S2E17 — The Two World Cups: Inside the AI fueled IPO Boom

S2E17 — The Two World Cups: Inside the AI fueled IPO Boom

Click here to watch a video of this episode. [https://www.youtube.com/watch?v=_SIirQKVqXI] On Friday 12 June 2026, a SpaceX rocket climbed out of Cape Canaveral as the opening bell rang in New York on the largest stock market debut in history. By the close, the company was worth more than $2 trillion and Elon Musk was the first trillionaire, and the traders still called the 19% pop disappointing. This summer America is hosting two World Cups at once. One is on the pitch: 48 nations, 104 matches, the final in New Jersey. The other is on Wall Street, where more than $2 trillion of company value is queuing for an IPO in the most concentrated listing run the public markets have ever seen. SpaceX has priced, Anthropic and OpenAI have filed within days of each other, and the question underneath all of it is why the smartest companies spent a decade refusing to go public at all. Aman Narain and Zubin Vandrevala break down the 2026 IPO boom: the five eras of the Silicon Valley listing, why a 1934 securities rule and the 2012 JOBS Act let trillion-dollar private companies exist, why Stripe still says no, the SPAC disaster and the discipline it left behind, the Goldman versus JP Morgan fee war, and the homecoming pulling Indian champions like Razorpay back to list at home. Key takeaways: 1. SpaceX's ~$75B raise and >$2T close is the largest IPO ever, yet the modest 19% pop may make it the most honestly priced debut in years. 2. The 2012 JOBS Act switched off the legal clock, Section 12(g), that once forced companies public. That is why Stripe and SpaceX could stay private for two decades. 3. The AI super-cycle has taken the market back to the 1980s: the money rewards asset-heavy steel and silicon, not asset-light software. 4. Stripe handles roughly 1.5% of global output and still refuses to list, because a payments IPO can be a relegation battle fought every 90 days. Adyen lost 39% in a single afternoon. 5. The 4,400 new millionaires from the SpaceX IPO are the seed capital for the next decade. The PayPal Mafia became Tesla and LinkedIn. The AI Mafia is being born now. Topics covered: - The five eras of the Silicon Valley IPO, and why 2026 rhymes with 1980 - Hyperscaler capex heading toward $650-700B in 2026, bigger than the GDP of Belgium - Section 12(g), the JOBS Act, and double-trigger RSUs explained - The June fortnight: Anthropic, OpenAI and the largest IPO in history days apart - Why Stripe ticks every box an IPO is meant to tick, without ringing the bell - The two doors out of payments: single-platform compounders versus PE roll-ups - The SPAC collapse and the January 2025 rules that killed the arbitrage - Goldman versus JP Morgan, and Bill Gurley's case that the IPO is rigged - China's Ant Group halt, Hong Kong's revival, and India's ghar wapsi homecoming Chapters: Referenced in this episode: SpaceX IPO and the $60B Cursor buyout; Anthropic and OpenAI June filings; CoreWeave, Figma, Klarna, Cerebras; Stripe's $159B tender and John Collison; Section 12(g) and the 2012 JOBS Act; Nuvei, Payoneer, Worldpay, Global Payments, Worldline; Adyen and Fiserv one-day falls; the SEC's 2025 SPAC rules; Goldman Sachs, JP Morgan, Bill Gurley; Databricks, Canva, Revolut, Kraken; Alibaba, Ant Group, Hong Kong; PhonePe and Razorpay; Microsoft, Google, Amazon, Uber and the PayPal Mafia. Related episodes: S2E16 (The People Who Knew First, the betting markets that priced the pop); S2E14 (Machines with Wallets) and S2E13 (The Google of Money) on Stripe. Hosted by: Creators & Guests * Aman Narain [https://www.a2zfintech.com/people/aman-narain] - Host * Zubin Vandrevala [https://www.a2zfintech.com/people/zubin-vandrevala] - Host Aman Narain writes at amanwhoblogs.substack.com. Zubin Vandrevala is your payments provocateur, reporting this week from dry land at last. Enjoying A2Z Fintech? Leave a rating and review on Apple Podcasts. It is the single biggest signal to the Apple algorithm and how new listeners in our world find us. For information and entertainment only. Not financial advice. Transcript: Click here to view the episode transcript. [https://share.transistor.fm/s/36665937/transcript]

