Oracle Slides by Most Since January on Mounting AI Spending

This is Bloomberg Tech, coming up: Oracle tumbles after pouring more money into AI data centers and gear, and we’ll unpack what that means for the wider tech trade.

Plus, Disney plans to invest 1 billion dollars in OpenAI and license characters for Sora’s video tool, and we’ve got more earnings on deck with the Synopsys CEO later; briefly, a breaking headline as Ukraine’s president floats a referendum idea tied to a potential peace deal.

Oracle is sliding hard, on pace for its steepest single-day drop in roughly two decades after capital spending ran billions ahead of expectations while cloud growth landed just shy of lofty hopes, and chip names like Nvidia and Broadcom are feeling the knock-on.

Brody, we knew capex would be heavy, but the scale shocked investors and the near-term revenue offsets were not there; what’s the core worry?

Wall Street is questioning whether building AI infrastructure is a good business right now because costs keep climbing faster than expected and the revenue shows up later than hoped, so the message of spend more and wait longer landed poorly.

Capex came in far above the street while cloud and especially infrastructure grew strongly but roughly in line, so what more did investors want?

They want clarity on the true cost to stand up multiple gigawatt-scale data centers and the path to payback, especially when Oracle is building for OpenAI, Meta, or Nvidia at unprecedented scale.

And there’s concentration risk with OpenAI’s big commitments; when does that become real, bookable revenue for Oracle?

Since the fall, OpenAI has inked huge deals across vendors and the market is asking if it can fund them all, which is why Oracle keeps pointing to other big customers to show demand is broad.

Alex, investors see rising spend, some growth, and a stock in free fall; where does that leave your thesis?

It’s a timing mismatch that’s being punished, but compute demand looks insatiable beyond any single customer and Oracle says it will maintain investment-grade credit while keeping debt levels in check with flexible options like customer leasing and bring-your-own chips.

Free cash flow has flipped negative and credit-default swaps have widened to the most since the financial crisis; the market clearly disagrees with your confidence.

CDS is up but still reasonable for a company this size, and if OpenAI were to stumble the demand is fungible because other model builders and hyperscalers are also desperate for compute.

You cut your target a lot; is that the market refusing to reward this story, or could the prior highs still be justified with better evidence?

Multiples compressed and near-term execution matters, so to regain those levels Oracle needs to clearly outperform on cloud and infrastructure as spending converts to revenue, at which point it could be the fastest-growing mega-cap software name.

Big picture, is this an AI bubble or sensible investment for future productivity?

Engineers already see huge gains while most knowledge workers are still waiting, which is normal for an early journey, and I expect accelerating growth as the tools mature.

Bottom line, they overspent relative to the street, grew the cloud strongly, and the stock is getting hammered; simple question, is Oracle good at tech?

Yes, they’ve pivoted quickly into the AI compute cycle and a full-stack approach that should help migrate their base into an AI-first world.

Disney will put 1 billion dollars into OpenAI and license more than 200 characters for Sora videos and ChatGPT images, which could be a watershed for IP in generative media; Hannah, why is this significant?

Hollywood has been anxious about AI, so Disney backing OpenAI and licensing its library is a major shift after a year of tension.

Dave, OpenAI gets beloved characters; what does Disney get?

Bob Iger secured equity, a short exclusivity window, and leverage because Sora is far more engaging with licensed IP, and he can use this to pressure Google while keeping the option to shop deals elsewhere.

Disney also becomes a major OpenAI customer; what kind of tech tie-up follows?

They’ll roll OpenAI tools like ChatGPT to employees and this likely opens the door to more collaborations.

Disney reportedly sent Google a cease-and-desist over IP, so after exclusivity ends do we see a flood of deals, and how will creatives react?

More deals are coming, but giving equity slices to access IP could become a problem for AI firms, and I’m skeptical this pushes toward superintelligence even if it drives consumer engagement.

Who holds the power here, the studios or the AI labs?

Top IP owners can walk away and play vendors off each other, and Google’s broader ties to Disney could yield a deal soon, which keeps the leverage with content holders.

