Big Technology Podcast – Friday Edition (December 12, 2025)

Episode: OpenAI’s 2026 Priority, Disney’s AI Play, Datacenter Buildout Trouble
Host: Alex Kantrowitz
Guest: Ranjan Roy (Margins)

Episode Overview

This episode dives into four big stories dominating the 2025 tech landscape:

• OpenAI’s new focus on enterprise AI for 2026, following an exclusive meeting with Sam Altman and media execs.
• Disney’s billion-dollar embrace of AI video through an OpenAI partnership (and all the implications for IP, fandom, and online creation).
• Major instability in the AI datacenter/infrastructure trade after disappointing moves from Oracle and Broadcom.
• Internal tensions at Meta: Is the Superintelligence team at odds with the rest of the company?

As always, Alex and Ranjan provide candid, witty analysis, memorable moments, and a touch of irreverence.

1. OpenAI Shifts Focus to Enterprise AI in 2026

The Scoop: Altman’s NYC Lunch (03:02–07:40)

• Sam Altman hosted a private lunch with major editors/CEOs (The Atlantic, The New Yorker, Fortune, NYT).
  • Main revelation: Enterprise AI is a “very big priority” for OpenAI in 2026, shifting from their 70% consumer/30% enterprise focus.
  • Greg Brockman, OpenAI cofounder, confirmed on Twitter: “Enterprise AI is going to be a huge theme of 2026.” (06:22)
  • Next-gen model (GPT-5.2) is tailored for business use cases: workforce planning, project management, deeper integrations.

Discussion: Is This a Smart Move? (04:55–13:48)

• Ranjan: “It makes for a splashy headline. But all it did for me is just affirm the lack of focus... Are they going to develop an AI cloud? Is it going to be a consumer devices company?... To me, the most interesting part is that this is what Sam Altman said in that room.” (04:55)
• Alex: Strong disagreement: “This is a big technology scoop... It makes sense from a logical standpoint. They’re going to need to show substantial revenue growth next year. And one way to do that is by attacking this enterprise area.” (06:00)
• Key insight: The unprecedented growth of enterprise AI as a software category ($37.5B projected revenue in 2026, up from “effectively zero” in 2022).

Product vs. Model Debate

• Alex: “What matters is what you do with that model and how you distribute it.” (12:01)
• Ranjan: “I might have to shut off my laptop... My work here is done.” (12:39)
• OpenAI, and the AI field at large, recognizes models are “commoditizing” and improvement is now application-specific, not general leaps (“it’s not a training problem, it’s an application problem.” – Altman, paraphrased by Alex, 10:55).
• General Agreement: OpenAI can be hugely valuable even if it doesn’t reach AGI—if it succeeds in making its models indispensable in different business and professional domains. (13:48–15:58)

The Challenge of Enterprise

• Enterprise AI is “a completely different beast.” OpenAI must build out an expensive, specialized sales and support infrastructure—a process that’s “much easier said than done.” (09:15)
• Risk: Focusing on enterprise may alienate OpenAI’s vast consumer user base.
  • “This is why I believe they're really gonna have to separate those two models out....eventually they're going to have to split out into separate models.” (22:40)
  • Humorous moment: “People are going to have relationships and they're going to have cyber sex with their ChatGPT and then they're going to ask it to organize their files. I think you might need two separate windows for that stuff.” – Alex (22:40)

2. Disney & OpenAI: The AI Content Deal of the Year

Deal Details & Implications (26:03–34:58)

• Disney invested $1B in OpenAI, taking an equity stake.
• OpenAI gets licensed access to 200+ Disney characters for AI-generated video creation on Sora, for three years.
• Fans can generate videos "surfing with Stitch" or "wielding a lightsaber with R2-D2".

Analysis: Smart for Disney? Smart for OpenAI?

• Ranjan: “From a creative standpoint, this is very interesting...Fandom and actually licensing IP for generative video, the moment I started playing with any of this, I was convinced had to become a thing.” (27:46)
• Alex: "Are content creators going to fight this or lean into it?... I'm surprised at how quickly Disney said yes. They like control... The second Mickey Mouse is doing cocaine or voicing your ChatGPT erotic pal, things are going to be bad. Disney's going to be mad." (28:47)
• Disney's calculated risk: Belief that OpenAI can enforce a “brand safe environment," and willingness to meet the trend head-on rather than playing constant whack-a-mole with copyright.
• Reference to Disney's cease & desist to Google over copyright and Google’s “pathetic and disappointing” response (30:33).

On Fandom and Monetization

• Disney wants to deepen fan investment; making it easy to put yourself in a scene with a Disney character translates into merchandising and fandom growth. (32:18–32:48)
• CEO Bob Iger: Curated Sora user videos may appear on Disney+. Disney aims to incorporate user-generated Sora content directly into its streaming ecosystem.

Notable Moment/Quote

• Ranjan: “Anything to actually boost that connection fans have with these characters should probably make them a lot of money. So from that standpoint I guess I can kind of see it.” (32:18)
• Alex: “Putting yourself in a scene with R2-D2 or whoever it is... It’s a cool thing for people. I think this is really going to work out well for them.” (33:17)

The Money Trail – Who Pays Whom?

• Why did Disney invest and commit to buying OpenAI enterprise licenses if it’s licensing its characters? (34:03–34:29)
• Speculation: Possible circular funding (OpenAI pays Disney with equity for IP; Disney invests the same billion back via equity/warrants).

