Sharp Tech with Ben Thompson – (Preview) A Spring Break Mailbag: RIP Sora, Ads and Surplus, F1 Going in Reverse, Elon Inc., Smartphone Parenting, and More

Date: March 27, 2026
Hosts: Andrew Sharp & Ben Thompson

Episode Overview

In this spring break mailbag episode, Andrew Sharp and Ben Thompson take a lighter approach, addressing listener questions while diving deep into the abrupt shutdown of the Sora app, the economics and challenges of consumer and enterprise AI, and musings on business models and market dynamics. The tone is conversational, occasionally playful, with the hosts trading jabs about their very different approaches to “vacation.”

Key Discussion Points & Insights

Spring Break, Podcast Setup, and Episode Theme

• Andrew jokingly calls out Ben for mixing work and vacation, comparing their travel habits (00:29–01:40).
• The episode’s format is an informal “mailbag,” focused on listener questions and lighter tech industry commentary.
• Promise of "at least 50% nonsense questions" for this episode (01:40).

The Demise of Sora: Domain Names, Social Apps, and Business Models

Sora's Shutdown Announcement & Initial Reactions

• The show's kickoff is the news that the Sora app (an AI video creation/social app) is shutting down.
• Ben calls out the app’s weak brand presence ("@Sora official app" rather than just “Sora”) and jokes about the classic Bob Lefsetz takedown of failed tech/media products (02:08–03:01).

  "If you are at Sora official app as opposed to Sora, maybe you are not long for this world." – Ben (02:50)

Did Sora Actually Matter?

• OpenAI’s shutdown statement said, “what you made with Sora mattered.” Andrew and Ben question this, with Ben wondering if anything created with Sora was truly valuable (03:32–03:36).

  "Did anything made a Sora matter?" – Ben (03:32)

Vibes and the Social Media Graveyard

• Both reflect on how Sora and similar products seldom succeed, and reference Meta’s “Vibes” app still surviving, somewhat miraculously (03:36–04:12).
• Ben notes that Sora looked impressive briefly, but its momentum quickly stalled—it was more novelty than lasting platform.

Sora’s Place in the Product Landscape

• Discussion of Sora as a standalone social network, its business model, and whether integrating its features into ChatGPT would have changed its fate (04:24–05:30).

  “The ones that have worked are the ones they just ended up buying because they sort of came out of the ether.” – Ben (04:56)

• Andrew and Ben muse that Sora’s “death” is largely inconsequential, except possibly for Disney, which had tried to engage with the platform (05:31–05:41).

The Hollywood Panic, Copyright, and Product Lessons

• Andrew highlights how Sora’s AI-powered video generation sparked serious copyright concerns in Hollywood and led to broader industry anxiety (05:45–06:11).

  "This app scared the crap out of Hollywood... Hollywood was all freaked out about copyright and what this could mean." – Andrew (05:45)

• Ben speculates that the clamp-down on copyright and lack of “shared cultural content” undermined Sora’s appeal. People were most interested in using famous IPs and memes (06:11–07:17).

  "Things that people have in common are going to matter and matter more and more...like live events, like live events, more and more important..." – Ben (06:11)

Sora Versus Instagram: The ‘Come for the Tool, Stay for the Network’ Thesis

• Ben compares Sora’s failed strategy to Instagram’s winning approach—starting as a tool, then evolving into a network. He references Chris Dixon’s canonical “come for the tool, stay for the network” argument (07:17–08:29).
• Sora may have erred by focusing too soon on social features, rather than growing from a viral single feature.

The Social Network Success Rate and The Cost of Running Sora

• Ben: Most new social networks fail—being a skeptic nets you a high batting average (08:31–08:49).

  "If you want to have a good track record, just come out and say every single one is going to fail and you're going to have a great nine out of 10 times." – Ben (08:41)

• OpenAI’s renewed focus and Sora’s surprising compute bill prompted its swift shutdown regardless of usage (08:49–09:53).

  "The amount of compute this appears to have been using, even now when no one's using it, feels astronomical." – Ben (08:49)

Notable Compute Statistic:

• Andrew cites an estimate: Sora may have cost $15 million per day to run (09:18).

  "I saw $15 million a day thrown out. I don't know what the actual number is, but yeah, I mean, it was... obviously compute-intensive." – Andrew (09:18)

Parting Thoughts on Sora and AI Era Misses

• Andrew: The AI era is still in early innings. Sora will be remembered as a brief, wild experiment—made possible by viral meme-making potential, ultimately doomed by copyright and business/cost realities (09:53–10:37).
• Ben: The zero marginal cost thesis is reversed here—since Sora’s compute costs never approached zero, growth did not produce a sustainable flywheel (11:13–12:45).