S2E16 — Polymarket, Kalshi and the People Who Knew First

A US Army sergeant bet roughly $33,000 on Polymarket on an arrest he had just been briefed on, and turned it into the price of a house. A Google engineer read his own company's unpublished data and moved about $1.2 million into a private wallet. On a public blockchain, both men left a trail the FBI could follow to the cent. That is the paradox at the centre of prediction markets: the transparency that makes them exploitable is the same transparency that makes them honest. Between September 2025 and April 2026, combined monthly volume on these prediction markets climbed from under $5 billion to about $24 billion, and in October 2025 the parent of the New York Stock Exchange committed around $1.6 billion to the largest of them. Aman Narain and Zubin Vandrevala break down how prediction markets went from a forgotten Wall Street betting ring to information infrastructure, and the harder question beneath the boom: when a market knows before the news does, is the system working perfectly, or is it being robbed? Key takeaways: 1. The transparency that makes these markets exploitable is exactly what makes them honest: the engineer who hid behind a wallet was traced by the same chain he trusted. 2. The political framing is a myth. On Kalshi, 80% of volume is sports and just 4% is politics. 3. ICE bought the data, not the casino: roughly $1.6 billion for a live probability feed it can sell to every bank and hedge fund on its network. 4. Volumes ran from under $5 billion a month to about $24 billion in seven months, with Piper Sandler projecting $8 billion in annual revenue by 2030. 5. A prediction market is not an oracle but a mirror, only as honest as the room it is in. Topics covered: - The two cold-open exploits: a Fort Bragg sergeant and a Google engineer who bet on what they already knew - The intellectual lineage: Hayek on price as information, Hanson's futarchy, Tetlock's superforecasters - Wall Street's unregulated political betting ring, and the $10m wagered on the 1916 election - The modern revival: the Iowa Electronic Markets, Intrade's collapse, and the fall of PASPA - How a prediction contract actually works, and the passport-versus-wallet divide between Kalshi and Polymarket - The full board: Polymarket, Kalshi, PredictIt, Manifold, Metaculus, Robinhood, Interactive Brokers and DraftKings - The data that reorders the story: prediction markets are now mostly a sportsbook in a derivatives licence - The three forces behind the 2024-2026 boom: the KalshiEX ruling, the GENIUS Act, and a presidential endorsement - ICE's ~$1.6 billion move on Polymarket, and why it bought the data and not the gambling - The prosecution: four insider cases, reflexivity, and gambling at derivative scale - The bull case in three layers: parametric insurance, macro hedging, and information infrastructure Chapters: Referenced in this episode: US Army Master Sergeant Gannon Ken Van Dyke and Operation Absolute Resolve; the Google engineer "AlphaRaccoon" and the Year in Search exploit; the Israeli Air Force major and the June 2025 Iran briefing; the MrBeast editor's $4,000 Kalshi trade and $20,000 fine; Friedrich Hayek, Robin Hanson and futarchy, Philip Tetlock and the Good Judgment Project; the Wall Street curb-market and Tammany Hall betting ring, the 1896 and 1916 elections, and Governor Charles Evans Hughes; the Iowa Electronic Markets and their 1.34-point average error; Intrade and John Delaney; Murphy v NCAA; Polymarket on Polygon and Kalshi as a CFTC-designated contract market; PredictIt and Victoria University of Wellington; Manifold, Metaculus and Augur; Robinhood, MIAXdx and Susquehanna; Interactive Brokers and ForecastEx; DraftKings, Railbird and DKeX; KalshiEX v CFTC and CFTC chair Michael Selig; the GENIUS Act; ICE / Intercontinental Exchange and Jeffrey Sprecher; Piper Sandler's $8bn-by-2030 estimate; Boaz Weinstein and Saba; the 12 January 2026 single-day record of $701.7m; the Arizona pre-emption ruling and Minnesota Governor Tim Walz's ban. Related episodes: S2E14 — Machines with Wallets, for the stablecoin-rails and GENIUS Act thread; [TBD — the Mastercard / BVNK stablecoin episode, confirm number]. Hosted by: Creators & Guests * Aman Narain [https://www.a2zfintech.com/people/aman-narain] - Host * Zubin Vandrevala [https://www.a2zfintech.com/people/zubin-vandrevala] - Host Aman Narain writes at amanwhoblogs.substack.com. Zubin Vandrevala is your payments provocateur. Enjoying A2Z Fintech? Leave a rating and review on Apple Podcasts. It is the single biggest signal to the Apple algorithm and how new listeners in our world find us. For information and entertainment only. Not financial advice. Transcript: Click here to view the episode transcript. [https://share.transistor.fm/s/6d85f105/transcript]

