Mistral Series D: Samsung Led, but Nobody Discloses How Much Chipmakers Paid

Mistral raised €3 billion in September 2026 at a post-money valuation above €21 billion, and Samsung led it. The tempting reading is that chipmakers now write the big AI cheques. The record is thinner than that. Mistral names its investors but no cheque sizes, so the strategic share of this round cannot be measured from its own announcement.

Short answer: Samsung led Mistral's €3 billion Series D in September 2026 at a valuation above €21 billion, with Nvidia and ASML among the participants. Strategic investors that also sell to the company supply a large share of late-stage AI money, at least 23% to 25% where the figures are public. Their cheque sizes mostly are not.

What Mistral's Series D actually disclosed

Start with what the primary source says, because the secondary coverage drifts from it. Mistral's own announcement, dated 8 September 2026, is short on numbers and long on names.

The terms, as Mistral states them

The round is €3 billion, a Series D. Mistral calls it "the largest equity fundraising round ever completed by a European technology company." The valuation is "a post-money valuation of more than €21 billion." No exact post-money figure appears, and no pre-money figure appears anywhere we looked.

Samsung Electronics led. The co-leads were Scaleup Europe Fund, managed by EQT, and PSG Equity, an existing investor. TechCrunch and Tech Funding News report the same structure. Tech Funding News adds that this is the Scaleup fund's first investment since EQT became its manager.

New investors are Advent, funds and accounts managed by BlackRock, and the Grand Duchy of Luxembourg. Existing investors that took part include a16z, ASML, Bpifrance, DST Global, General Catalyst, Index Ventures, Lightspeed, NVIDIA and Salesforce Ventures. Mistral gives no amount for any of them.

The step-up from the Series C

Mistral's Series C post from 9 September 2025 records €1.7 billion raised at €11.7 billion post-money, led by ASML. Twelve months later the valuation is at least 1.8 times higher. That is a large step for a company that, on its own reported figures, was still building its revenue base.

Revenue is where the primary record goes quiet. Neither Mistral's post nor TechCrunch gives a revenue or ARR figure. ActuIA reports a company-declared target of €1 billion in recurring revenue by year-end, set against roughly $300 million annualised in September 2025. It cites no primary source for the second number, so we do not build on it.

Samsung's investment in Mistral: confirmed, reported, unknown

The Samsung Mistral investment is the headline, and it is also the least documented part of the deal. It helps to sort every claim into three bins.

What is confirmed

Samsung led. Nvidia and ASML participated. Mistral's post lists ASML among its "125+ global enterprises," so ASML is a customer as well as a shareholder and past lead. Tech Funding News and ActuIA both call ASML, Samsung and Nvidia "both suppliers and shareholders." That is the press framing, not Mistral's.

Nvidia's supplier role is the easiest to document. Tech Funding News names it as a hardware partner and mentions an $830 million debt facility from March 2026 used to buy Nvidia GPUs. ActuIA reports that Mistral's first data centre will house about 13,800 Nvidia GB300 GPUs.

What is reported but not confirmed

ActuIA cites Sifted for a Samsung commitment of up to €1 billion. We did not open the Sifted article, and Mistral, TechCrunch and Tech Funding News give no Samsung figure. ActuIA also cites Korean press for talks on high-bandwidth memory supply and on putting Mistral models into Samsung devices. In its own words, "neither party has confirmed the second aspect."

There is one more caution from the same article. ActuIA notes that "a stake in the company does not contractually guarantee any delivery." A shareholder that supplies you has an interest in your success, but that interest is not a purchase order.

2026 AI mega-rounds, split into strategic and financial money

The Mistral angle invites a comparison. We took 5 of the largest 2026 announcements and sorted the named investors by type, using only what each company or a named outlet published. The table is what we could verify. It is not a complete census of 2026 rounds.

Sources: Mistral, Anthropic, SiliconANGLE and Quartz, The Next Web, StorageNewsletter, listed in the references. SoftBank is shown among the financial leads. Quartz calls it a strategic partner, so a stricter reading would raise the OpenAI strategic share.

How we classified investors

We called an investor strategic if it has a commercial relationship with the company: it sells to it, buys from it, or supplies compute or distribution. Everyone else is financial. That includes sovereign funds such as GIC, Temasek and MGX, and asset managers such as BlackRock and T. Rowe Price.

The line is blurry. Salesforce Ventures is strategic by type, though we found no confirmed commercial tie to Mistral. The Grand Duchy of Luxembourg is a customer per ActuIA and a state investor. We kept both calls conservative and flagged them where they matter.

How much of the money is strategic, measured as a floor

No source we opened gives a clean strategic share for any of these rounds. What we can compute are floors: the strategic dollars that are named and sourced, divided by the round.

Why these numbers are floors, not shares

Anthropic's Series H post says the round "includes $15 billion of previously committed investments from hyperscalers, including $5 billion from Amazon." Fifteen divided by 65 is 23%. It excludes Micron, Samsung and SK hynix, which Anthropic names as strategic infrastructure partners without amounts.

