Hard Fork

ChatGPT’s Platform Play + a Trillion-Dollar GPU Empire + the Queen of Slop

October 10, 2025

Key Takeaways Copied to clipboard!

  • OpenAI is aggressively pursuing a platform strategy by integrating third-party apps directly into ChatGPT, aiming to make it the new front door to the internet, echoing a high-risk strategy previously attempted by Facebook. 
  • OpenAI is securing massive infrastructure commitments, including a major deal with AMD, as part of a stated plan to build a trillion-dollar GPU empire necessary to power its future products, signaling an unprecedented level of compute constraint and investment in the AI ecosystem. 
  • The initial public reception to OpenAI's Sora 2 is highly polarized, showing massive initial adoption as a novelty toy while simultaneously facing backlash from copyright holders and revealing a clear gender skew in early usage due to inherent risks associated with likeness generation. 
  • AI-generated video, specifically Sora, is currently limited by technical hurdles like clip length and lack of fine-grained control over likeness and voice, but is expected to gain greater mainstream adoption as these issues are resolved. 
  • The trajectory of ChatGPT, which initially saw limited use but achieved massive adoption as it improved, serves as evidence that AI-generated video is here to stay and should not be underestimated. 
  • Guest Katie Notopoulos, the 'Queen of Slop,' humorously received a mock cease and desist for her trolling activities using AI tools and promised to continue creating provocative content with future platforms. 