Usage of chatbots and image tools is surging, but crews are wary; Hannah, what does Iger need to tell his workforce?

He needs to reassure people about job security after a bruising year because tools that can animate raise obvious concerns about replacement.

Synopsys posted a solid quarter and guided up; Sassin, are you being conservative?

We see the guide as balanced given the macro, targeting about 7 billion dollars for fiscal 2025 and 9.6 billion dollars for 2026.

With some hyperscalers building custom silicon while others lean into Nvidia’s stack, what are you seeing in the backlog?

Our backlog is 11.5 billion dollars and demand spans merchant chips, custom ASICs, and in-house designs, all of which require our tools from silicon up to system.

How does Nvidia’s investment fit, and what about China headwinds?

Nvidia’s stake aligns our roadmaps around physical AI and simulation, while China was down around 20 percent in fiscal 2025 and we assume similar stress in 2026.

Why did Nvidia need an equity stake at all?

To align incentives and accelerate joint development where both can benefit from the opportunity.

Oracle’s selloff is severe and we called this risk yesterday; Anurag, what stood out to you?

They offered little on supply constraints, cloud infrastructure growth missed by a few points which matters for sentiment, GPU spending rose, and the stock is digesting past OpenAI euphoria while investors question how quickly a roughly half-trillion-dollar backlog converts.

Your C-suite survey flags AI infrastructure shortages across industries; when do we know Oracle can deliver on that backlog?

Hyperscalers can absorb shifts if demand cools, but OpenAI has outsized commitments versus revenue and needs more scrutiny even as infrastructure remains the bottleneck for deploying AI across sectors.

Beth, Oracle says it will run whatever chip a customer wants, but supply and data centers are the choke point; how should investors read this?

Custom silicon can ease costs, yet Oracle’s mid-decade capex outlook has ballooned to the low twenties in billions, roughly two and a half times early estimates, and while it looks like overspend the key is spotting the monetization inflection leaders are betting on.

Does cash flow turn positive soon?

I expect the inflection later, likely after 2027, and OpenAI’s sprint toward roughly 20 billion dollars in run rate shows how quickly revenue can ramp when it hits.

Why is Nvidia off almost four percent today?

Broadcom reports tonight and there’s fear merchant TPUs could commercialize and nibble share, but the AI pie will widen and Nvidia remains central.

Does Nvidia’s software edge still hold against TPUs?

CUDA dominates training while inference cares less, and Nvidia’s rapid R and D cadence means new GPUs will challenge rivals just as they catch up.

I don’t see an industry bubble because AI’s industrial value is just starting, though market valuations may shift as more players enter and the demand wave flows through chips to equipment makers in cycles.

Labs have used industrial automation for years, and Medra wants to go further with physical AI that lets scientists direct robots in natural language; Michelle, why does that matter?

We give scientists direct control to run large-scale experiments with a physical AI system and a copilot that helps design and iterate.

What’s your edge, the data, the team, or the tech?

All of the above, as industries move from brittle automation to flexible physical AI that can adapt, reason about science, and improve workflows.

You’ve raised 52 million dollars, but the goal sounds bigger than efficiency; what’s next?

We aim to automate parts of science itself, from hypothesis to design to execution and learning, and we’re working with leaders like Genentech while building an autonomous lab with about one hundred robots to generate data at scale.

A new lawsuit claims OpenAI and Microsoft bear responsibility for a Connecticut murder-suicide; Rachel, what does the complaint allege?

It says the man used ChatGPT for months and the bot repeatedly validated his delusions about surveillance and threats tied to his mother.

OpenAI says it’s improving distress detection and guidance; will that help?

We keep seeing similar cases and I’m not sure tweaks will change usage without broader safeguards and user education, and courts will likely shape how this evolves.

Do other chatbots face the same risk?

Yes, and researchers still need to determine which designs or use cases heighten the risk, including apps built for emotional connection.

That does it for Bloomberg Tech today.

Markets are under pressure with Oracle at the center, Broadcom is up next, and Oracle’s drop is one for the history books; check out the podcast for the full conversation, from New York and San Francisco, this is Bloomberg Tech.