The short version
Confirmed today: OpenAI signed four agreements with Abu Dhabi entities at ADIPEC, announced by the Abu Dhabi Media Office.
- 🛢️ ADNOC becomes the first energy company to deploy OpenAI Frontier, through G42's subsidiary Inception42.
- 💼 Mubadala, Inception42 and OpenAI will apply AI to investment analysis, opportunity identification and risk assessment.
- 🏛️ The Abu Dhabi government launched a machine-readable government initiative, already piloting with the food safety authority.
- 📚 A five-year Inception42 agreement to roll out frontier AI across government, enterprise and education in the UAE.
One technical detail worth noticing: the AI processing for these agreements runs on UAE inference residency, meaning the computation happens on GPUs inside UAE data centres rather than abroad.
Reported, but not confirmed: OpenAI is seeking around $20 billion at a $1.25 trillion valuation, co-led by UAE investors, with state-backed MGX expected to be the largest participant. The target was reportedly cut from an original $30 billion.
To be precise about the connection: these are the same jurisdiction and, in MGX's case, related institutions. But no source states that the agreements and the funding round are contingent on each other, and we are not suggesting they are.
OpenAI did not respond to requests for comment on the funding reports.
People are searching:
- "OpenAI Abu Dhabi"
- "ADNOC OpenAI"
- "MGX OpenAI investment"
- "OpenAI valuation"
- "OpenAI funding round"
- "Mubadala AI"
- "G42 Inception42"
- "can I invest in OpenAI"
Explore global stocks and ETFs from $1 on Nemo.money — zero commission, and uninvested cash earns 6% AER interest, paid daily in USD.
What "inference residency" means, and why it matters
Inference is the moment an AI model answers a question. Training is building the model; inference is running it.
Residency means where that happens physically. The Abu Dhabi Media Office said AI processing for these agreements runs on OpenAI's UAE inference residency, with real-time computation on GPUs housed inside UAE data centres.
Why organisations care:
- 🔒 Data stays in the country. For an energy company, a sovereign fund or a government, that is often a requirement rather than a preference.
- ⚡ Lower latency. Processing closer to the user is faster.
- 🏗️ It requires physical infrastructure locally, which is why these agreements connect to data centre construction rather than just software licences.
That last point links this story to everything we have written about AI spending this month. Someone has to build and power those data centres, which is why GE Vernova's power-equipment orders nearly doubled and hard drive prices rose around 50%.
The four agreements, in plain terms
ADNOC and OpenAI Frontier. The first energy company to deploy it, integrating AI across operations for what the parties describe as more connected operational, investment and commercial decision-making.
Mubadala and Inception42. Applying AI to investment analysis, opportunity identification and risk assessment, with reporting pointing to scaling Inception42's Catalyst platform across Mubadala group companies and the wider GCC.
The Abu Dhabi government. A machine-readable government initiative, already piloting with the food safety authority. Machine-readable means structuring official information so software can process it directly rather than a person reading a document.
Inception42's five-year agreement. Supporting adoption of frontier AI across government, enterprise and education across the UAE.
Sam Altman's framing, as quoted in coverage, was that Abu Dhabi and the UAE are setting a global standard for AI integration across public and private sectors. That is his characterisation of a commercial partnership.
The funding round, and what the numbers imply
Everything in this section is reported rather than confirmed. OpenAI did not respond to requests for comment.
- 💰 Around $20 billion, cut from an original target of about $30 billion.
- 🏷️ At a $1.25 trillion valuation, with UAE investors and Coatue reported as co-leading and MGX expected to be the largest single participant.
- 📈 Up from $852 billion in March, when a $122 billion round closed. MGX co-led that one too.
Here is the arithmetic that puts it in perspective. Reporting puts OpenAI's annualised revenue at roughly $50 billion as of the end of September, with the company expecting $70 billion by year-end.
At $1.25 trillion, that is about 25 times the September revenue run rate.
For comparison, from our own coverage: Nvidia trades on a price-to-earnings ratio of about 26, which is a multiple of profits. OpenAI's is a multiple of revenue, before costs. Those are different measures and the comparison only shows how differently private and public companies get valued.
The timing detail nobody should miss
OpenAI reportedly wants to close before Anthropic's expected listing in mid-November.
The reason is straightforward: both are approaching the same pool of investors, and reporting indicates four investors expected to join OpenAI's round have also backed Anthropic this year. MGX co-led Anthropic's $30 billion raise in February and OpenAI's $122 billion raise in March.
We wrote about Anthropic's prospectus when it was filed, including the detail that OpenAI filed confidentially about a week later.
OpenAI's own listing is further out. Sam Altman ruled out going public in 2026, calling it ill-advised given AI safety concerns and saying the company felt no pressure to rush. Reporting now points to 2027 at the earliest.
Which explains the order of events. Anthropic faces public buyers first. Reporting suggests some investors want to see what price Anthropic achieves before committing to OpenAI privately, and that this is part of why OpenAI reduced its target. Raising privately before a rival's listing sets a visible benchmark is a very different exercise from raising afterwards.
