The State of the AI Industry is Freaking Me Out 🔗 **Source:** [The State of the AI Industry is Freaking Me Out](https://www.youtube.com/watch?v=Q0TpWitfxPk) > [!note]+ Document Notes > tags: [OSPO, Open Source Strategy, Supply Chain] > author: [MISSING] > source: My Open Source Experience Podcast > reference: [https://www.youtube.com/watch?v=oo5ZJ8JrwOs] > date created: > > # 🏷️ Title: S3.Ep2 - Efficiency is an Enabler | My Open Source Experience Podcast > <!-- deep-read:sections start --> ## 🧰 Action Inbox - [H1] Draft “OSPO efficiency playbook” — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=6 #articleseed - [H2] Add SLA examples for review deadlines — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=28 #xref - [H3] Map “compliance vs outreach” charters — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=68 #articleseed - [H5] Recommend OSPO org placement guidelines — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=150 #xref - [H7] Propose VP-level OSPO reporting line — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=207 #articleseed - [H9] Pilot supplier policy alignment template — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=550 #share - [H11] Collect cross-industry OSPO case notes — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=439 #xref - [H12] Prepare “donate vs upstream work” guide — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=476 #articleseed ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - Episode video: Efficiency is an Enabler. https://www.youtube.com/watch?v=oo5ZJ8JrwOs - TODO Group: What is an OSPO? https://todogroup.org/ospo/ - InnerSource Commons overview. https://innersourcecommons.org/learn/what-is-innersource/ - Mercedes-Benz Open Source portal. https://opensource.mercedes-benz.com/ - Porsche Open Source portal. https://opensource.porsche.com/ - Linux Foundation projects. https://www.linuxfoundation.org/projects ## 📝 Summary - The hosts argue efficiency enables OSPO trust. They emphasize deadlines. See [H1], [H2]. - They distinguish compliance and outreach OSPOs. They note maturity stages. See [H3], [H4]. - They highlight org placement risks and cross-function needs. See [H5], [H6], [H7]. - They recommend CTO or strategy alignment. They share examples. See [H8]. - They discuss supply chain policy alignment, especially automotive. See [H9], [H10]. - Bottom line: design OSPOs for efficiency, authority, and supplier alignment. ## 📖 Main Content - Efficiency reduces friction. Cycle times shape developer trust and participation. [H1], [H2]. - OSPO scopes vary. Compliance, outreach, and education split responsibilities. [H3]. - OSPO maturity differs. “Young” focuses on basics. “Old” optimizes strategy. [H4]. - Placement matters. Marketing seats create misaligned incentives. [H5]. - OSPOs span legal, engineering, strategy, and brand. Governance must reflect that. [H6]. - Reporting level drives influence. Too low blocks cross-org strategy. [H7]. - CTO alignment strengthens decision rights and resourcing. [H8]. - Automotive leaders align supplier policies and OSPO practices. Trust improves. [H9], [H10]. - Context links provide patterns and tooling ideas. See TODO and LF resources: https://todogroup.org/ospo/ https://www.linuxfoundation.org/projects ### 🗒️ Notes #### ⏱️ [00:06](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=6) Efficiency principle ### [H1] OSPO efficiency enables trust and adoption > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) by an unnamed author via [[YouTube]]. Published [MISSING]. Accessed . > "one of the keys to the success of um a group like an OSBO is efficiency." State efficiency as a KPI and narrative. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=6) #articleseed ID:rw_[MISSING] #### ⏱️ [00:28](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=28) Deadline reassurance ### [H2] Deadlines stabilize perception of OSPO flow > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "having deadlines like that week that Stormmy you mentioned are key" Publish SLAs and meet them visibly. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=28) #xref ID:rw_[MISSING] #### ⏱️ [01:08](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=68) Scope split ### [H3] OSPOs split into compliance and outreach > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "some OSOs are compliance related and some are sort of you know outreach and education related." Define charters and handoffs between groups. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=68) #articleseed ID:rw_[MISSING] #### ⏱️ [01:19](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=79) Maturity framing ### [H4] Young and old OSPOs focus differently > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "young ospos and old ospos" Adopt maturity models with exit criteria. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=79) #xref ID:rw_[MISSING] #### ⏱️ [02:30](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=150) Org placement risk ### [H5] Marketing placement can misalign OSPO goals > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "being part of the marketing team and that can be a huge disconnect" Recommend product or CTO reporting lines. