#Mistral€3BRahoitus
About Mistral€3BRahoitus
Mistral sulki 3 miljardin euron Series D -rahoituskierroksen yli 21 miljardin euron jälkeisellä arvolla, mikä on suurin osakerahoitus eurooppalaiselle teknologiayritykselle. Samsung, eurooppalaiset rahastot ja PSG Equity johtivat kierrosta, ja Nvidia sekä ASML jatkoivat mukana. Tuotot käytetään mallin tutkimukseen ja kehitykseen, laskentatehoon, infrastruktuuriin ja kansainväliseen laajentumiseen. Mistral tähtää hallituksiin, rahoitukseen ja valmistukseen avoimilla painoilla, paikallisella käyttöönotolla ja datan suvereniteetilla. Samsung, ASML ja Nvidia samassa omistuspohjassa yhdistävät mallit, sirut ja valmistuslaitokset yhdeksi teolliseksi kokonaisuudeksi.
Suosittu
Viimeisin
Mistral€3BRahoitus Suositut postaukset
Mistral AI has raised €3 billion at a valuation of more than €21 billion from investors led by Samsung
CEO Arthur Mensch tells @TomMackenzieTV what the money will be used for#ZECBreaksIntoTop10 #SamsungHynix10DaySupply #OracleAdobeEarnings
Three founders. About $5.5 billion between them. Three years. And unlike Hugging Face last week, they did not sell.
Mistral closed $3.5B at a $24.4B post-money valuation on Tuesday, led by Samsung. Largest equity round ever completed by a European technology company. Five days after Nvidia bought French-founded Hugging Face outright for $12.93B.
The round:
- $3.5B raised at more than $24.4B post-money
- Samsung Electronics led. Co-leads: the EQT-managed Scaleup Europe Fund and existing investor PSG Equity
- New money: Advent, BlackRock funds, and the Grand Duchy of Luxembourg
- Returning: a16z, ASML, General Catalyst, Lightspeed, Nvidia, Salesforce Ventures, Index, DST, Bpifrance, BNP Paribas, Eurazeo
- Macron posted about it, framing France and South Korea as building a third way in AI
The ladder, three years and four rounds:
- June 2023 seed: $120M at a $280M post. Four weeks after founding, on a seven-page deck, with no product. Largest seed in European history
- December 2023 Series A: $450M at roughly $2.3B, a16z leading
- June 2024 Series B: $700M at $6.7B, General Catalyst leading
- September 2025 Series C: $2B at $13.6B, ASML leading and taking roughly 11%
- September 2026 Series D: $3.5B at $24.4B, Samsung leading
- Plus $840M of debt in March 2026 for its first large data centre near Paris
- Nearly doubled in twelve months
Who made money, working the dilution round by round:
- Lightspeed and the seed syndicate: bought in at $280M. Roughly 46x net of dilution. The seed took 43.75% of the company, so the dollars behind that multiple are large
- a16z at the Series A: about 6.9x in under three years
- General Catalyst at the Series B: about 2.65x
- ASML at the Series C: about 1.5x on $1.5B in twelve months, and it holds close to 9.4% post-dilution, worth roughly $2.3B
- Samsung at $24.4B: par, for now
- Nvidia is on this cap table and also just bought Hugging Face outright. It owns a piece of both French AI champions
Founder wealth. Arthur Mensch, ex-DeepMind, is CEO. Guillaume Lample, ex-Meta, who led LLaMA, is chief scientist. Timothee Lacroix, ex-Meta, is CTO.
- Working the dilution stack, roughly 30% of original founder ownership survives the four rounds
- A European option pool, which Index Ventures data puts nearer 10% than the American 20%, leaves a retention factor around 0.26
- If the three started with 85% to 90%, they hold roughly 22% combined today
- That is about $5.5B across three founders, so roughly $1.8B each, with Mensch likely higher as CEO
- Range: $1.5B to $2.3B apiece. No cap table has been published, so this is modeled, not disclosed
The business under it:
- Mensch told CNBC the company will pass $1B in ARR before the year ends
- At $24.4B that is roughly 24x forward revenue. Cheaper than Perplexity at 40x, cheaper than Anthropic's forward multiple
- More than 125 enterprise customers across 20 countries, including Airbus, ASML, and HSBC
- Open-weight models from day one, the same category GLM and Kimi occupy
- The plan is to own the compute. Mensch: the amount Mistral owns will grow around 100% over the next five years
The frame worth keeping. Two French-founded AI companies, one week apart. Hugging Face, headquartered in New York, sold itself to the most valuable company on earth for $12.93B. Mistral, headquartered in Paris, raised the largest round in European history to stay independent and build its own data centres. Roughly $5B to three Hugging Face founders in cash and Nvidia stock. Roughly $5.5B to three Mistral founders in paper they still control.
Same nationality, same category, opposite endings. Sovereignty turned out to be a business model, not just a slogan, and being not-American is now a revenue line.
Ramp is in talks at a $60bn valuation, up from $44bn in June and $13bn eighteen months ago.
What. A. Day. I can't keep up!
Bloomberg reports early talks for roughly $1bn of primary capital. @tryRamp declined to comment.
The ladder since March 2025 is genuinely wild: $13bn, $16bn, $22.5bn, $32bn, $44bn, and now $60bn on the table. Six marks in 18 months, about $3bn raised in total.
A company that is free cash flow positive rarely needs $1bn.
Ramp says it passed $1bn in annualized revenue in June, with purchase volume up about 170% year on year in March at roughly 20 times the size it was when it last grew that fast.
So I'm guessing investors keep pre-empting, and Ramp is smart enough to take the money while the window is wide open.
That turns the private market into a slow-motion IPO. The cap table already reads like a public register: GIC, Ontario Teachers', Goldman, Morgan Stanley IM, D.E. Shaw.
Ramp started as a corporate card that killed expense reports. Then bill pay, procurement, treasury, and finally accounting.
But what makes Ramp interesting is they're one of the few genuinely positioned to finance the intelligence economy. R
Ramp has described intelligence as the third form of capital, paid by the token and invisible to every system built to manage cost.
Ramp now issues cards to AI agents with Visa and tracks token spend across providers. Its data on AI spend is cited in places like the FT. It calls tokens the fastest-growing cost in business, and no ERP has a field for them.
Whoever holds the ledger for machine spend holds the control point when agents do the buying.
That $60bn is around 40x estimated annualized revenue. Brex, same category, sold to Capital One in January for $5.15bn. So there's a lot of growth expectation priced in here.
Still.
You have to respect the sheer speed this company moves at. Outside of the AI labs, they're one of the fastest-growing things.
Ever.
Q-Day 2028, according to @IonQ_Inc
IonQ
The first complete, end-to-end fault-tolerant resource estimate for running Shor’s algorithm—the team used IonQ’s Walking Cat architecture to demonstrate how a specific application can be optimized for a trapped-ion quantum computer.
This work concludes that a 20,000-physical-qubit IonQ quantum computer, once developed, is expected to be able to break secp256k1, the 256-bit elliptic curve used by blockchain technology such as Bitcoin, in just under 26 days.
This architectural and resource-estimation study is consistent with IonQ's publicly stated roadmap, which targets systems with the relevant capabilities in the 2028 timeframe.
Watch Chris Ballance share the full view in our livestream today, September 8, at 12:30 PM ET →
Read the announcement →
Read the full paper →
#IonQ #QuantumIsNow #QuantumComputing #WalkingCat
