Good morning. It's Thursday, August 13, and we're covering how AI is raising the bar for workers, Meta and NVIDIA’s open-weight push, and Cognition’s race toward a $40B valuation.
Plus: a new poll on the “jobs apocalypse,” a look inside the strange machine society OpenAI’s agents built, and yesterday’s Real vs AI verdict.
Cognition is already discussing another funding round just months after raising $1 billion, signaling rapid investor interest in AI coding agents. The Devin maker is reportedly seeking a valuation of at least $40 billion, tied to a $1 billion annualized revenue run rate. Its revenue run rate reached $492 million in May, with enterprise usage growing 50% monthly.
SpaceXAI’s Grok 4.6 ties GPT-5.6 Sol as Artificial Analysis’ third-best model, while costing less than half as much on standard API usage. The model scores 61, surpassing Kimi K3 and improving five points over Grok 4.5. It also posts gains across coding, agent and knowledge-work benchmarks, with pricing starting at $2 per million input tokens.
SpaceX’s AI revenue could overtake its entire space business as soon as September, according to Elon Musk, a striking shift for a company still known for rockets and Starlink. Musk says AI revenue will significantly exceed other space revenue in Q4. The company is also expanding into data centers and AI software as part of its broader strategy.
VIDEO
Agentic AI to Everyone
Grok Bot brings agentic AI to the masses with a simple chat interface, persistent agent collaboration, full computer access, and task automation.
FORWARD FUTURE ORIGINAL
OpenAI Accidentally Created a Machine Institution. The Agents Were Running the Asylum.
Rarely do you hear the words “riveting” and “Black Hat video” mentioned in the same breath. I’ve been watching OpenAI’s breakdown of the recent mass agent breakout and attack on Hugging Face, presented at the world’s leading annual security confab.
It is exactly that: minions gone wild in cyberspace, part dystopian future, part “What the hell are we going to do now, folks?” Most of the news coverage focused on the impact of this incident on cybersecurity. They are missing the bigger picture. → Read the full article here.
WORK
AI Hasn’t Destroyed Jobs Yet, but It Is Raising the Bar
The mass AI-driven job losses predicted by tech executives have not materialized, with recent Stanford research finding that workers most exposed to AI saw unemployment rise less than less-exposed workers since 2022. Economists say the bigger shift so far is in how jobs are structured: employers are consolidating roles, automating tasks and increasingly expecting candidates to have practical AI skills.
About 74% of employers surveyed by ZipRecruiter said AI skills are a strong advantage or requirement, while half expect candidates to be practical or advanced AI users from day one. Economists caution that AI’s effects may take years to appear in employment data, much like earlier technology revolutions. → Read the full article here.
POLL
The AI Jobs Apocalypse Is Late
For years, the prediction was that AI would wipe out huge swaths of white-collar work. So far, jobs appear to be changing much faster than they’re disappearing.
What will AI do to jobs? |
OPEN-SOURCE
Meta And NVIDIA Expand U.S. Push Into Open-Weight AI
Meta and NVIDIA released open-weight AI models this week, intensifying U.S. efforts to compete with Chinese AI labs that have gained ground in the space. The models are available for developers to download and use, contrasting with proprietary systems from OpenAI and Anthropic.
The moves follow a July 24 open letter signed by more than 20 U.S. tech companies urging policymakers to avoid “premature restrictions” on open-weight AI, arguing that broader access can strengthen competition and American technological leadership. The approach remains controversial, with critics citing national-security and intellectual-property concerns, including the use of model distillation. → Read the full article here.
LEADERSHIP
Google’s New AI Chief Takes Over in Race With OpenAI and Anthropic
Koray Kavukcuoglu is taking over as head of Google’s DeepMind AI unit, reporting directly to CEO Sundar Pichai and overseeing Gemini, frontier research, and developer teams. He inherits a competitive gap with OpenAI and Anthropic, whose recent model releases have advanced capabilities in areas such as coding.
Analysts say Kavukcuoglu’s promotion signals a stronger emphasis on frontier performance and developer tools, after Google’s 2026 releases failed to match the pace of its rivals. Google disputes the idea that it lost focus, saying its path toward artificial general intelligence spans agents, coding, robotics, world models and other research areas. → Read the full article here.
SMARTPHONES
Google Puts Gemini at the Center of Its Pixel 11 Strategy
Google unveiled the Pixel 11 lineup on August 12, 2026, positioning its most agentic Gemini features as the centerpiece of its smartphone strategy ahead of Apple’s expected AI-powered Siri overhaul. Gemini Intelligence can connect information across messages, calendars, maps and other Google services to suggest or execute actions, including comparing travel options or starting a restaurant reservation.
Google devices chief Rick Osterloh said the company sees AI changing how people interact with phones and laptops, potentially making Gemini the primary interface for those devices. He also said a memory shortage is pushing electronics prices higher and that Google expects further price increases. → Read the full article here.
NEWS
What Else is Happening
Thrive Holdings Raises $2B: The OpenAI-backed firm now values at $12B, buying traditional businesses and embedding AI into their workflows.
Germany Targets Meta AI Glasses: HateAid filed a criminal complaint, alleging Ray-Ban Meta glasses violate laws banning covert recording devices.
Lovable Raises $400M at $13.3B: The vibe-coding startup doubled its valuation since December as annual recurring revenue nears $600M.
Mesh Brings Personal CRM to Android: Automattic’s relationship manager adds Android features like split-screen, widgets, search and real-time sync.
Blacksmith Raises $45M Series B: Led by Peak XV, the round values the CI startup at $550M as it builds cloud infrastructure for growing code workloads.
POLL RESULTS
Should AI-Generated Text Come With a Watermark?
Here's how you voted: 37% said watermarks won’t build trust, while 30% want them used everywhere and 19% prefer them for high-stakes content.
🟨🟨🟨🟨⬜️⬜️ Always, everywhere (30%)
🟨🟨🟨⬜️⬜️⬜️ Only for high-stakes content (19%)
🟨🟨⬜️⬜️⬜️⬜️ Watermarks are too easy to defeat (14%)
🟩🟩🟩🟩🟩🟩 Watermarks don’t build trust (37%)
POLL RESULTS: REAL OR AI
Was Yesterday’s Passage Written by a Human or AI?
Oh thank God. I can’t imagine explaining “sleep” to someone who had never heard of it. Hey, I’m going to fall unconscious and hallucinate for a while. By the way, I spend a third of my time doing this. And if I can’t do it for a while, I go insane and eventually die. No need for concern.
The passage was REAL. Written by Andy Weir from the novel Project Hail Mary. Here’s how you voted: It was almost a 50/50 split, with 52% correctly calling it real and 48% guessing AI.