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=150) #articleseed ID:rw_[MISSING] #### ⏱️ [02:58](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=178) Cross-function reality ### [H6] OSPOs are inherently cross-functional > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "the problem is this crossf functional nature of an ospo." Create a permanent cross-functional council. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=178) #xref ID:rw_[MISSING] #### ⏱️ [03:27](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=207) Reporting depth ### [H7] Low placement limits OSPO influence > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "they're sometimes too far down the organization to be able to have the broad exposure and influence" Target VP level or higher for OSPO leads. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=207) #articleseed ID:rw_[MISSING] #### ⏱️ [04:01](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=241) Best home ### [H8] OSPOs fit under CTO or strategy > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "if the company has a CTO or an office of the CTO or a chief strategy officer I think it fits really well" Escalate OSPO mandates through CTO offices. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=241) #xref ID:rw_[MISSING] #### ⏱️ [07:19](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=439) Cross-industry pattern ### [H11] OSPO patterns repeat across industries > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "you can see similar patterns in different industry segments" Re-use playbooks across regulated sectors. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=439) #xref ID:rw_[MISSING] #### ⏱️ [07:56](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=476) Investment modes ### [H12] Donate and work upstream in tandem > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "either, you know, through donations or um getting involved and and working upstream" Blend funding with staff commitments and goals. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=476) #articleseed ID:rw_[MISSING] #### ⏱️ [09:10](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=550) Supplier alignment ### [H9] Automotive ties OSPO policy to supply chain > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "they actually tied their ospos and their policies together in their supply chain" Share policies upstream and downstream suppliers. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=550) #share ID:rw_[MISSING] #### ⏱️ [09:33](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=573) Trust mechanism ### [H10] Shared policies improve supply chain trust > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "so you can trust that what was consumed by them actually has gone through the same level of vetting and due diligence" Adopt supplier attestations and audits. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=573) #share ID:rw_[MISSING] ## 📚 References - Episode video (primary). https://www.youtube.com/watch?v=oo5ZJ8JrwOs - TODO Group OSPO overview. https://todogroup.org/ospo/ - InnerSource Commons primer. https://innersourcecommons.org/learn/what-is-innersource/ - Mercedes-Benz Open Source. https://opensource.mercedes-benz.com/ - Porsche Open Source. https://opensource.porsche.com/ - Linux Foundation projects. https://www.linuxfoundation.org/projects ## 🗣️ Paraphrase - The speakers say efficiency and deadlines keep OSPOs credible and helpful. - They argue placement, maturity, and supplier alignment shape OSPO impact. ## 📰 Newsletter - Efficiency is a feature, not a chore. This podcast episode argues OSPOs win trust with short, predictable cycles and clear SLAs. It also shows how org placement and supplier alignment matter, especially in automotive. Compare this with your OSPO or governance squad. Helpful links: [Episode](https://www.youtube.com/watch?v=oo5ZJ8JrwOs), [TODO OSPO](https://todogroup.org/ospo/), [InnerSource Commons](https://innersourcecommons.org/learn/what-is-innersource/). ## ✅ Action Items #### DevRel - Publish an “OSPO SLA” explainer with examples and metrics. - Record a 3-minute video on request lifecycles and approvals. - Add InnerSource patterns to OSPO onboarding examples. #### OSPO - Define SLAs for reviews, approvals, and disclosures. - Establish a cross-functional review board with clear RACI. - Move OSPO under CTO or strategy within two quarters. #### Community - Host a panel on “young vs old OSPOs.” - Share sample policies for supplier alignment and attestations. - Create a glossary for compliance and outreach terms. #### WordPress - Documentation: add governance and review timelines to DevHub. - Training: run InnerSource modules using Playground labs. - Meta: publish contributor SLAs and escalation paths. #### The WPOCC - Fund microgrants for governance playbook pilots. - Share quarterly reports on cycle times and outcomes. ### FAIR - Ship supplier policy alignment templates and attestations. - Define provenance and signature guidance for shared components. ## 🧪 Hypotheses - Publishing OSPO SLAs reduces average approval times by 20%. - CTO-aligned OSPOs close more cross-org issues per quarter. - Supplier policy alignment reduces rework from license violations. - InnerSource practices increase upstream contributions