Segments

DevDay Vibe and Disclosures
Copied to clipboard!
(00:03:01)
  • Key Takeaway: OpenAI DevDay 2023 at Fort Mason evoked parallels to Facebook’s 2014 developer platform announcement, suggesting a similar ambition to become the internet’s central hub.
  • Summary: The event hosted around 1,500 people and featured demos like a Sora cinema and a phone booth station for interacting with ChatGPT. Hosts provided mandatory disclosures: The New York Times is suing OpenAI/Microsoft, and one host’s boyfriend works at Anthropic.
OpenAI Growth Metrics
Copied to clipboard!
(00:05:27)
  • Key Takeaway: ChatGPT has achieved massive scale, boasting over 800 million weekly users and processing 6 billion tokens per minute via its API, a significant increase from 300 million in 2023.
  • Summary: Sam Altman announced that Sora 2, GPT-5 Pro, and a new smaller voice model will be coming to the API for developers. Mattel is already using Sora 2 to prototype toy designs. OpenAI also introduced Agent Kit for building AI agents via a drag-and-drop interface.
ChatGPT Platform Strategy
Copied to clipboard!
(00:06:59)
  • Key Takeaway: OpenAI is launching apps within ChatGPT, allowing users to interact with services like Expedia and Zillow directly within the chat interface, positioning ChatGPT as the primary gateway to the web.
  • Summary: This strategy aims to make ChatGPT the new homepage, replacing Google or Facebook as the starting point for online activities. Functionality involves opening sub-windows within ChatGPT to perform tasks like browsing Zillow listings based on personalized criteria. The underlying promise is safely sharing personalized user data with vetted external services.
Platform Strategy Risks and History
Copied to clipboard!
(00:10:27)
  • Key Takeaway: OpenAI’s platform play mirrors Facebook’s early 2010s strategy, which ultimately led to the Cambridge Analytica scandal due to overly permissive data sharing rules, posing a greater privacy risk now due to the intimate nature of ChatGPT conversations.
  • Summary: Facebook’s past platform success involved sharing user data with developers, monetized via virtual currency cuts, but this ended when data misuse became public. ChatGPT data, potentially including therapy transcripts, is far more sensitive if breached or misused by third-party developers like Zillow or Target.
Data Sharing and Trust
Copied to clipboard!
(00:14:22)
  • Key Takeaway: OpenAI claims it will only share the minimum necessary information for transactions, but connection prompts explicitly warn users that data, including memories, may be shared with developers, making opting out a rational choice for privacy-conscious users.
  • Summary: Executives stated a commitment to user trust, but the connection process requires users to confirm they understand data sharing, including conversational history. The hosts noted the inherent conflict between serving the user and monetizing the platform through these new integrations.
Ive/Altman Fireside Chat Critique
Copied to clipboard!
(00:19:50)
  • Key Takeaway: The highly anticipated fireside chat between Sam Altman and former Apple designer Jony Ive provided zero substantive details about their rumored new hardware product, consisting only of vague, uninformative rhetoric.
  • Summary: The 30-minute session was described as ‘GPT2 level’ content, featuring unfinished sentences and philosophical musings on design. The hosts issued a ‘ship it or zip it’ ultimatum, demanding concrete specifications rather than abstract ruminations on design essence.
Trillion-Dollar GPU Empire
Copied to clipboard!
(00:24:29)
  • Key Takeaway: OpenAI is building infrastructure requiring the energy equivalent of 20 nuclear reactors by locking in multi-year deals with chipmakers like NVIDIA and AMD, committing to roughly a trillion dollars in GPU spending.
  • Summary: The AMD deal involves OpenAI purchasing six gigawatts worth of GPUs (half of the NVIDIA commitment) in exchange for cheap AMD stock, signaling a strategy to diversify suppliers and encourage increased production. This massive infrastructure buildout is propping up the U.S. economy, with data centers accounting for 92% of GDP growth in the first half of the year.
AMD Partnership Details
Copied to clipboard!
(00:30:56)
  • Key Takeaway: The OpenAI-AMD partnership extends beyond chip sales to software collaboration, aiming to improve AMD’s ROCm software to compete with NVIDIA’s dominant CUDA platform, thereby benefiting the broader AI ecosystem.
  • Summary: OpenAI gains infrastructure access and a financial stake in AMD, while AMD secures massive future demand and crucial software development assistance. This deal structure suggests OpenAI is strategically spreading its bets across multiple infrastructure providers.
Financing Ambitions and Systemic Risk
Copied to clipboard!
(00:34:24)
  • Key Takeaway: OpenAI plans to finance its trillion-dollar infrastructure needs through fundraising, vendor financing, and product revenue, but the reliance on ’novel financial instruments’ and the deep interconnectedness of the AI ecosystem create systemic risk if the bubble bursts.
  • Summary: Analysts suggest Sam Altman now holds the power to either crash the global economy or lead it to a promised land, highlighting the high stakes of the current AI investment paradigm. The industry’s reliance on continuous, massive capital inflow means failure could cause widespread collapse.
Sora Backlash and Creator Concerns
Copied to clipboard!
(00:46:44)
  • Key Takeaway: The Sora 2 app achieved the number one spot on the U.S. app store despite significant backlash from Hollywood copyright holders and concerns from major YouTube creators about potential revenue impact.
  • Summary: OpenAI has adjusted generation filters in response to copyright complaints, but creators like MrBeast and Casey Neistat recognize the dual nature of the tool: a creative asset versus an economic threat. The novelty factor of Sora is currently high, but its long-term mainstream viability depends on technical improvements like longer clips and better voice cloning.
Sora Likeness and Bullying
Copied to clipboard!
(00:59:30)
  • Key Takeaway: The Sora app’s current user base skews heavily male, suggesting women are proactively opting out of likeness generation due to obvious downside risks, contrasting with OpenAI’s prediction that users will eventually demand their likenesses be used.
  • Summary: Katie Notopoulos observed that the platform is currently dominated by teenage boys engaging in in-group trolling, exemplified by videos making fun of other tech reporters. OpenAI offers settings to restrict cameo usage, but the inherent risk of misuse for bullying remains a significant concern.
Sora Mainstream Hurdles
Copied to clipboard!
(01:05:51)
  • Key Takeaway: Sora’s current limitations include inability to change outfits, inaccurate voice replication, and a 10-second clip restriction, preventing full mainstream adoption.
  • Summary: The technology behind AI video generation, while amazing, faces technical hurdles preventing immediate mainstream success. Specific user desires, such as changing an avatar’s outfit or having accurate voice cloning, remain unfulfilled. These product moves are necessary before broader embrace occurs.
ChatGPT Adoption Parallel
Copied to clipboard!
(01:06:48)
  • Key Takeaway: ChatGPT’s initial lukewarm reception followed by massive adoption after iterative improvements suggests AI video will follow a similar path to success.
  • Summary: The initial user experience with ChatGPT was often limited, leading many to abandon it after initial trials. However, continuous improvement led to 800 million users adopting the tool over time. This historical pattern suggests that AI-generated video is a technology that will persist and grow.
Guest Farewell and Trolling
Copied to clipboard!
(01:07:09)
  • Key Takeaway: Host issues a mock cease and desist to guest Katie Notopoulos, who vows to continue creating ‘deranged videos’ and escalate trolling with future platforms.
  • Summary: The host jokingly demands Katie Notopoulos cease her trolling activities involving AI video generation. Notopoulos confirms she will not stop, promising to use even worse methods on the next emerging platform. The segment concludes with thanks and a final warning to the guest.