S2E15 — Sundar's Long Game: Google I/O 2026

Click here to watch a video of this episode. [https://www.youtube.com/watch?v=Z3U1YMGKUNM] Twelve hours after Sundar Pichai walked off the Shoreline stage, the recap industry was already ranking his Google I/O 2026 announcements top ten this, best five of that. Two former Googlers with 50 years between them think that misses the story. On Wednesday 20 May, Sundar Pichai closed Google I/O 2026 with a 150-minute keynote that pulled together a 25-year arc. Google built the transformer in 2017. It declared itself an AI-first company at I/O 2016, six years before ChatGPT shipped. The press spent two years arguing Google was behind. Google I/O 2026 was the company quietly revealing it never was. Aman Narain and Zubin Vandrevala break down five non-obvious reads on Google I/O 2026, drawing on their combined years inside Google and Google Pay. Not a top-10 list. Five reads on the long game. Key takeaways: 1. Google has spent 25 years building the AI foundation everyone else is racing to construct in 24 months, and at Google I/O 2026 it stopped pretending that head start did not exist. 2. Gemini Spark is not a chatbot. It is a category response: AI woven into the Google surfaces where most users already live their digital lives. 3. The Universal Commerce Protocol and AP2 are Google writing the TCP/IP of agentic commerce, with Shopify, Visa, Mastercard, PayPal and Gr4vy inside the standards body and AP2 donated to the W3C. 4. SynthID got two minutes of stage time and is the most underrated announcement of the day. Apple made privacy a moat; Google is making trust the next one. 5. Android XR, with Samsung as OEM and Warby Parker and Gentle Monster as the face brands, is the platform play applied to wearable hardware. Fashion is how you blend in, not how you stand out. Topics covered: * Why search grew 19 per cent the year AI Overviews was meant to kill it * Vidhya Srinivasan, Antigravity, and the rebuilt search box * Gemini Spark as a category response, not a product * UCP, AP2, and the protocol coalition routing agentic commerce * The $5 trillion agentic commerce market and the fraud-liability question nobody is answering * Visa Trusted Agent Protocol versus Mastercard Agent Pay * SynthID, OpenAI, Nvidia, Eleven Labs, and the web's immune system * Android XR with Samsung, Warby Parker, and Gentle Monster * What banks, fintechs and merchants should build before Google I/O 2027 Chapters: Referenced in this episode: Google I/O 2026 keynote (Shoreline Amphitheatre, 20 May 2026); Sundar Pichai's "AI-first" keynote at Google I/O 2016; "Attention Is All You Need" (Google, 2017); AI Overviews and Search +19 per cent year-on-year; Antigravity demo from Vidhya Srinivasan; Gemini Spark from Josh and team, formerly NotebookLM; Universal Commerce Protocol with Shopify; Agent Payments Protocol with Visa, Mastercard, PayPal and Gr4vy, donated to W3C; Visa Trusted Agent Protocol; Mastercard Agent Pay; SynthID coalition with OpenAI, Nvidia and Eleven Labs; Android XR with Samsung as OEM and Warby Parker and Gentle Monster as face brands; AlphaGo demo on Android XR. Related episodes: [TBD] Hosted by: Creators & Guests * Aman Narain [https://www.a2zfintech.com/people/aman-narain] - Host * Zubin Vandrevala [https://www.a2zfintech.com/people/zubin-vandrevala] - Host Aman Narain writes at amanwhoblogs.substack.com. Zubin Vandrevala is your payments provocateur, recording live from the Shoreline floor at 2 per cent battery. Enjoying A2Z Fintech? Leave a rating and review on Apple Podcasts. It is the single biggest signal to the Apple algorithm and how new listeners in our world find us. For information and entertainment only. Not financial advice. Transcript: Click here to view the episode transcript. [https://share.transistor.fm/s/9fa1a267/transcript]