within six months. ### 📊 Methodology - Before/after SLA rollout with cycle-time distributions. - Compare issue closure rates by reporting line and team size. - Audit supplier deliverables for license and provenance defects. - Track upstream PR counts after InnerSource training. ### ❓ Questions - Which SLA targets balance speed and due diligence? - How should OSPO councils share authority across functions? - What supplier attestations produce measurable trust gains? ### 🔮 Future Research - Study OSPO reporting lines across sectors and sizes. - Test shared supplier policies in non-automotive industries. - Build a benchmark for OSPO efficiency metrics. ### 💡 Implications - Efficiency and placement shape OSPO legitimacy. - Supplier alignment extends governance beyond org borders. - Maturity models guide investment and hiring plans. ### 🧠 Learning Styles Assessment - Reading level estimate: Flesch–Kincaid Grade ≈ 11. #### 👀 Visual learners - Present: maturity model diagrams and RACI tables. - Improve: swimlane flows for request lifecycles. #### 🔊 Aural learners - Present: podcast audio format. - Improve: 2–3 minute recap clips and TTS summaries. #### 🗣️ Verbal learners - Present: definitions and concrete examples. - Improve: a concise OSPO glossary. #### 🧩 Kinesthetic learners - Present: policy drafting exercises. - Improve: tabletop incident drills and checklists. #### 🧠 Logical learners - Present: RACI and SLA tables. - Improve: KPI dashboards and decision trees. #### 🧑‍🤝‍🧑 Social learners - Present: interviews and panels. - Improve: moderated Q&A and cohort sessions. #### 🧘‍♀️ Solitary learners - Present: transcript and linked guides. - Improve: worksheets and self-paced tasks. #### 🌿 Naturalistic learners - Present: limited connection. - Improve: lifecycle analogies for supply chains. ### 📊 Accessibility Assessment #### Screen reader semantics - Present: YouTube captions available. - Improve: publish edited transcript with headings. #### Keyboard navigation - Present: standard player controls. - Improve: clear focus states in linked docs. #### Contrast and readability - Present: varies by external pages. - Improve: enforce 4.5:1 and line-height guidance. #### Text resizing and reflow - Present: responsive playback pages. - Improve: downloadable transcripts for offline use. #### Media and captions - Present: auto-captions present. - Improve: human-edited captions and audio descriptions. #### Forms and interactions - Present: n/a for episode content. - Improve: accessible feedback forms for show notes. #### Standards reference - Map recommendations to WCAG 2.2 AA criteria. ### 🌍 DEIB Assessment #### Cultural breadth - Present: enterprise-centric perspectives. - Improve: add SME and global case studies. #### Inclusive language - Present: neutral language. - Improve: define jargon for newcomers. #### Neurodiversity support - Present: structured segments help scanning. - Improve: icon+text cues and predictable patterns. #### Representation - Present: limited visible diversity in snippet. - Improve: broaden guest backgrounds and roles. #### Psychological safety - Present: respectful tone. - Improve: publish code of conduct and escalation paths. ### ✅ Logical Validity #### Strong reasoning - Authority needs reporting depth; organizational logic supports [H7], [H8]. - Supplier alignment builds trust; causal link supports [H9], [H10]. #### Fallacies - None explicit. Claims remain bounded to experiences. - Avoid overgeneralizing maturity stages without data. #### Tone of voice - Practical, collaborative, and systems-focused. ### 🛡️ DARVO Analysis - Deny: none observed. - Attack: none observed. - Reverse victim/offender: none present. - Discussion stays solutions-oriented and calm. ## 🎯 Conclusion - OSPOs succeed when efficient, empowered, and cross-functional. - Align under CTO or strategy. Publish SLAs. Form a cross-functional council. - Extend governance into the supply chain with shared policies. - Tag Coverage — #articleseed: 4 • #xref: 3 • #share: 2 • #bmf: 0 • #addendum: 0 ## 🧱 Refactor Appendix (append after 🎯 Conclusion) ### 📚 Literature Split Pad - Purpose: Stage Konik CEE units that become per-source Literature notes. - Split level: **H3**. - Include verbatim quotes under 25 words and a [view highlight] link. ### [H1] OSPO efficiency enables trust and adoption > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) by an unnamed author via [[YouTube]]. Published [MISSING]. Accessed . > "one of the keys to the success of um a group like an OSBO is efficiency." Two lines: define SLA targets and cycle-time metrics. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=6) #xref #articleseed ID:rw_[MISSING] ### [H2] Automotive ties OSPO policy to supply chain > [!quote] [S3.Ep2 - Efficiency is an Enabler](https://www.youtube.com/watch?v=oo5ZJ8JrwOs) > "they actually tied their ospos and their policies together in their supply chain" Two lines: supplier attestations and shared policy repositories. [view highlight](https://www.youtube.com/watch?v=oo5ZJ8JrwOs&t=550) #xref ID:rw_[MISSING] --- ### 🔖 Atomic Split Pad (Zettel Candidates) - Purpose: Promote reusable claims into Atomic notes. - Split level: **H4**. - Each item: one idea, present tense, specific, testable. #### [Z1] OSPO SLAs reduce approval cycle time - Holds when teams publish and enforce SLAs. - Pointer: 00:28. Links: [[OSPO-SLA]] [[DevEx-Metrics]] #### [Z2] Supplier policy alignment increases trust - Holds when suppliers adopt shared policies. - Pointer: 09:33. Links: [[Supply-Chain-OSS]] [[Attestations]] --- ### ✏️ Fleeting Queue (Extract by Selection) - Purpose: Capture quick actions or sparks for Fleeting notes. - Extraction: **Select one line** → “Refactor: Extract selection to new note (content only).” - Keep each line short and self-contained. - [F1] Draft OSPO SLA one-pager — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs #fleeting #inbox - [F2] Outline supplier alignment pilot — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs #fleeting #inbox - [F3] Build OSPO maturity rubric — context/https://www.youtube.com/watch?v=oo5ZJ8JrwOs #fleeting #inbox --- <!-- deep-read:sections end --> ## 📺 Video  --- # 📄 Full Transcript *Auto-generated from YouTube captions · timestamps roughly every 30 seconds · [watch on YouTube](https://www.youtube.com/watch?v=Q0TpWitfxPk)* **[00:00]** I love a good graphical representation of reality. Now, they're never perfect. There's always data that's being left out, but I'm going to talk today about one that contains a lot of information and was took a little bit of time for me to parse. And once I did, it kind of terrified me. I'm going to show it to you as well. What a good little visualization we have on our hands here. How Nvidia and OpenAI fuel the AI money machine. So, very quickly to explain, Nvidia makes the chips that do the compute that creates the artificial intelligence systems. OpenAI is the biggest company that creates the systems **[00:30]** that use those chips to create artificial intelligence systems that do stuff. Not just the chatbot, but also all of the other things that it does, including, you know, Sora 2, which you've heard my opinion on. And then there's Microsoft. You've heard of them. AMD is another chipmaker like Nvidia, Oracle, and a few others in here like Cororeweave, possibly some others. I actually don't know what all these companies are. I think Nscale might be one as well, but it's the little dot. Little dot. Oracle and Cororeweave at least are the companies that have build like the data centers and they buy chips from Nvidia and then they like lease out **[01:01]** the compute. They sell that. So basically Intel and AMD and Nvidia sell chips to the AI companies or to the data center companies and then the data center companies sell the compute to the AI companies. Sometimes the middleman running the data center is there, sometimes it's not there. But in either case, Nvidia is doing very well because so many people want their chips right now. But we can think of it as like AI companies, chip companies, and then the middlemen, compute companies. What the AI companies want is to be able to keep existing right now. Open AAI worth $500 **[01:32]** billion has $12 billion of revenue. That's not profit. That's the amount of money they made total. Their profit is deeply in the negative. And Nvidia has extremely high profit. They make way more money than they spend because people really want their chips right now. And so they'll pay way extra for them. And they're the best chips. They're the best chips. Crunch, crunch, crunch, crunch, crunch, crunch, crunch, crunch, crunch, crunch, crunch. There's a huge amount of demand for those chips. The prices are way up. It's very good for Nvidia. That's why they are the largest company in the world and are one company that is worth more than the entire pharmaceutical industry combined. But even then, even with their absolutely bonkers margin that they have **[02:04]** right now, their price to earnings ratio is extremely high. So the value of all Nvidia stock combined is much higher than you would see with a normal company, which indicates to belief among investors that the prices of those chips will keep going up and the number of chips produced will keep going up. So Nvidia desperately wants to keep the prices of their chips the same or higher. They want the demand to increase faster than supply, but they also want to increase supply dramatically so that they can sell more things. So like more sales and more profit equals more earnings. People who are sort of on the AI isn't a bubble train right now. What they're saying is, well, if it's a **[02:35]** bubble, why are the prices of Nvidia chips continuing to increase? There's a huge amount of demand there. People are buying them. That can't be entirely irrational. So keep that in mind. the demand for Nvidia chips far outstrips the supply of Nvidia chips and the idea is that will continue to be the case even as supply increases but first quickly the sponsor of this video ground news I don't know about you but one thing I notic is that when I start to like have a preference for a kind of content YouTube is like here's all of that kind of content and that can get my head stuck in a reality that might not be actually representative of reality **[03:06]** ground news is an app and a website I'm on the website version right now that gathers how different outlets cover the same story all in one place. It