  "For video generation in particular, we're nowhere close to a world [with] zero marginal cost... not having zero marginal costs is sort of debilitating." – Ben (11:52)

OpenAI’s Business Shift: Consumer vs. Enterprise (14:05–19:52)

Listener Question: Did OpenAI Make a Mistake Pivoting to Enterprise?

• Listener Adrian questions whether OpenAI's strategic pivot away from consumer products limits its upside, given historical precedent like Microsoft's path (13:18–14:05).

Ben’s Take: Profitable Paths in AI, Monetization, and Uncertainty

• Ben emphasizes the unique uncertainty of AI ("My answer is I don't know... the degree of uncertainty is so high") and the resource reality: compute costs drive strategic decisions, and enterprise remains the most reliable revenue source (14:05–15:55).

  "You only get a big new consumer company like every 15 to 20 years. There's usually like one defining one per paradigm... consumer is the great prize." – Ben (15:20)

• He laments OpenAI possibly missing a “flywheel” opportunity to build both ad-based and consumer-centric revenue, but concedes enterprise deals (like Codex) pay bills when compute is expensive (15:55–16:59).

  "It's so frustrating... We don't know if that {consumer flywheel} would have worked because they were so late to get started. And now... the best way to make money in the short to medium term... is in enterprise." – Ben (15:55)

Andrew’s Point: Microsoft’s Monopoly vs. Modern Market Complexity

• The “everyone learned Microsoft at work” advantage doesn’t quite apply today. Enterprise AI tools will be more fragmented, and Anthropic may be gaining share quietly (16:59–17:38).

  "OpenAI's story is going to be more complicated... because I'm sure there are going to be a lot of different tools that are used at enterprises..." – Andrew (16:59)

Sizing the AI Players: Revenue Metrics and Market Share

• Ben discusses how private AI company revenue claims can be misleading: sometimes numbers are quoted as “gross merchandise volume” (GMV), rather than actual profit; Anthropic’s vertical revenue trajectory appears impressive, but apples-to-apples comparisons are difficult (17:38–19:31).

  "Anthropic is basically... talking about their skyrocketing annual recurring revenue numbers... that's actually like the total that people are paying..." – Ben (17:48)

What’s Next? Public Markets

• Both hosts expect OpenAI and Anthropic may go public soon due to capital needs. The opacity of their current business metrics will diminish (19:43–20:00).

Notable Quotes & Memorable Moments

• On Social Network Prognostication:

  “If you want to have a good track record, just come out and say every single one is going to fail...” — Ben (08:41)

• On Sora’s Economics:

  “The amount of compute this appears to have been using, even now when no one's using it, feels astronomical.” — Ben (08:49) “I saw $15 million a day thrown out...” — Andrew (09:18)

• On the AI Era:

  “We're all living through the early innings of the AI era here and there are going to be misses and weird experiments and all sorts of things that are charming in retrospect. Sora is already one of those things.” — Andrew (09:53)

• On Zero Marginal Costs:

  “For video generation in particular, we're nowhere close to a world [with] zero marginal cost... not having zero marginal costs is sort of debilitating.” — Ben (11:52)

• On AI Business Model Uncertainty:

  “My answer is I don't know, like the degree of uncertainty is so high and OpenAI is a great example.” — Ben (14:05)

Timestamps for Key Segments

• Spring Break and Mailbag Setup: 00:00–01:40
• Sora App Shutdown Discussion: 02:08–04:12
• Sora, Copyright, and Social Network Analysis: 04:24–07:17
• Instagram Comparison and Social App Failure Rates: 07:17–08:49
• Sora’s Compute Costs & Zero Marginal Cost Challenge: 08:49–12:45
• OpenAI’s Pivot: Consumer vs Enterprise Thesis: 14:05–16:59
• Enterprise Adoption and AI Market Fragmentation: 16:59–17:38
• Revenue, Metrics & Anthropic’s Growth: 17:38–19:43
• Prospects of AI Companies Going Public: 19:43–20:00

Tone, Style, and Host Dynamics

• The episode is relaxed and somewhat tongue-in-cheek, with both Andrew and Ben bringing a sense of perspective to flashy tech trends and industry hype cycles.
• The hosts mix sharp analysis with casual banter, poking fun at each other and the industry as a whole.

Summary for Listeners:
This episode is a thoughtful but chill deep dive into why so many hyped apps—like Sora—don't last, what separates real consumer platforms from novelties, and why AI company business models are so challenging today. The hosts walk listeners through the realities of compute cost, business focus, and the unpredictable path toward building the tech giants of the AI era—reminding us to remain skeptical, but not cynical, amid the hype.