OpenAI has just put a very large number on the private market: $852 billion.
But the more interesting part is not the valuation itself. It is how that valuation was created.
OpenAI completed a roughly $7 billion employee tender offer, allowing current and former employees to sell shares back to the company. The deal was completed at the same $852 billion valuation as its most recent primary funding round.
At first glance, that may look like a straightforward confirmation that OpenAI’s valuation has held steady. But there is an important distinction for anyone following private markets: OpenAI itself funded the transaction rather than bringing in new outside investors.
That difference matters because a company-funded buyback tells us something different from a transaction where an independent investor is willing to put fresh money into the company.
And that is where the latest private-market activity gets particularly interesting.
OpenAI’s $852 billion valuation comes with an important caveat
The $7 billion tender gives employees a chance to turn some of their private shares into cash without waiting for an IPO or another funding round.
The valuation remained at $852 billion, matching OpenAI’s previous primary round.
That sounds like stability. But it does not necessarily mean the market has independently reassessed OpenAI at exactly $852 billion.
Why?
Because OpenAI was the buyer.
In a typical secondary transaction, an outside investor buys shares from an existing shareholder. The price is negotiated between someone who wants to sell and someone willing to invest their own capital.
That creates a useful market signal.
In OpenAI’s case, the company used its own cash to buy the shares. There was no new outside investor establishing a fresh price through the tender.
So the $852 billion figure is best understood as the price at which OpenAI chose to buy shares from employees, rather than a fresh third-party market clearing price.
That does not make the valuation meaningless. It simply means investors should be careful about comparing it directly with prices generated through other types of private-market transactions.
Why the buyer matters more than the headline
Private-company valuations can look much more precise than they really are.
A headline might say that a company is worth $852 billion, $20 billion or $11 billion. But behind every number is a transaction, and the structure of that transaction matters.
Ask three questions before taking a private valuation at face value:
- Who was buying? Was it the company, an existing shareholder, a venture fund or a new institutional investor?
- Why was the transaction happening? Was it a funding round, employee liquidity event or secondary sale?
- When was the price established? A valuation from several months ago may not reflect the company’s current performance.
This is especially important in private markets because there is no continuously traded stock price updating every few seconds.
A private company can operate for months while its last official valuation remains unchanged.
That is exactly why OpenAI’s tender is worth watching. The company has moved about $7 billion of shares, but the transaction has not produced a new outside investor price.
OpenAI is also moving closer to the public markets
The tender comes at an interesting point in OpenAI’s journey.
The company submitted a confidential draft registration statement to the SEC in June, although that does not mean an IPO is scheduled or guaranteed.
For employees, the tender offers another way to access liquidity before any potential public listing.
For the broader market, it raises a bigger question: how should investors think about OpenAI’s private valuation when the company is potentially moving toward the public markets?
A future IPO would introduce a very different pricing mechanism.
Public-market investors would have access to more information, the shares would trade continuously and the valuation would change with market conditions.
Until then, private transactions remain the main reference points.
And those reference points need context.
The bunq decision shows that fintech valuations are not just about growth
OpenAI is not the only company providing an interesting look into how private markets are being priced.
The Office of the Comptroller of the Currency rejected Dutch neobank bunq’s application for a US national bank charter in a decision dated August 4.
The rejection was not simply about paperwork.
The OCC raised concerns about bunq’s business and marketing plans, including whether they were realistic in the highly competitive US unsecured credit card market.
The regulator also questioned whether the company had adequately planned for the costs of building a US presence without significant name recognition.
Another issue was the proposed board’s understanding of the differences between European and US credit markets.
The decision is notable because bunq had already attempted to pursue a US banking charter before, withdrawing an earlier application in 2024.
The company has said it plans to try again.
Erebor is taking a very different path
The bunq decision comes as Erebor, the bank founded by Palmer Luckey and Joe Lonsdale to serve crypto, AI and defense companies, is reportedly discussing a roughly $1.5 billion fundraising round at an $8 billion valuation.
The round has not closed.
Still, the contrast is interesting.
One company has had its national banking charter application rejected, while another is reportedly attracting significant investor interest around a multibillion-dollar valuation.
The lesson is not that the OCC decision changes the valuation of fintech companies across the board. It does not.
Instead, it highlights something more practical: regulatory approval is part of the business model for financial companies, not just a box to check later.
For investors, the quality of a business plan, regulatory strategy, management team and understanding of the market can matter just as much as headline growth.
Notion shows another problem with private valuations
Then there is Notion.
Sacra estimates that Notion reached $865 million in annual recurring revenue in July, representing roughly 82% year-over-year growth.
That is a significant increase.
But Notion’s latest reported valuation is still based on an $11 billion tender completed in January 2026.
Put those two numbers together and the company is sitting at roughly 18 times the estimated annual recurring revenue.
There is an obvious limitation here.
The $865 million revenue figure is an outside estimate, not an audited company disclosure.
