There is a golden rule that applies to many things, especially in business and finance. That rule is: “If you want to know the truth, follow the money”. Let’s follow OpenAI’s money today, and let’s see where it leads.
Here is the official record of OpenAI’s fundraising history so far:
- 2015–2019 $130 million (collected) from Elon Musk and other donors
- 2019 $1 billion from Microsoft
- 2021 $2.1 billion from Microsoft ($2B), Venture funds ($100M)
- 2023 $10 billion from Microsoft
- 2024 $6.6 billion from Microsoft, NVIDIA, Thrive Capital, Tiger Global, Altimeter, SoftBank
- 2025 $10 billion syndicated by SoftBank (a second $30 billion tranche is still pending to complete the $40 billion commitment)
In total, OpenAI has fundraised $29.83 billion in capital. However, we know for a fact that not all the capital contribution was cash. For example, in 2023 Microsoft did not invest $10 billion in cash but only $1 billion, with the rest being “Azure Credits” as exposed by the article: “OpenAI has received just a fraction of Microsoft’s $10 billion investment“. Various sources confirmed this was the structure of all previous Microsoft investments in OpenAI as well. This brings down the total cash OpenAI truly raised through its fundraising to $28.13 billion.
On top of that amount, in October 2024, a pool of banks granted a $4 billion revolving credit facility to OpenAI (“OpenAI secures $4 billion credit line after big funding round“), with the company issuing a press release stating that in October 2024 it had “over $10 billion of liquidity”. Hold on a second, let’s take a step back. If in October 2024 OpenAI raised $6.6 billion in capital and got granted a $4 billion revolving credit facility for a total of $10.6 billion, considering the company’s press release announced it had over $10 billion in liquidity after the closing of the 2 deals, wouldn’t that logically imply that by September 2024 OpenAI’s liquidity position was down to just a few hundred million dollars? The answer is: yes.
Let’s take a look now at the latest OpenAI cash projections as reported by The Information
Complex math is not necessary to realize that by March 2025, when OpenAI announced the new fundraising round led by SoftBank (“Announcement Regarding Follow-on Investments in OpenAI“), the company had already burned through a good chunk of the $6.6 billion it fundraised back in October 2024. Hence, the need to raise more capital quickly; otherwise, it’s hard to imagine banks would have kept their revolving credit facility commitment in place without triggering a request for immediate full repayment. Clearly, once again, OpenAI’s cash position was in a very poor shape just a few months ago.
At this point, it is important to clarify that a big portion of OpenAI’s October 2024 fundraising was done in the form of Convertible Notes that will only become equity if OpenAI manages to turn itself into a “for-profit” corporation (“Exclusive: OpenAI’s huge valuation hinges on upending corporate structure“). The same applies to the $40 billion committed by SoftBank, which is why only $10 billion from that agreement has been transferred to OpenAI. This does not even consider the fact that SoftBank does not have all the $40 billion it committed to invest in OpenAI. As I reported in my article “SOFTBANK GOES ALL IN ON AI WITH THE LITTLE MONEY IT HAS LEFT“, the company led by Masaponzi Son was forced to syndicate to other investors ~$2 billion of the first $10 billion investment tranche and to raise debt to finance the amount it had to pay directly.
Putting all this together, according to its own latest cash projections (which I bet are still optimistic), OpenAI won’t have enough cash on hand to last until the end of 2026. Afterwards, assuming OpenAI manages to fundraise the $30 billion left from the agreement with SoftBank, the company projects it will need $60 billion more to make it till the end of 2029, when it assumes its operations will finally start to generate a ton of cash. I would not hold my breath on this last assumption.
Doesn’t all I just said imply that every single commitment OpenAI signs with any partner from now till the end of 2029 is subject to the company raising a ginormous amount of capital funded in cash? The answer, again, is: yes. However, even if there are still gullible investors willing to take OpenAI’s projections at face value, the company is facing a very practical problem: who can put down all that cash regardless of any valuation number? Very hard to answer this question.
What we can say for sure is that any commitment signed by OpenAI must be taken with a grain of salt. This is what the company agreed to directly pay to suppliers to purchase computing power so far this year:
- $15.9 billion total to CoreWeave in the next 5 years
- $30 billion total to Oracle in the next 5 years
Furthermore, we can confidently assume that the extra $19.4 billion computing capacity Microsoft decided to commit to with Nebius (“Stock Market Today: Nebius Soars on $19.4B Microsoft AI Deal“) is on behalf of the additional demand it received from OpenAI, similar to the additional capacity it is currently buying from CoreWeave, again very likely on behalf of OpenAI demand, worth a total of ~$15 billion in multiple years.