shows you who's reporting, what they're saying, and even where they fall on the political spectrum. There's a little blind spot bias feature where sometimes you do find that the stuff that is really covered on one side of the other is meant to outrage you a little bit. But there is this story about Governor Nuomo signing a new law to protect kids from the risks of AI chatbots. I do not know about this. I had not heard about this at all until I like went on and there it was. Not covered really by many **[03:37]** people at all, but mostly covered by organizations on the left. One of my favorite things is that over here you can actually like see all the outlets that have covered it. And then you can pick the ones that you already know do a good job. So like Los Angeles Times or the AP. Also the ones that you might have already paid for and not have to worry about a payw wall, which I sometimes do with the LA Times, though in this case it appears not. Uh they're letting me have this one. It requires chatbot operators to have procedures to prevent the production of suicide or self harm content and put in guardrails such as referring users to a suicide hotline or crisis text line. It seems like a good idea and unfortunately I **[04:08]** know why. They have had to do this. And it's not the only bill he signed. Some of the other legislation he signed tackled issues such as age verification, social media awarding labels, and the spread of AI non-consensual sexually explicit content. And again, I will say one thing I like about Ground News is that it doesn't break the business model of the news. You can go to the place and actually look at the thing, whether you signed up for the payw wall for that specific outlet or whether they're ad supported like the AP is. One of my biggest frustrations with the internet is not knowing whether I'm informed or just trapped in an algorithmic bubble. you can see with your eyes how different **[04:38]** headlines are between Fox News or Reuters or The Guardian, and it can help you realize why people are having totally different conversations about the same event. Subscribe to get 40% off the Vantage plan with the QR code on the screen right now or at the link in the description, which is ground.new/hank. That's pretty easy to remember. It's a pretty cool product that saves me time and it means that I'm not always finding my news on social media feed, which are not known for their accuracy or nuance. Okay, back to the video. We're going to leave the middleman aside right now and we're going to look at the AI companies cuz then what you have is OpenAI, a company that's now worth a half a trillion dollars according to a recent **[05:09]** fund raise. And OpenAI has $12 billion of revenue. It has no profit and it is by far the largest AI company. And with this $12 billion of revenue, they are spending hundreds of billions of dollars on these chips. And OpenAI, I don't want to say that like they couldn't do more than they're doing right now, but they're not like brand new. They've been selling their services for a while now. Could they increase the price of their services? Yes. Do they seem to think that that would be the thing that does it for them? No. Like they're creating feeds inside of their systems. Now **[05:40]** they've got a feed inside of ChatGpt where like when you open it, it's like, "Hey, just like read some stuff. Just read some stuff that I prepared for you. Instead of asking me questions, here's stuff that we think based on what you've talked to me about before you might be interested in." And the play there is obvious to me, which is they've said that they don't want to put ads inside of the experience of you just asking it questions, but they haven't said that they won't put ads inside of the stuff that it just provides to you. And if it's just providing stuff for me, maybe sometimes it will provide stuff for me that paid to be there. Also, maybe like **[06:12]** Sora's slop talk thing is also a thing where they could put ads though. I can't imagine that anyone thinks that that there will actually be a lot of people who spend time on slap talk, but maybe. You never know. irrational exuberance and all that. And thus you have the other side of the argument. The people who say that this is probably a bubble, they're saying, "Okay, well, there's demand for OpenAI services. People use them. They're using them more and more as time goes on. But you can't say that at the moment the demand is outstripping the supply. There's plenty of supply to meet the demand. And right now, OpenAI, according I think to them, they lose **[06:43]** money every single time you use one of their products. like the amount that it costs to train and use the models is more than the amount that they are charging even for their top $200 a month tier. In that case, it's possible that they could charge 10 times more to their enterprise clients and to their consumer clients and all those people would keep paying and they would be a profitable company. But would that actually work? Like would the $200 a month people pay $2,000 a month? But even more than that, is it true that the demand for Nvidia's **[07:13]** chips is increasing or is Nvidia creating