🥛 OpenAI just broke the AI deal 🧨

GM. This is Milk Road AI, built for the people who want to understand what's happening in tech before their group chat does.

Here’s what we’ve got for you today:

✍️ The Microsoft-OpenAI divorce nobody saw coming.
🎙️ The Milk Road AI Show: OpenAI Missed Revenue Targets… Is the AI Bull Market in Trouble?
🍪 DeepSeek says it’s now near frontier AI.

Consensus Miami is one of the largest digital asset conferences that’s going all in on crypto and agentic commerce. Grab your passes at 20% off.

Prices as of 10:00 a.m. ET.

THE DEAL THAT REWROTE THE AI POWER MAP

In 1803, Napoleon Bonaparte had a problem.

He was broke, overstretched, and staring down a war with Britain he couldn't afford, and he needed cash, fast.

So he did something that, on the surface, appeared to be total surrender.

He sold Louisiana to the Americans for $15M, which is roughly 828,000 square miles for about 3 cents an acre.

Source: HistoryMemes/Reddit

Everyone called it a humiliation, a desperate fire sale from a man with no options.

But here's the part they skip in the history books:

Napoleon walked away with hard cash, a funded war, and zero obligation to defend a territory he couldn't hold anyway.

He turned an illiquid asset he was about to lose into a weapon he could actually use.

Microsoft did something that looked almost identical, they tore up their exclusive deal with OpenAI.

Gave back the exclusivity, dropped the famous AGI clause, and let OpenAI run its models on AWS and Google Cloud.

And the internet immediately called it a loss, but it was the opposite.

What Microsoft actually gave up (and what they kept)

To understand why this looks like a loss but isn't, you need to know what the original deal actually said.

When Microsoft first invested $1B in OpenAI in 2019, and another $10B in 2023, they didn't just write a check and wish Sam Altman good luck.

They bought a cage.

Every OpenAI model trained on Azure, every API call routed through it, and 20% of revenue kicked back to Microsoft.

Source: Uncoveralpha

And if OpenAI ever declared it had achieved AGI, Microsoft's exclusivity died immediately.

It was a brilliant structure, right up until it wasn't.

The moment Amazon announced a $50B investment in OpenAI in February 2026, with OpenAI's new enterprise platform Frontier running natively on AWS, the cage became a courtroom.

Microsoft reportedly started drafting legal threats and one source told the Financial Times, “We understand our contract. We will take legal action if they violate it.”

And this is where the Napoleon move happened.

Instead of litigating a partnership into rubble, Microsoft traded the cage for something better, a guaranteed revenue structure that works whether OpenAI runs on Azure, AWS, or a server farm on the moon.

Here's what Microsoft kept:

$250B in Azure cloud commitments from OpenAI. Already representing 45% of Microsoft's entire $625B revenue backlog.
20% revenue share on all OpenAI sales through 2030. Every ChatGPT subscription, API call, 20 cents on every dollar, capped at an undisclosed ceiling.
27% equity stake in OpenAI Group PBC, now valued at over $135B on an original investment of $13.8B.
Non-exclusive IP license through 2032. They can still build Azure AI products on top of OpenAI's models.
Zero revenue share obligation going the other direction. Microsoft stopped paying its own 20% cut to OpenAI for Azure OpenAI Service revenue. Net positive on margins.

The exclusivity clause was an asset turning into a liability in real time and that shift created a deeper problem.

OpenAI had a simple, terrifying problem: they were trying to justify an $852B valuation while a single partner held legal leverage over their entire product distribution.

Try selling that S-1 to public investors, and the math was also ugly.

OpenAI needs more than 30GW of compute over the next decade, and Azure can't supply that alone.

At the same time, Amazon Web Services’ Trainium chips offer 30–40% better price-performance for inference workloads, which at OpenAI’s scale compounds into billions in savings.

Source: Carmen Li

So what started as a strategic advantage became a constraint and Azure exclusivity wasn’t a moat anymore but rather a ceiling.

THE ENTIRE INDUSTRY IS HEADING TO MIAMI

Where do you go to hear people talk about crypto, AI, and real capital?

If you want surface-level takes, then it’s probably in my group chat.

But if you want institutions with deep pockets, you’ll have to head to Consensus Miami.

Consensus Miami is one of the largest digital asset conferences that’s going all in on crypto and agentic commerce.

Here are the key details:

20,000+ global attendees
$4T AUM managed by finance giants attending Consensus Miami
The ultimate intersection of crypto and AI

The best part?

You can get an exclusive 20% discount on passes with code MILKROAD.

Grab your tickets at 20% off.

THE DEAL THAT REWROTE THE AI POWER MAP (P2)

There is a concept in finance called a circular deal.

Company A invests in Company B. Company B commits to spending that money back on Company A's products.

Company A books the commitment as revenue and that revenue validates more investment.

The AI industry has turned this into an art form.

Microsoft invests $13.8B in OpenAI. OpenAI commits $250B to Azure, and Microsoft earns 20% of all OpenAI revenue.

Microsoft now holds a 27% equity stake that appreciates as OpenAI grows.

Amazon is running the same play. They invested $33B in Anthropic, and they committed $100B+ to AWS.

