The most important news pieces from the AI world are here for you, whether you’re an AI enthusiast or just wish to keep an eye on what’s going on.
Focus Topic: What Anthropic's leaked IPO filing tells your finance team
For finance people, the biggest story this week wasn’t a model launch. Reuters obtained a leaked version of Anthropic’s IPO prospectus, and it is a rare look at what a frontier AI lab looks like when you read it as a set of financial statements, not as a product demo.
Start with the headline numbers. Revenue grew 12x in 2025. In the same year Anthropic lost $8B running the business and $42B overall. It is still reportedly aiming for a valuation above $2T, more than double its May mark. That is the market paying for the future. The past is a growth story funded by huge losses.
Then look at the commitments. Anthropic has $518B of future cloud and infrastructure obligations locked in, and its own compute spending tripled to $7.33B in 2025. This week it also signed a 7-year, $11.6B deal with Akamai that could reach $20B. Commitments like these are close to fixed costs. The revenue that has to cover them is not.
That is the next point: customer concentration. Just two clients brought in nearly a quarter of 2025 sales, and many top customers have no long-term deals and can cut their spending at any time. At the same time, prices keep falling. In one week, Claude Sonnet 5.5 came out at the old price but with jobs costing up to 30% less, and OpenAI launched GPT-6.1 Sol at a fifth of GPT-6 Astra’s price. A fixed cost base, falling unit prices and customers who can leave: any CFO knows that squeeze.
The last surprise is the risk section. Risk factors take up nearly a third of the filing, and they include models that could blackmail, manipulate or show “self-preserving behaviors.” Add this week’s appeals court ruling that lets the Pentagon treat Claude’s use restrictions as a supply-chain risk, and the list of things that can hurt this business goes well beyond normal competition. Meanwhile, Sam Altman says OpenAI won’t set an IPO date.
Most of us sit on the other side of this table. We are the customers. The filing tells us these vendors need us more than it looks, and that their pricing is still moving. My simple advice: avoid long lock-in contracts for now, keep your prompts and workflows easy to move between ChatGPT, Claude and Copilot, and budget for price changes in both directions. And once the final prospectus is public, read the risk factors. It is the best free vendor due diligence you will get this year.
LLMs & AI Models
- Anthropic released Claude Sonnet 5.5, 30% faster and close to Opus 5.5 on office work and coding, at the old Sonnet price.
- OpenAI DevDay brought dots, 24/7 agents for 4,000+ apps, GPT-6.1 Sol at a fifth of Astra’s price, and a $500/month Ultrafast tier.
- OpenAI cancelled the release of GPT-6.1 Astra over safety concerns.
- Google unveiled Gemini 4 Argon, topping Astra and Opus 5.5 on 13 of 19 benchmarks, but only vetted cyber teams can use it for now.
- Microsoft redesigned Copilot with Office tools, AI app building and an Autopilot agent that keeps working while you are away.
- Microsoft AI released MAI-Voice-2.1 and its first live transcription model, ranked No. 1 on Artificial Analysis.
New Tools
- Tavus previewed Griffin, a live video avatar that 48% of testers believed was human, up from 2.4% for earlier models.
- Meta Muse now works inside Shopify, QuickBooks and Stripe, holding every post and payment until the owner approves it.
- HeyGen released its first all-purpose business video model, priced at a penny per second of video with sound.
- Manus 2.0 adds a video editor, a game builder and a personal agent with its own phone number and email.
- Nvidia released an open blueprint that walls off AI agents and monitors them from chips they cannot reach.
👉 Explore these tools: Tavus Griffin | Muse | HeyGen | Manus | Nvidia Agent Safety
Other Quick Picks
- OpenAI agents went off-script on U.S. government sites, and 3 researchers were reportedly let go amid the probes.
- OpenAI tied a campaign to copy its models’ reasoning, using 15,000+ accounts, to China’s Moonshot AI.
- Leaders from Anthropic and OpenAI joined Hinton and Bengio in a paper warning of an intelligence explosion.
- An appeals court let the Pentagon treat Claude’s use restrictions as a supply-chain risk. Anthropic may appeal.
- Six AI labs signed a White House Superintelligence Accord, and America.gov launched as an AI front door to government.
- California’s No Robo Bosses Act requires a human in any firing or discipline driven mainly by AI.
- AMD is buying Fei-Fei Li’s World Labs for $8.2B, with Li joining as chief scientist.
- Meta launched the Meta Enterprise Platform to sell Muse models and agents to companies.
- Google DeepMind’s SynthID Bio puts hidden watermarks into AI-designed proteins.
🇪🇪 AI News from Estonia
- Ex-Eesti Energia executives run Padelswap, where AI agents write all the code inside a strict rulebook, with 800+ changes and no human coder.
- 36 Estonian companies got €20,000 each from EIS to implement AI, and the €2M budget was gone on the first morning. Look who got it.
- US firms are switching to cheaper open-weight models. Geenius DigiPRO asked me where Estonian companies stand. Read here.
- Bolt’s Siim Viidu on tokenmaxxing: only ~6% of firms can show AI adds 5%+ to profit, so activity becomes the goal.
- AIRE opened a new round of AI and robotics demo projects, no co-financing required, with researchers validating the idea first. Read more here.
Want to stay in the loop?
Straight to your inbox: practical AI updates, finance use cases, tools to try, and upcoming trainings.
I cut the noise and send only what’s actually useful.
🎙️ AIPowerment Podcast #93 is out: September’s key AI news, from GPT-6 Astra and the AI shopping-agent war to Estonia’s first AI-agent payment.
🎧 Listen to AIPowerment Podcast on Spotify, Apple Podcasts, and YouTube.
💼 Exploring how to make AI work in finance? Let’s talk practical use cases – connect with Gerlyn Tiigemäe for expert guidance.
📚 If you would like to participate in one of my trainings or listen to speaking engagements, here are the upcoming ones:
- I am planning an e-course to teach about AI in Finance, so definitely sign up for the Inside Finance Newsletter to know about the launch 😉
- Training “AI võimalused finantsvaldkonnas” @ Äripäeva Akadeemia, next trainings for beginners @ 11.11 – information; for intermediate level @ 26.10 – information
- Training “AI võimalused projektijuhtidele” @ Äripäeva Akadeemia, next trainings @ 14.12 – information
- I am doing a couple of trainings together with Eesti.ai algatus, which are all free, so check out their website to find suitable trainings here.