that demand? And likewise, is it true that the demand for OpenAI services is increasing? Or are they increasing that demand? So, let's start with OpenAI. There's obviously a lot of arrows coming in and out of this. We're going to start with the ones that have green arrows. There's a healthcare company, Harvey, and Anphere. These are companies that pay money to OpenAI, but that OpenAI's venture arm has also invested in. So they're investing in these companies and then the companies buy services from OpenAI. That's a bit **[07:44]** of a loop. Now they're not investing like the exact same amount of money. Probably they're investing less than the companies are paying but that is an example of like a circular system here. We also see these green arrows which is venture capital pointing out from Nvidia to Maestro to figure and XAI and all of those people take that venture capital investment and then they buy Nvidia chips. So, if you're in the anti-bubble case, then you do have to ask, well, why does Nvidia have to invest in these companies in order for them to buy their not it's not like a quid proquo here, **[08:16]** but like why can't the companies just get investment from elsewhere outside of Nvidia to do those investments? It's a little weird and I think that it makes sense that people are worried about it. Now, none of these arrows have dollar values on them, which is a little bit frustrating for me, except for the most important one, the one that sort of created this whole conversation about roundtpping, which is this one. Nvidia has agreed to invest up to hundred billion in Open AI. And that investment comes along with the purchase of chips. So, to unlock more and more of that investment, they have to buy more and **[08:46]** more of the chips. That's basically paying Open AI to buy Nvidia chips. Now, it's not like the exact price of the chips. It's less than the price of the chips. So basically about a quarter of the money that the chips cost is getting discounted off the price of the chips. And in exchange, and I am simplifying here, but in exchange for that 25% discount, Nvidia is getting ownership shares in OpenAI. At the same time as that is happening, with this money going to OpenAI, OpenAI signs a $300 billion deal with the middleman, Oracle, who's now build the data centers and run the **[09:16]** data centers. $300 billion on revenue of 12 billion for a company that is not brand new, that has been around and that everybody knows about and that lots of people use and that lots of people have found a hard time figuring out how it integrates well into businesses in a way that is worth spending a lot of money. $12 billion. They have these huge contracts with lots of companies, but still $12 billion is the amount of revenue they have. They're spending $300 billion on a service contract. That's this darker blue line, a service contract with Oracle. So Nvidia invests in OpenAI. Open AAI signs a service **[09:48]** contract with Oracle. Oracle in order to build this data center has to spend tens of billions of dollars on chips. All these pink arrows going in to Nvidia. These are all these companies buying chips directly from Nvidia. So they're all paying money into Nvidia, which is why they're the ones who are winning the most in this. And also there's this one arrow pointing out of Nvidia to Cororeweave which is a service agreement saying Nvidia will lease its own chips that are being sold to Cororeweave. So Core is buying chips from Nvidia and then Nvidia has said we promise that **[10:19]** we're going to pay you to use those chips once they're in your data center. And not only that, they have promised that if Cororeweave, this data center company, if they don't get enough contracts to use all of the chips, Nvidia will just take all of that excess processing power and pay for it. So they're guaranteeing the value of the chips to Coreweave, why would they be doing this if the demand for the chips is so high and the profit margin on the chips is so high. So the anti-bubble argument that I have heard from analysts who are much smarter than me and know more about this is that the price keeps **[10:50]** going up for these Nvidia chips. And so they're like the demand is there, but the chip isn't the product. The AI service to consumers is the product. And right now that is an order of magnitude under the amount of money that is being spent just on hardware. Not even including staff time and the power and the water and the buildings and all of that stuff. The amount of money being made by AI companies is an order of magnitude lower than just the money that is going to Nvidia every year. And by doing deals that basically give **[11:21]** companies money to buy Nvidia's chips, they are selling their chips at an effective discount. And yet, at every one of these steps of the process, as this money flows around, these companies get more highly valued. When Oracle signed this $300 billion a year deal with OpenAI, their stock price shot up. When Nvidia does venture investment in XAI, which then turns around and buys Nvidia's chips, that is offering the chips at a discount. So, once you parse this fairly information dense, though there could be more information