In partnership with
Welcome to OHMO AI. A US start-up is renting out humanoid robots to clean homes, and the catch is in the small print. Mistral has closed a round that leaves it worth far more than it was a year ago.
China has named how much AI computing it wants by the end of the decade, and the US jobs numbers show no AI layoff wave.
In today's newsletter:
Tau Robotics rents out humanoid robots as home cleaners, one hour at a time
Mistral raises €3 billion in a Samsung-led round at a valuation of more than €21 billion
China wants to more than quadruple its AI computing by 2030
US jobs data shows no AI-driven surge in layoffs
LATEST IN AI
TAU ROBOTICS
A US start-up is renting out humanoid robots to clean homes
Image source: Tau Robotics
OHMO AI: US start-up Tau Robotics has started hiring out its humanoid robots as home cleaners, sending them into people's homes to do the varied, messy domestic work that robotics firms have spent years trying to hand to machines. The service has been running for about a month.
The details:
For now, a robot can work for only one hour at a time, so a booking buys a single hour of cleaning.
A session with one robot costs $30, about €25.7, and sending two to the same home costs $60.
In its first month the service has cleaned more than 50 apartments and condominiums across the city it operates in.
Why it matters: The price is the part that makes the headline, and the one-hour cap is the part that decides what this is. A cleaner who comes for a morning does more than a machine that stops after an hour, whatever the rate, so at $30 this reads like a ticket to watch a humanoid in your kitchen rather than a replacement for anyone. The tell will be the session length: if Tau lets the robots run past the hour, cleaners have a competitor; if it doesn't, the people in more than 50 apartments paid for a show, not a cleaner.
PRESENTED BY ROKU
Blu Dot surpasses 2,000% ROAS with self-serve CTV ads
Home furniture brand Blu Dot blew up on CTV with help from Roku Ads Manager. Here’s how:
After a test campaign reached 211,000 households and achieved 1,010% ROAS, the brand went all in to promote its annual sales event. It removed age and income constraints to expand reach and shifted budget to custom audiences and retargeting, where intent was strongest.
The results speak for themselves. As Blu Dot increased their investment by 10x, ROAS jumped to 2,308% and more page-view conversions surpassed 50,000.
“For CTV campaigns, Roku has been a top performer,” said Claire Folkestad, Paid Media Strategist, Blu Dot. “Comping to our other platforms, we have seen really strong ROAS… and highly efficient CPMs, lower than any other CTV partner we've worked with.”
Using Roku Ads Manager, the campaign moved from a pilot to a permanent performance engine for the brand.
MISTRAL AI
Mistral raises €3 billion in a Samsung-led round that values it at more than €21 billion
Image source: OHMO AI
OHMO AI: Mistral has raised €3 billion in a Series D that values the French company at more than €21 billion once the new money is counted. Samsung Electronics led the round, and Mistral is pitching itself as the European alternative to OpenAI and Anthropic.
The details:
The co-leads are the Scaleup Europe Fund, managed by EQT, and PSG Equity, which had already invested in Mistral.
Advent, funds and accounts run by BlackRock, and the Grand Duchy of Luxembourg all joined as new investors.
A year ago Mistral was valued at €11.7 billion, after a round led by ASML, the Dutch maker of chipmaking equipment.
Arthur Mensch, Mistral's chief executive, said earlier this year that he expects annual recurring revenue, the yearly rate its current sales run at, to exceed $1 billion this year.
Why it matters: Mistral is pitching itself as the European alternative to OpenAI and Anthropic, and this cap table reads like the pitch: a memory chip giant in front, a Scaleup Europe Fund managed by EQT beside it, and the Grand Duchy of Luxembourg joining as a new investor. A government on the shareholder list is the detail to hold onto. If more European states follow Luxembourg in, the alternative stops being a slogan and starts being who each of them buys from.
CHINA
China wants to more than quadruple its AI computing by 2030
Image source: OHMO AI
OHMO AI: China's Ministry of Industry and Information Technology has told the country to reach 9,800 exaflops of intelligent computing capacity by 2030, in a five-year industry plan released on Monday. An exaflop is a unit of raw computing speed, and the plan pairs the target with a spending figure to match.