For OpenAI, the $30 billion from Nvidia reported by The Next Web is 25% of the $122 billion round. That leaves out Amazon, whose amount we could not confirm in the articles we opened, and Microsoft. The real strategic share is higher in both cases. By how much, nobody outside the cap table can say.

Mistral cannot be computed from primary sources at all. If the Sifted figure of up to €1 billion for Samsung is right, it is one third of the round. We show that bar as dashed because it is unverified.

Nvidia as a venture investor with a product to sell

One strategic investor appears in nearly every row, and it deserves its own section. The Next Web, relaying CNBC reporting based on public filings, says Nvidia committed more than $40 billion to AI equity in the first four months of 2026. About $30 billion went to OpenAI in late February. The rest spread across 7 multi-billion-dollar deals in public companies and roughly 24 private rounds, Mistral among them.

Two details matter for pricing. First, the same article says Nvidia's commitments to Corning and IREN were warrants or structured instruments, "with cash outflow timed at Nvidia's discretion." Headline amounts are not always cash on the day of the announcement. Second, it cites last fiscal year's figure of $17.5 billion into private companies and infrastructure funds, taken from the 10-K, so the 2026 pace already exceeds it.

Nvidia's chief financial officer, Colette Kress, has described the logic on an earnings call, as paraphrased by The Next Web: invest where needed to make sure compute is built around its hardware. We take that at face value. It is a coherent business reason. It is also the definition of an investor whose return function differs from a fund's.

The wider circular-financing debate is covered in the explainer on how the circular AI deals work, and the older precedent in the comparison with Lucent and Nortel vendor financing. This post does not re-run that argument. It asks a narrower question about the price that results.

What strategic money does to price discovery in AI rounds

Price discovery is the process by which a market finds what an asset is worth. In a private round, the lead investor effectively proposes the price and the rest of the syndicate accepts or declines. Who leads therefore matters.

A financial investor prices a stake on expected exit value, discounted for risk. A strategic investor can add a second term: the value of sales, supply security or platform position that the relationship protects. A chipmaker whose product the company buys has a reason to pay a little more than a fund would.

Our view is that this does not make the price wrong. It makes the price answer a different question. A valuation set by a supplier tells you what the company is worth to that supplier, which is a ceiling on its worth to an arm's-length buyer, not an estimate of it.

The second effect is the one buyers and later investors feel. Valuations from one round become comparables for the next, and for secondary sales, employee equity and acquisition talks. A strategic-led price flows into all of them without a label saying how it was set.

We would not treat any strategic-led AI valuation as a clean comp. When the market turns, the structure of the cap table decides who absorbs the reset, a point examined in our look at AI down rounds and cap table terms. For the lab-level version of the same question, see the breakdown of OpenAI's valuation against its run rate.

Where this argument is weakest

Four things cut against the thesis. A post that skipped them would be selling the headline rather than the evidence.

Financial investors still lead most of these rounds

Look at the table again. Anthropic's Series H and G were led by Altimeter, Dragoneer, Greenoaks and Sequoia, then GIC and Coatue. OpenAI's round was co-led by SoftBank with a16z, D.E. Shaw, MGX, TPG and T. Rowe Price. Mistral is the only row in which a supplier takes the top slot. "Chipmakers now write the big AI cheques" is true of Mistral and of Nvidia's own book, and not clearly true of the market.

Strategic investors may know more

A supplier sees the customer's order book. If Nvidia or Samsung invests, that can be a signal that real demand sits behind the valuation, because they have information a fund lacks. That is a legitimate counter-argument, and nothing in the public record lets us rule it out.

We cannot see who set the price

Mistral's announcement names Samsung as lead and EQT's fund and PSG Equity as co-leads. It does not say which of them negotiated the valuation. Financial co-leads took part in the same round, which is weak evidence that the price cleared with them too. We would not push this reading far.

The evidence base is secondary in places

The OpenAI figures come from news coverage because OpenAI's own page returned an access error when we tried to read it. The Nvidia total is CNBC's, relayed. The Samsung amount is Sifted's, relayed again. Where we could not open a primary document, the table says so. Treat the percentages above as arithmetic on reported numbers.

How to use a strategic valuation as a founder, buyer or investor

The practical question is what to do differently once you accept the argument. The table below sets a weak habit against the version that works, and its last row concedes where the weak habit is fine.

If you are raising, our position is simple: take the strategic investor's money if the commercial terms help you, and do not let their price become your reference point for the next round. If you are a buyer, a vendor's last valuation tells you very little about its survival. Cash on hand and gross margin tell you more, a theme in our analysis of AI's share of venture capital and in the piece on sovereign AI funds and compute demand.

Frequently asked questions

Where to start this week

Pick one vendor you depend on and open its latest funding announcement. Write down who led, which investors also sell to or buy from it, and how many of the dollars are stated. If the answer to the last question is none, record that as the finding.

Then do the same for the next competitor round that lands in your inbox. After three of these you will have your own table, and it will tell you more than any headline valuation does.

Related reading