☕️ OpenAI's revenue run rate tops $40B

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

💰 OpenAI's revenue run rate tops $40B

🚀 SpaceX completes $60B Cursor deal

🤖 Zuckerberg's AI manifesto draws backlash

🐫 Alibaba releases open-weight Qwen 3.8

🖼️ Gemini lets you turn off AI watermarks

⚖️ France court blocks under-15 media ban

Plus: 🎁 13 other news you might like, 🧰 6 tools, and 📚 5 papers.

💰 OpenAI's revenue run rate tops $40B LINK

OpenAI now brings in more than $40 billion a year, roughly double its rate from late 2025, giving the ChatGPT maker fresh momentum as it prepares to go public, according to a Bloomberg report.

The jump was driven by OpenAI's AI coding software, subscription sales, and a new advertising business, with cofounder Greg Brockman saying the annual revenue rate grew more than 20% month over month in July.

OpenAI trails rival Anthropic, which said its rate had passed $47 billion and is expected to go public this fall first, and it has cut some model prices to compete for cost-conscious customers.

🚀 SpaceX completes $60B Cursor deal LINK

SpaceX has completed its $60 billion purchase of AI coding startup Cursor, closing the deal after clearing the final regulatory approvals, according to The Information, which cited comments made to staff at an all-hands meeting.

Staff learned at an all-hands meeting that the Cursor brand name will likely be phased out, with an upcoming agent code-named "Sand" possibly renamed "Grok Bot" after SpaceX's chatbot, though existing tools keep their names for now.

SpaceX first announced the option to buy Cursor in April and confirmed the bid after going public in June, framing Cursor's product and customers alongside its Colossus supercomputer in Memphis as a way to build stronger AI models.

🤖 Zuckerberg's AI manifesto draws backlash LINK

Mark Zuckerberg's 6,500-word manifesto on "personal superintelligence" built by Meta AI drew criticism for glossing over the real harms of AI while painting an upbeat picture of how the technology will reshape society.

Critics say Zuckerberg's examples are out of touch, like calling AI a personalized tutor when students mainly use chatbots such as ChatGPT to skip homework, since no reliable watermarking lets teachers detect AI-written essays.

The essay lands as Facebook and Zuckerberg stay unpopular, with 64% of Americans saying social media harms democracy and a court fining Meta $567 million for harming children, deepening public distrust of tech leaders.

🐫 Alibaba releases open-weight Qwen 3.8 LINK

Alibaba made Qwen3.8-Max generally available on August 3, 2026, releasing the flagship model with a published benchmark table, per-token pricing, and a production API, and has now officially published the open weights on Hugging Face and ModelScope.

The model uses a Mixture-of-Experts design with 2.4 trillion total parameters and about 95 billion active per token, handles a 1-million-token context, and accepts text, image, and video input, with output limited to text.

On Alibaba's own tests, Qwen3.8-Max leads Claude Fable 5, GPT-5.6, and Opus 4.8 on computer-use and long-document tasks, but trails Fable 5 by 12 points on the harder SWE-bench Pro coding test and finishes last on broad reasoning.

🖼️ Gemini lets you turn off AI watermarks LINK

Google's Gemini app now lets people turn off the visible spark watermark that shows up in the corner of images, videos, and songs made with its Nano Banana, Omni, and Lyria tools.

The option won't appear in countries where law requires keeping the mark, and Gemini still embeds invisible SynthID watermarks plus C2PA metadata so the media stays traceable even with the corner icon removed.

On the web, a new "Media watermark" section under the settings gear offers "On" and "Off" choices, with On staying the default, and the feature is reaching all platforms over the coming days.

⚖️ France court blocks under-15 media ban LINK

France's Constitutional Council has thrown out the country's planned ban on social media accounts for under-15s, which was set to take effect on September 1, ruling it violated free expression and privacy protections.

The Council said the ban swept too widely, covering TikTok, Snapchat, X, Instagram, YouTube, and messaging apps like WhatsApp, without proof that each posed established risks to the health and safety of minors.

Parliament had already weakened the law before its July 21 approval, dropping age-verification rules and penalties, so any revival would need a narrower scope and stronger privacy safeguards, marking a setback for Emmanuel Macron.

Other news you might like

• Waymo receives permission to offer rides in Sacramento and San DiegoLINK

• Ads Coming Soon to Apple MapsLINK

• Canva was the rare startup that grew fast and made money—then AI costs slashed its growth forecast by a thirdLINK

• Person Hides Prompt Injection in Legal Filing Telling AI to Side With ThemLINK

• Nvidia’s new $500B plan is risky but brilliant, especially for aging GPUsLINK

• Deepseek ships improved V4 Pro, open-sources its agent software, and raises API pricesLINK

• JPMorgan shuttered its banking relationship with predictions platform Polymarket: FTLINK

• OpenAI Staff Blame Rush to Ship for Rogue Agent HackLINK

• US courts will start publishing how often the government uses spywareLINK

• Apple sends new ‘Threat Notification’ alerts over mercenary spyware attacksLINK

• Self-driving trucks are officially testing on California highwaysLINK

• PBS station fears losing 50TB of data after being ghosted by cloud storage providerLINK

• Trump family-backed crypto firm World Liberty gets conditional bank charter approvalLINK

🧰 Trending tools

Inferock Bench: a local proxy for OpenAI, Anthropic, Gemini, and OpenRouter calls that tracks token usage, failures, and retries to reveal billing discrepancies and overpayment.LINK

isolate.video: converts raw screen recordings into polished product videos by adding automatic motion zoom, AI-generated music, and spotlight effects.LINK

Zetik: an AI agent team that collects, filters, and analyzes podcasts, papers, code, tweets, and news to deliver real-time intelligence briefings on topics you choose.LINK

SalesCloser.ai: an AI sales rep that qualifies leads, schedules calls, runs live demos, handles objections in 32 languages, then updates your CRM automatically.LINK

Big Mike: texts betting picks across MLB, NFL/CFB, and NBA/WNBA with specific lines and sportsbooks, plus roster-based start/sit advice via Sleeper or ESPN league integration.LINK

GLM-5.3: a free platform for testing Z.ai's MIT-licensed GLM models, Base, Reasoning, and Rumination, through a simple, distraction-free chat interfaceLINK

📚 Trending papers & reports

AI code reviews now work best when a separate watchdog agent, not the same model that wrote the code, checks it at fixed checkpoints, since testing across twelve projects found each of nineteen critique angles caught bugs the others missed.LINK

AI-generated test verdicts often check whether software works without ever consulting a written spec, since a review of 83 studies found just over half rely purely on what the model learned, making its judgments harder to defend when challenged.LINK

Physics AI models can be compressed far more aggressively, in some cases by orders of magnitude, without losing accuracy, making scientific simulation tools cheaper to run and easier to deploy.LINK

Time-series pattern detection spots the specific warning signs unique to each data sequence rather than forcing one generic pattern on everything, beating existing methods across 128 benchmark tests while staying easy to explain.LINK

Attention shortcuts in AI models can merge two of the three internal signals models use to focus on information, cutting the memory needed during use by up to ~97% with only a small ~3 point drop in accuracy, enabling AI to run practically on phones and other on-device hardware.LINK

See you tomorrow for a new dose of ☕️ Techpresso!