3. AI Infrastructure Trade: Reality Check for Oracle & Broadcom

What Happened? (38:17–41:43)

• Oracle announced delays on data center projects for OpenAI (pushed from 2027 to 2028).
• AI data center buildout is proving less profitable (lower margins) than hoped.
• Broadcom’s hardware order flow was similarly underwhelming.
• Both stocks took a major hit.

Takeaways

• Ranjan: “We see real numbers each one of these times... Starting to understand the logistics of building out these data centers... Not going to be that straightforward.” (39:33)
• “This is just bringing a degree of reality to the entire infrastructure trade... Reminding people it’s not that simple.”

Financial Risk: Credit Default Swaps (42:16–43:54)

• Oracle’s CDS (credit default swap) costs hit their highest intraday level since April 2009—a direct market signal of rising risk for default.
• Ranjan: “Anytime [CDS] is entering the mainstream conversation, it’s not a good thing for anyone.”

Is the AI Bubble Bursting? (44:57–47:06)

• Not systemic yet, but…
• Alex: “Is this the beginning of the unraveling of the AI trade?”
• Ranjan: More of a “shift in what the AI trade is.” The focus is moving to application layer efficiency and value, rather than just compute/infrastructure spending. (45:04)
• “Building applications that run in a more compute efficient way start to become a competitive advantage, especially in enterprise.”

4. Meta's Superintelligence Team vs. Product Execs

NYT Story Recap (47:06–52:03)

• Alexander Wang, brought in to run Meta’s Superintelligence Lab, reportedly clashing with product officers Chris Cox and Andrew Bosworth.
• Disagreement over using Meta’s social data to improve Meta’s models and ad products—or aiming for general superintelligence.

Panel Reactions

• Alex: "Ideally, if you're running a super intelligence operation within Meta, you're going to use Meta's Instagram and Facebook data to train, right? Very weird story to me." (48:59)
• Ranjan: “Meta properties probably have one of the richest data sets in the world... To have some kind of difference in opinion and purity or preciousness... was surprising to me.”
• The real concern is intra-company politics (splashy public hires, early departures); Meta must avoid a breakdown between the pure research and business/product units.
• Alex: “The way we’ll know this is really over is when Meta Super Intelligence Lab says it’s focused on enterprise. That’s the end.” (52:03)

Notable Quotes & Moments

• On OpenAI's Priorities:
  “When you sell to enterprises, you need to build out an enterprise sales infrastructure. This is something that took Google like what, a decade?... OpenAI is going to have to figure this out.” – Alex (09:15)

• On Consumer vs. Enterprise AI Experience:
  “Trying to develop the product simultaneously in those two directions, which they kind of have to. It's tough. It's tough.” – Ranjan (21:38)

• On Disney AI Licensing:
  “Disney, of all companies, is the one that did it.... They know that people are going to do it anyways because OpenAI has a hands-off policy. So they might as well try to bring some structure to it.” – Ranjan (29:32)

• On AI Infrastructure Economics:
  “This is just bringing a degree of reality to the entire infrastructure trade… reminding people that it’s not going to be that straightforward.” – Ranjan (39:33)

• On Meta's Internal Tensions:
  “If the politics of this are actually getting in the way of the technology, that’s not a good start for this.” – Ranjan (49:37)

• On the AI Hype Cycle:
  “Here's how the AI bubble bursts. You can compute much more efficiently. We still don't have real clear indications that this stuff is going to be economically valuable on the scale that people imagine.” – Alex (45:51)

Segment Timestamps

• 00:00–02:24: Intros, show setup, OpenAI/Disney/infra/Meta preview
• 02:24–07:40: Inside Altman's lunch, OpenAI’s shift to enterprise, reactions
• 07:41–13:47: Product vs. Model debate, business model implications
• 13:48–23:30: Enterprise challenges, risk of consumer alienation, AI erotica
• 26:03–34:58: Disney–OpenAI deal breakdown, creative and business stakes, monetization
• 34:59–38:01: Sora usage, AI-generated video culture & syndication
• 38:01–47:06: Oracle/Broadcom woes, infrastructure trade reality, market risk
• 47:06–52:22: Meta’s AI team divisions, data dilemma, role confusion
• 52:22–52:33: End of episode

Tone and Style

• Irreverent, humorous, lightly skeptical.
• Both hosts balance deep industry knowledge with playful banter (e.g., recurring "Mickey Mouse doing drugs" jokes as an emblem of AI content moderation chaos).
• Willingness to ask uncomfortable business questions and call out hype.
• Regular callbacks to earlier episodes/conversations, rewarding long-time listeners.

For Listeners Who Missed the Episode

• OpenAI will chase enterprise dollars in 2026, following market pressure and model commoditization. The era of breakthrough model leaps is likely over—now it’s about killer apps built on top of very similar models.
• Disney’s billion-dollar leap into AI video will probably drive a raft of similar IP/content deals, with huge potential for both fan engagement and "brand safety" headaches.
• The tech infrastructure stack is facing a tough reckoning—overhyped profits are meeting slow, expensive real-world buildouts.
• Internal politics at AI super-labs (Meta, and by implication, everyone else) remain a huge wild card—success may depend as much on organizational design as on technical advances.

Verdict: A must-listen episode for anyone trying to decode the next moves of the AI giants, the shifting focus of AI revenue, and how tech titans are gambling on the new digital goldrush.