At the same time, the $11 billion valuation comes from a real transaction, but that transaction took place months ago.
So we have two different timestamps describing the same business.
The company may have changed considerably since the valuation was established, while the valuation itself has not yet been formally reset.
That is one of the defining quirks of private markets.
The private-market valuation problem: appraisal smoothing
There is a useful concept for understanding all of this: appraisal smoothing.
The idea is simple.
Public companies are repriced constantly because investors can buy and sell their shares every trading day.
Private companies do not work that way.
A private valuation typically changes when there is a significant event such as:
- A new funding round
- An employee tender
- A secondary transaction
- An annual valuation process
- A new strategic investment
Between those events, the previous valuation can remain in place.
That can make private-company valuations appear remarkably stable even when the underlying business is changing quickly.
Imagine a private company whose revenue jumps significantly over six months. If it does not raise capital or conduct another liquidity event, its official valuation may remain exactly where it was before that growth happened.
The business moves. The valuation does not.
Eventually, another transaction happens and the valuation may jump in one step.
This is why the date and structure of a private transaction can be just as important as the valuation itself.
Private markets are sending mixed signals
The latest deals also show that investor appetite is not moving in one direction.
On one side, some companies are seeing huge increases in their valuations.
Whatnot closed a $545 million Series G at a $20 billion valuation, roughly double the $11.5 billion valuation attached to its previous round in October.
The company operates in live shopping and collectibles, meaning its business is not directly tied to AI infrastructure or model development.
Applied Compute is also reportedly in talks to raise hundreds of millions of dollars at around a $3 billion valuation, roughly double its previous mark.
The deal has not been confirmed as closed.
At the same time, secondary-market data suggests investors are becoming more selective about software.
Evercore data reported by PitchBook showed first-half secondary deal volume above $120 billion, a record level.
But software’s share of sponsor-led secondary activity fell eight percentage points.
One explanation is growing concern around companies that could face disruption from AI, alongside weaker public-market comparables and uneven operating performance.
That creates an interesting split.
AI is attracting enormous amounts of capital, while AI is also forcing investors to rethink the value of parts of the existing software market.
The $500 billion AI infrastructure number
Perhaps the biggest number in this entire update is not a company valuation at all.
It is $500 billion.
Nvidia and six alternative asset managers have announced plans to mobilize more than $500 billion of third-party capital for AI compute infrastructure.
The partners include:
- Apollo
- BlackRock
- Blackstone
- Brookfield
- Goldman Sachs
- KKR
- Nvidia
The proposed financing platforms are expected to cover areas such as chips, power generation and data centers.
That is a significant shift in how the AI buildout is being financed.
The next stage of AI infrastructure will require far more than venture capital. It will require power, data centers, chips and enormous amounts of long-term financing.
That is where pension funds, sovereign wealth funds and large alternative asset managers come into the picture.
The AI infrastructure boom is increasingly becoming an institutional capital story.
The secondary market is becoming harder to read
There is another important takeaway here for anyone tracking private companies.
Private-market pricing is becoming more active, but that does not necessarily make it easier to understand.
There are more tenders, more secondary transactions, more continuation vehicles and more private-company financing rounds.
But these transactions do not all mean the same thing.
A $10 billion valuation established by a new investor putting fresh capital into a company is different from a $10 billion valuation used by the company to buy shares from employees.
A valuation established seven months ago is different from one established last week.
A reported valuation for a potential future round is different from a completed financing.
And an outside revenue estimate is different from audited financial statements.
The numbers can all be useful. They just need to be read in context.
What to watch next
A few developments could give investors a clearer picture of where private markets are heading.
OpenAI’s next valuation:
The $852 billion tender gives us a reference point, but a future outside funding round or public offering could provide a more meaningful market test.
Etched’s reported $20 billion raise:
The AI chip company reportedly held discussions around a $20 billion valuation after raising $300 million at $10.3 billion in July. The reported terms are not settled, so the next confirmed transaction will matter.
Formlabs and a possible IPO:
The 3D-printing company is reportedly considering an offering that could raise around $500 million. The company was valued at about $2 billion in its 2021 funding round, making any future public-market valuation particularly interesting.
National bank charter applications:
The OCC’s reasoning in the bunq decision could become more relevant if other fintech companies receive decisions on their own applications.
The bigger picture
The biggest takeaway from all these transactions is not that private-company valuations are rising or falling.
It is that the way those valuations are being created matters more than ever.
OpenAI can buy $7 billion of employee shares at an $852 billion valuation without bringing in a new investor.
Notion can grow rapidly while its last official valuation remains tied to a transaction from January.
Whatnot can nearly double its valuation in a new funding round.
AI infrastructure can attract hundreds of billions of dollars from institutional investors while traditional software companies face growing questions about AI disruption.
These developments can all happen at the same time.
For investors, the important question is not simply “What is this company worth?”
It is:
“Who decided that price, when was it decided, and what money actually changed hands?”
That is the context that turns a private-market valuation from a headline into something worth analysing.