If the whole IPO question is new to you, our IPO explainer covers what a listing actually is, what a lock-up period means, and what the data says about how IPOs really perform. The short version of the research: IPOs have tended to underperform comparable established companies over the following three to five years, and more than half of individual IPOs lose money over that period. Worth knowing before the next one arrives.
What that competition tells you: capital for frontier AI is large but not unlimited, and the two leading labs are now explicitly racing each other to reach the same cheque-writers first. That is a meaningful change from a year ago.
The loop worth noticing
Follow the sequence, using only what has been reported.
MGX, affiliated with Mubadala and G42, has invested in OpenAI. Mubadala has now signed an agreement to use OpenAI's technology to help analyse investments and assess risk. Inception42, a G42 subsidiary, delivers the technology and runs the inference locally.
Money goes in. Technology comes back. The technology then helps decide where money goes next.
This is not unique to Abu Dhabi. We wrote in October about Nvidia investing $3 billion in Meta's share offering while Meta buys Nvidia's chips. Same shape, different participants.
Two readings, and both are reasonable:
- 🤝 The benign one: partners invest in what they use and use what they invest in. That is ordinary in technology, and it aligns incentives.
- ⚠️ The sceptical one: when the same money circulates between a small number of parties, it becomes harder to tell genuine external demand from internal recycling.
We are not adjudicating between them. The point is to notice the structure, because it is now a defining feature of how AI is being financed worldwide.
What any of this means for a reader here
Honestly? Not much that is directly investable, and it is worth saying so.
- 🚫 OpenAI is private. You cannot buy shares.
- 🚫 MGX, Mubadala, G42 and Inception42 are not publicly listed either.
- 🏛️ ADNOC has listed subsidiaries on the Abu Dhabi exchange, but those trade locally rather than in the US.
What you can research are the companies supplying the picks and shovels, which is where the money physically goes: the chip designers, the manufacturers, the power equipment makers and the storage suppliers.
That is the pattern we keep finding: the most exciting part of a theme is frequently the part you cannot own, and the investable part sits one or two steps away from the headline.
And the simplest route of all is the one most people overlook. If you hold an S&P 500 tracker, you already own a great deal of the AI build-out: the ten largest companies make up roughly 38% of the index, and most of them appear somewhere in this story as customers, suppliers or investors. The index itself now leans heavily on the AI theme, which is worth understanding whether or not you ever buy a single share directly.
What could go wrong
- 📄 The funding round is reported, not confirmed. Terms, size and valuation could all change, and OpenAI has not commented.
- ✂️ The target was already cut from about $30 billion to about $20 billion, which tells you conditions moved.
- 📉 The backdrop is a selloff. Coverage this week put the value lost by large AI-exposed stocks in October at around $3.17 trillion, and Nvidia closed about 16% below its October high.
- 🧾 Memorandums of understanding are not deployments. Several of these agreements are frameworks, and the measure of them is whether they convert into working systems.
- 💸 OpenAI is loss-making at scale, and a valuation of roughly 25 times revenue assumes a great deal about what comes next.
Research AI infrastructure on Nemo.money
- From $1. Fractional investing, in research-sized amounts.
- Zero commission on US stocks and ETFs.
- Nemes: curated themed collections to start from.
- 6% AER interest on uninvested cash, paid daily in USD.
FAQs
What did OpenAI sign in Abu Dhabi? Four agreements announced by the Abu Dhabi Media Office: ADNOC deploying OpenAI Frontier via G42's Inception42, a Mubadala and Inception42 agreement applying AI to investment analysis and risk, an Abu Dhabi government machine-readable government initiative piloting with the food safety authority, and a five-year Inception42 agreement covering government, enterprise and education.
Is OpenAI raising money from the UAE? Reporting indicates OpenAI is seeking around $20 billion at a $1.25 trillion valuation, co-led by UAE investors with MGX expected to be the largest participant, reduced from an original $30 billion target. This is reported rather than confirmed, and OpenAI did not respond to requests for comment.
What is inference residency? Inference is when an AI model produces an answer. Residency refers to where that computation physically happens. These agreements run on GPUs located inside UAE data centres, so data and processing stay in the country.
Can I invest in OpenAI? No. OpenAI is a private company and its shares are not available to retail investors. MGX, Mubadala, G42 and Inception42 are also not publicly listed.
How can I get exposure to AI from the UAE? Through listed companies that supply the industry, such as chip designers, manufacturers, power equipment makers and storage suppliers. On Nemo.money you can research and invest in eligible US-listed stocks and ETFs from $1 with zero commission, with uninvested cash earning 6% AER interest, paid daily in USD. Your capital is at risk.
Nemo = Never Miss Out.
Stay informed. Stay ahead.
#OpenAI #AbuDhabi #UAE #AI #Investing #NemoMoney
Terms and conditions apply. This is not investment advice. Past performance is not indicative of future results. Your capital is at risk. See website for Risk Disclosure. Exinity ME Ltd (https://nemo.money) is regulated by ADGM's Financial Services Regulatory Authority.
