Porter’s music already plays in Nordstrom stores and high-profile broadcasts, and he has 35 songs queued for 2025 while he seeks a U.S. label deal. He has shared his AI stance with groups such as The Music Business Association, Byta, and A2IM.

Porter distinguishes assistive tools like iZotope and LANDR—which automate tedious studio tasks—from generative systems that mimic artists without consent. He stresses that responsible AI can boost workflow, but unlicensed imitation threatens creativity and rights.