S2E14 — Machines with Wallets: From Buy Button to Agentic Commerce

Click here to watch a video of this episode. [https://www.youtube.com/watch?v=7c-UWxqoly0] Patrick Collison says agents will soon account for most online transactions. The Machine Payments Protocol that replaces the buy button is now live, and Stripe, OpenAI, Google, Meta and Microsoft have all signed on. This week at Stripe Sessions 2026, the architecture of agentic commerce stopped being a thesis and started being plumbing. The Machine Payments Protocol, co-authored by Stripe and Tempo, is live. ACP and UCP have settled the discovery layer. Streaming payments are real. Software-as-a-subscription is becoming software-as-a-stream. The buy button is being retired in real time. Aman Narain and Zubin Vandrevala close out the anniversary Double Digest with part two: how money moves when the buyer is no longer human. The 2026 Agentic Stack refreshed, the four horsemen of agentic commerce mapped, the dark horse from China called, the streaming-payments primitive that is quietly rewiring every business model in the next five years, and a six-twelve-twenty-four-month playbook for operators. Key takeaways: 1. The buy button is being replaced by the Machine Payments    Protocol, Stripe and Tempo's "OAuth for money" primitive. 2. The four horsemen are playing distinct hands: Stripe wants    the rails, OpenAI wants the brain, Google wants to protect    search, Visa and Mastercard are fighting to remain a rail    rather than the rail. 3. Streaming payments end batch settlement; software-as-a-    subscription becomes software-as-a-stream. 4. The operator playbook is fix data hygiene first, pick a    protocol stack second, reimagine for streaming third. 5. Stripe is not the next Visa. Stripe is the next Google. Topics covered: - The 2026 Agentic Stack: discovery, execution, safety - ACP and UCP, the two protocols solving merchant-to-agent   discovery - MPP, the Machine Payments Protocol, and why it is becoming   the TCP/IP of agent payments - Verifiable Credentials, Verifiable Intent, and the agent as   an actor with reputation - One in six AI sign-ups being a bad actor, and the rise of   token theft as the new fraud vector - The four horsemen of agentic commerce: Stripe, OpenAI,   Google, Visa and Mastercard - The dark horse from China: Alibaba, Qwen, Taobao, Alipay,   and 120 million agent transactions in a week - Streaming payments, software-as-a-stream, and the death of   batch settlement - The CFO problem: reconciling billing at the granularity of   tokens consumed - The six-twelve-twenty-four-month playbook for founders,   operators, and CFOs - Why Stripe is not the next Visa but the next Google - Anniversary reflections: 30 episodes, 128,000 words, and   the founder-mode payoff Chapters: Referenced in this episode: Stripe Sessions 2026; the Machine Payments Protocol (Stripe + Tempo); the Agentic Commerce Protocol (Stripe + OpenAI); the Universal Commerce Protocol (Google); JD Sports as UCP launch partner; Visa Agentic Ready APAC and LatAm expansion; the Mastercard-BVNK acquisition; Lightspark on Bitcoin Lightning; Stripe's Tempo, Bridge, Privy, Link and Metronome stack; the Fido Alliance agent-identity work; Stripe Radar's token-theft defences; Alibaba's Qwen, Taobao and Alipay; Reed Hastings, Steve Jobs and Andy Grove as re-platforming references. Related episodes: S2E13 — [TBD part one title], the "who" of the Anniversary Double Digest; the Mastercard-BVNK breakdown; [TBD prior agentic commerce episode covering OpenAI's Etsy Instant Checkout]. Hosted by: Creators & Guests * Aman Narain [https://www.a2zfintech.com/people/aman-narain] - Host * Zubin Vandrevala [https://www.a2zfintech.com/people/zubin-vandrevala] - Host Aman Narain writes at amanwhoblogs.substack.com. Zubin Vandrevala is your payments provocateur. Enjoying A2Z Fintech? Leave a rating and review on Apple Podcasts. It is the single biggest signal to the Apple algorithm and how new listeners in our world find us. For information and entertainment only. Not financial advice. Transcript: Click here to view the episode transcript. [https://share.transistor.fm/s/1bbc380b/transcript]