Amazon invested $50B in OpenAI, committed to a $38B, seven-year cloud contract with AWS.

The largest infrastructure commitment ever signed by a private AI company.

Oracle has a $300B compute contract with OpenAI. NVIDIA invested up to $100B in OpenAI and sells them the chips to run everything.

Source: @lydiadepillis

This is essentially a closed-loop economy where the same dollars keep circling between the same few players, getting slightly larger each lap.

This whole structure requires OpenAI to actually monetize at the scale implied by an $852B valuation.

And this morning, the Wall Street Journal reported they aren't.

OpenAI missed multiple monthly revenue targets in early 2026.

They missed their internal goal of reaching 1B weekly active ChatGPT users by year-end.

Subscriber defections are rising as Anthropic wins coding and enterprise, while Google’s Gemini takes the rest.

It gets worse inside the building.

CFO Sarah Friar has reportedly told company leadership that if revenue doesn't accelerate, OpenAI might not be able to pay for its own future data center contracts.

The market reacted accordingly, with all the AI infrastructure names sliding: Oracle dropped 5.5%, CoreWeave fell 5.4%, and NVIDIA, Advanced Micro Devices, and Broadcom all slipped.

When OpenAI sneezes, the entire AI infrastructure complex catches a cold, which tells you everything about how tightly the circular loop has been wound.

The circular deal only works in one direction. When OpenAI grows, everyone wins, but if OpenAI misses, the whole loop tightens at once.

The quiet winner nobody's talking about

If you squint at the entire scoreboard, one company comes out structurally ahead of everyone else.

It’s Amazon.

And today, while everyone was still processing the renegotiation headlines, AWS made it official.

They just announced that OpenAI models are now live on Amazon Bedrock.

Codex, OpenAI's coding agent used by over 4M people weekly, is available on Bedrock.

And they launched Amazon Bedrock Managed Agents, powered by OpenAI, a one-stop shop for enterprises to deploy production-ready AI agents on AWS infrastructure.

Translation: the ink on the renegotiation isn't even dry, and Amazon is already collecting.

The timing is not a coincidence, and this is what the $50B investment bought.

A full product suite, live today, that lets any enterprise run OpenAI's most capable models through the AWS accounts they already have, with IAM controls, PrivateLink, CloudTrail logging, and existing compliance frameworks already baked in.

For years, 'you need Azure to access OpenAI' was a real enterprise sales proposition that drove Azure workloads, but that proposition is now dead, and it’s been replaced by running OpenAI on whatever cloud you already use.

They already had Anthropic locked in with $33B in investment and $100B+ in cloud commitments, now they have OpenAI too.

Both of the two most important AI companies on earth, on Bedrock, with no exclusivity drama and no legal risk.

Amazon shares jumped ~5% the day the OpenAI partnership was first announced. $140B in market cap in a single session.

That's what it looks like when you're the silent winner in a room full of people fighting over credit.

What to watch next

The IPO is the whole game now.

OpenAI's CFO, Sarah Friar, has reportedly flagged the Q4 2026 timeline as aggressive internally.

The company raised $122B in a round where most of the capital is tied to purchase commitments rather than free cash, which makes unit economics deliberately difficult to assess.

When that S-1 drops, it will be the first objective test of whether the circular capital structure creates durable value or just inflates reported revenues.

Watch Azure's next two earnings calls with exclusivity gone, whether Azure's growth rate holds will reveal how much of the AI moat was OpenAI-specific versus genuine enterprise migration.

And watch Anthropic as they're targeting a late-2026 public listing.

As the only frontier AI model currently available across all three major clouds, they have a structurally cleaner multi-cloud story than OpenAI.

If they file first and absorb retail investor appetite for AI exposure, the entire dynamic shifts.

The real question isn't which cloud wins but rather whether anyone can actually monetize this thing at the scale these valuations require.

Napoleon sold Louisiana and funded his war, but he still lost at Waterloo.

Alright, that's it for this edition of Milk Road AI. We want to hear from you.

Who's the real winner of the Microsoft-OpenAI renegotiation?

OPENAI MISSES TARGETS, AI BULL MARKET SHAKEN? 📉

In today's episode, we sat down to break down why OpenAI's missed revenue and user targets sparked a short-lived selloff and what it reveals about the real bottleneck holding AI back.

Here's what you'll hear:

Why compute scarcity, not demand, is now the ceiling on AI revenue growth thanks to permitting, grid queues, and energy limits.
How Microsoft quietly won the OpenAI restructuring, giving up exclusivity but locking in a 20% revenue share through 2030.
Why Elon's rumored $60B Cursor move signals just how desperate the race for AI talent and compute has become.
How ETH, SOL, and BTC are being reframed as core infrastructure for AI agents, settlement, and a post-scarcity economy.

Hit play and dive in 👇️

YouTube | Spotify | Apple Podcasts

Midnight is a fourth generation blockchain that just launched. Check out their launch announcement here.

Real Finance Blockchain is an EVM-compatible L1 that is built specifically for RWA tokenization. Read more about Real Finance Blockchain here.

Nexo is back in the U.S. - and new clients get 30 days of Wealth Club Premier perks! Higher yields, lower borrowing rates, and crypto cashback - start here.