The details:
The plan calls for 3.8 trillion yuan, about $532 billion, of cumulative spending on information infrastructure between 2026 and 2030.
By the ministry's own count, capacity had reached 2,185 exaflops by the end of June, up 177 per cent on a year earlier, and China had built 52 intelligent computing facilities, each fitted with more than 10,000 accelerator cards, the chips that do the AI maths.
Getting from the June figure to the 2030 target means more than quadrupling capacity.
Why it matters: Against a year in which capacity rose 177 per cent, more than quadrupling by 2030 reads less like a stretch than a floor. What changes is that everyone upstream of a data centre, from chipmakers to grid operators, now has a state buyer's number to plan around, and it looks like less than the pace they have been supplying at. If the next count shows the June pace holding, the 2030 target gets passed early, and the plan's job was never the target, it was the 3.8 trillion yuan.
PRESENTED BY ATTIO
Introducing The First Agentic CRM
Get revenue agents, workflows, and automations across every stage of your motion. Access customer data in real time through Attio's web app, MCP, API, and SDK.
Then Ask Attio anything about your business and get instant answers.
It's the CRM that runs the work behind every win.
US JOBS
The latest US jobs data shows no AI layoff wave
Image source: OHMO AI
OHMO AI: The latest US employment data offers no sign that AI is quickly displacing workers, with employers choosing to adapt rather than cut staff. Across the first eight months of 2026, planned job cuts dropped to a four-year low while hiring plans reached their highest level since 2023.
The details:
Both figures come from Challenger, Gray & Christmas, a firm that helps laid-off workers find new jobs and tracks announced cuts, in data released on Thursday.
Apollo Global chief economist Torsten Sløk said on Thursday that there is still no evidence of AI replacing workers.
Ben Zipperer, an economist at the Economic Policy Institute, said AI's impact on jobs has so far been more limited than some of the doomsday scenarios predicted.
Why it matters: If employers are adapting rather than cutting, the cost of AI hasn't left the labour market. It has moved off the payroll line and onto the person who keeps the job and now learns the tool on top of it, and nothing in these figures measures that. For most readers the question this year isn't whether a machine takes the job; it's what the job they keep turns into.
AI GUIDES
Get an answer you can act on by telling AI what to leave for you
Paste some customer feedback and ask for patterns and you get a confident summary that quietly fills the gaps with guesses. Write the request in four parts, and end with what the AI must not decide, and you get the same analysis with the guesses labelled, the holes flagged and the final call still in your hands. It takes one extra minute and works in any chatbot.
Copy this, swap in the bracketed parts, and paste it in.
Goal: Read the material below and tell me the main patterns in it.
Context: This is [what you pasted, e.g. 40 survey replies from customers who cancelled last month]. I need it to decide [the decision you are making].
Output: Give me the top [3] patterns as a numbered list. Under each one, put what you actually saw in the material and, separately, what you are assuming. End with [3] concrete actions I could take.
Boundary: Use only the material I pasted, nothing from general knowledge. If anything is missing, unclear or contradictory, flag it at the top instead of fixing it. Do not draft replies, and do not state a conclusion the material cannot support. Leave the final decision to me.
Material:
[paste your feedback, notes or data here]The four parts give the AI a job, the situation, the shape of the answer and the line it must not cross. The boundary is what keeps guesses labelled and the decision with you.
COMMUNITY
Featured Reader
Every newsletter, we feature how a reader is using AI to work smarter, save time, or make life easier.
Today's reader is Paweł W. from Lublin, Poland:
“I told the whole board in March that ai wasn't going to count our water. 60 plots in the ogród działkowy, readings on paper, dues on a calculator at the kitchen table, three evenings gone every June.
Now it's a small page Claude built in the browser after I typed the columns as I'd say them: plot number, last reading, this reading, rate, who shares a meter with whom. It asked about the shared meters before I'd said they were the problem which is when I shut up. Readings go in on my phone at the plot, it splits each share and prints the noticeboard list. Dues still come by BLIK to the secretary's phone and I type those in myself, it's not getting near money. The board has not let me forget March.”
Want to be in the next newsletter? Tell us how you use AI here.
AI Meme
Go Deeper
Instagram: The AI page they pray you’ll never find.
Facebook: For the ones who never left Facebook.