S2E13 — The Google of Money: From Seven Lines of Code to Indexing the Economy

Click here to watch a video of this episode. [https://www.youtube.com/watch?v=xOvbJ8ULNdw] Stripe shipped 288 features in a single morning at Sessions 2026 in San Francisco. The right pattern match for what the company is becoming is not Visa, Mastercard, or PayPal — it is Google in 2006. On the final Wednesday of April, almost ten thousand founders, engineers, and CFOs queued around Yerba Buena Gardens, the same plaza where Steve Jobs unveiled the iPhone nineteen years earlier, to hear Patrick and John Collison announce the largest single-morning product launch in fintech history. (334 if you count the entries on the public roadmap, per Stripe's own footnote.) The 288 launches grouped into five buckets, three of which would have been laughed out of a payments conference five years ago. Aman Narain and Zubin Vandrevala break down what Stripe Sessions 2026 actually shipped, why payments was always the wedge rather than the company, and the long-arc historical parallel that explains the architecture. This is part one of an A2Z Fintech anniversary special, marking 30 episodes and one year since the show's first recorded cameo, on Stripe Sessions 2025. Key takeaways: 1. Stripe shipped 288 product launches at Sessions 2026 in a single morning, grouped into five buckets that share one thesis: payments was the wedge, not the company. 2. Three transformations are running simultaneously: rails (Tempo, Bridge, Privy, Link, the Machine Payments Protocol), mission (GDP of the internet to economic infrastructure for AI), and customer (developer-led to enterprise-led). 3. The right pattern match for Stripe is not Visa, Mastercard, or PayPal. It is Google in 2006: a company that started narrow, ate every adjacency, and built its moat on the index rather than the algorithm. 4. Stripe processed roughly two trillion dollars in payment volume in 2025, equivalent to about 1.5 percent of global GDP, at a 40 basis point take rate that can only be grown by stacking software on top. 5. The unit of growth has changed: Stripe used to grow when more humans bought things online; it now grows when more businesses are created, which is parabolic in the AI era. Topics covered: - The five buckets of Sessions 2026: stablecoin shadow banking, agentic commerce, the full-stack business bank, AI-native revenue, trust at agent scale - Tempo, Bridge, Privy, Link, and the Machine Payments Protocol as the kit you build to be a network, not a payment processor - Why a Brazilian merchant paying a freelancer in Manila no longer needs to touch SWIFT - The Visa-in-1976 parallel for Stripe's stablecoin posture - The mission shift: GDP of the internet was horizontal; economic infrastructure for AI is vertical - Eileen O'Mara, Tyler Bryson, and the quiet construction of a Salesforce-grade enterprise sales motion - Why financial infrastructure now sits in the same enterprise budget category as cloud infrastructure - Two trillion in TPV, 40 basis points, and how that take rate grows from here - The Google of money thesis: phase one to phase four, search ads to Vertex AI, the index as the real moat - Founder leverage and why Patrick and John can run R&D budgets that look more like Bell Labs - A tee-up for Part 2 (S2E14, publishing next week): the Machine Payments Protocol, the protocol war between Visa, Mastercard, OpenAI, Google and Meta, and the playbook for founders Chapters: Referenced in this episode: Stripe Sessions 2026; the 288 launches and 334 footnote; Tempo (layer 1 blockchain co-built with Paradigm, mainnet March 2026); Bridge ($1B+ stablecoin orchestration acquisition); Privy (110M programmable wallets); Link (250M consumer wallets); the Machine Payments Protocol; Metronome; Patrick Collison's "GDP of the internet" mission; Stripe Atlas, Treasury, Billing, Tax, Capital, Issuing, Connect, Radar, Sigma; Stripe's internal tender at ~$107B (reports as high as $159B); TPV ~$2T in 2025, up 34%; net revenue $5.1B; free cash flow over $2B; 40bps take rate; Fortune 100, 500, top US software, and Forbes AI 50 penetration; Eileen O'Mara, Tyler Bryson, Will Gaybrick, Emily Sands, Mike Clayville; the Steve Jobs iPhone keynote at Yerba Buena Gardens, January 2007; Andy Grove, Jensen Huang, Sergey Brin, Max Levchin and the immigrant founder pattern; Google's 1998 mission and the search-to-cloud arc; Google Cloud at ~$50B run rate in 2025. Related episodes: S2E14 — The Google of Money, Part 2: Rewriting the Rules of Agentic Commerce [link TBD] (publishing next week); A2Z Fintech's first cameo on Stripe Sessions 2025 [TBD]; the Mastercard / BVNK breakdown [TBD]; S2E11 — Pit Wall, Podium and Pie: Q1 2026 Fintech Scorecard [TBD]; S1E16 on the one-person enterprise [TBD]. Hosted by: Creators & Guests * Aman Narain [https://www.a2zfintech.com/people/aman-narain] - Host * Zubin Vandrevala [https://www.a2zfintech.com/people/zubin-vandrevala] - Host Aman Narain writes at amanwhoblogs.substack.com. Zubin Vandrevala is your payments provocateur. Enjoying A2Z Fintech? Leave a rating and review on Apple Podcasts. It is the single biggest signal to the Apple algorithm and how new listeners in our world find us. For information and entertainment only. Not financial advice. Transcript: Click here to view the episode transcript. [https://share.transistor.fm/s/ca994355/transcript]