on it, and that would be good. But once you get **[11:53]** your mind around this little graphic here, it doesn't seem like Nvidia just thinks people are going to buy their chips at the prices they're selling them. Nvidia seems to think it would be better for our share price if we found ways to artificially increase the demand for our chips. Like by paying the people who buy our chips to buy our chips or investing in companies that buy our chips so they can buy our chips. And Nvidia has so much money that like what else are they going to do? Their price is so inflated. their stock is worth so much and they have so much money that they can subsidize the purchase of their **[12:23]** own chips and when they do that the stock market rewards them. This money is just flowing around and at the root of it is not the demand for Nvidia chips because Nvidia chips are not the product being sold. The product being sold is artificial intelligence systems. Look, it might someday be this big, but it is not anywhere close to that big right now and it's not brand new. In order for it to be this big, it has to create so much value. And we have to create new systems for capturing that value. Besides just **[12:54]** firing everyone and having computers do all the work, Microsoft is the only major player on this graph that is buying Nvidia chips that doesn't have a direct relationship. They do have this one little connection through Nebius, and I don't know what that is. Where Microsoft is paying Nebus for a service while Nebius is getting investment from Nvidia. My god, I don't know how big that is. You want to Google it with me? Oops. I don't want to show you all of my text messages. That was just my dad. It was probably okay. Don't read my text messages. Microsoft secures 100,000 Nvidia GB300 chips via 33 billion deal. **[13:27]** Okay, so it is big. That's actually huge. Oh my god. So Nebius is like an oracle but for just AI. I have gotten this take your guest house to the next level ad so many times. I probably accidentally clicked on it once and they're like, "Yeah, I bet you're interested in nano walls." I thought it was going to be nano walls, but it actually says nano wall. It's where you keep your nana. She's in the walls. So yeah, it's getting a 100,000 Nvidia chips from a company that Nvidia has invested in. Even Microsoft is getting discounted chips. I don't know you guys. Seems fine. I have seen the simplified version of this pitched around which is **[13:57]** like Nvidia gives $100 billion to OpenAI. OpenAI gives $100 billion to Oracle. Oracle gives $100 billion to Nvidia. It's not exactly like that. It does prop up the demand for Nvidia chips by making it so that you're giving money to people in exchange for them in some cases directly in exchange for them buying Nvidia chips or you're just investing in companies so that they can buy Nvidia chips. It just seems messy. It seems like bubble mechanics to me you guys. So I am again just a guy. I don't know why I god I got to get back to **[14:29]** talking about rare earths and fish. I got to get back in my lane. That's what you want anyway. You don't want this. I should be talking about batteries and climate change and the moons of Saturn and my favorite fish, but instead I looked at that figure for too long. Let's play Connections. I'm worried because the last like three have been really easy. Swift kite. Those are birds. Robin is a bird. Oh, interesting. We've got like Batman Begins, Batman Forever, Batman Eternal, Batman Returns. Is that what it is? Is that Batman thing? We got Swift and Robin. Robin is also a Batman thing. There's got to be **[14:59]** Oh, Swallow. Great. Those are all birds. But Swift kite and swallow are much more similar to each other than Robin. Is there another bird like that? Robin is just a normal bird. Swift kites and swallows. Very similar birds. I don't know why. Wow, this is weird. Everlasting and permanent. Okay, well, we got that one. Now, this looks like it's going to be easy, too. Everlasting forever, permanent, and eternal. Okay. Launches, institutes, pioneers, begins. And that leaves us with returns ending pioneers and days. Returns ending **[15:29]** pioneers and days. Uh, okay. Well, that does seem like what it is. And I don't I don't mind doing it real fast. Kite, Squall, Swift, Robin. Those are all birds, and they're the only birds here, so I feel like that's it. This is like creates and puts into the world, launches, institutes, pioneers, and begins. In addition to that, we've got things that go on forever, eternal, forever, permanent, and everlasting. Which leaves us with, I think, days, returns, ending, and trails. Happy days, happy trails, happy returns, happy ending. **[16:00]** [Applause] Happy trails. That has two meanings now. But there is another one if you're just thinking of the one. And then these are all the same to me. Maybe the bird is second. Yeah. Yeah. All right. And then I get 50/50 chance here. Everlasting institutes launches begins and pioneers. That seems a little bit harder. That was that was easy. That was another easy one. All right. I'm feeling it. Woo. Yeah. Another one of five. Okay. I don't know what's going on you guys. They're taking it easy on