The total value of these agreements, not surprisingly, is ~$80.3 billion, and that’s EXACTLY the capital OpenAI needs to fundraise till the end of 2029. As a consequence, it is now evident how OpenAI’s cash burn projections are totally unrealistic since increasing operational costs are totally ignored. If you also consider that OpenAI just signed a $10 billion agreement with Broadcom to manufacture its own chips, we can already say that the total cash burn the company will face till 2029 rises to $90.3 billion, assuming its revenues grow fast enough to cover its operational costs after the 20% cut OpenAI has to pay on its revenues to Microsoft (and that the company is desperately trying to renegotiate).
Can we now conclude that all this “boom” in computing demand, mostly driven by OpenAI, has very small chances to materialize? Yes, we can. Both because the company is running out of pockets it can extract cash from and because those left still willing to invest in OpenAI at an asinine valuation will only do so if they can purchase real shares in the company. What is the likelihood of OpenAI successfully turning itself into a “for-profit” company? They are fewer every passing day, which is why the WSJ just reported how its executives are “scrambling” to find a solution (“Exclusive | OpenAI Executives Rattled by Campaigns to Derail For-Profit Restructuring“). Personally speaking, I believe the end of the story is already written: sooner or later, OpenAI will hit a brick wall and implode spectacularly, ultimately bursting the giant nonsensical bubble formed in the AI and semiconductor space similar to what happened with the big telcos companies in the early 2000s (“HOW BIG TECH SPENDING SPREE MORE AND MORE RESEMBLES 2000s TELECOMS EPIC FAIL“).
ARTICLE UPDATE [11th September]
After I published this article, the WSJ reported that the vast majority of Oracle’s RPOs massive jump, which triggered a single-day ~36% increase in the value of the company’s shares, was related to a $300 billion agreement signed with OpenAI (“Exclusive | Oracle, OpenAI Sign Massive $300 Billion Cloud Computing Deal“). The WSJ article doesn’t clarify whether this sum will be on top of the previous $30 billion agreement the two companies signed back in July (“OpenAI agreed to pay Oracle $30B a year for data center services“), so I can only update my previous calculations on OpenAI’s cash burn to a range between $360 and $390 billion for the next 7 years (because the new agreement between Oracle and OpenAI will not start untill 2027).
Let’s be serious. The chances OpenAI could find $90 billion of cash to fulfil its agreements in place were already very slim, but surely now there is not a single chance it could find up to $390 billion to pay all it committed to its suppliers. What’s going on here then? My opinion is that OpenAI is more and more desperate to raise cash, and its chances of success depend on how many people with deep pockets are still willing to buy into the more and more nonsensical AI narrative Sam Altman has been preaching nonstop. Consequently, OpenAI is running around signing these types of bloated agreements that make great headlines and are very effective in continuing to feed investors FOMO. Furthermore, I would not be surprised at all to read in the future the announcement of Oracle’s investment in OpenAI, why? Think about it, Oracle’s market capitalisation jumped ~$300 billion in a single day because of the ~$300 billion agreement signed with OpenAI. What’s going to happen to Oracle’s share price if OpenAI runs out of cash and files for bankruptcy? The magnitude of the tumble is not hard to imagine. Moreover, the same will happen to all those companies that tied their fortunes to OpenAI one, like Microsoft or Nvidia.
While OpenAI might not be “too big to fail” for the real economy, it surely is “too important to fail” for all the Big Tech companies that bet hard on the “AI promises of salvation” narrative of the evangelist Sam Altman. This is the reason why I believe Big Tech companies and heavily invested firms like SoftBank will figure out a way to keep OpenAI in business as long as possible, but unless they understand how to start generating positive cash flow from this first generation of AI fast, they will ultimately be forced to pull the plug on OpenAI simply because they will not be able to afford doing otherwise without putting in danger their established and healthy businesses.
JustDario on X | JustDario on Instagram | JustDario on YouTube
SYNNAX SUMMER PROMO
Get exclusive intelligence on stocks you can’t find anywhere else – insights that could give you a serious edge. Claim your free Synnax trial month now: https://synnax.app/PROMO/JUSTDARIO
If you love what we built at Synnax (and I think you will), this advanced market intelligence will cost you less than a cup of coffee a month for the next 12 months. After the 30th of September, the annual subscription price will be 149$ – redeem your free access now and give it a try.