A2Z Fintech

Click here to watch a video of this episode.

On 28 October 2026 Jio Platforms listed in Mumbai at 1,358 rupees a share, 21% above its issue price, a value of about $130 billion. The Jio IPO was 22 times subscribed, yet retail investors, the people Dhirubhai Ambani built Reliance on, barely turned up. The twist came from Reliance itself: answering Musk's charge that oligarchs were blocking Starlink, it disclosed that Jio's own satellite service is waiting for the same security clearance.

Aman Narain and Zubin Vandrevala break down what the Jio IPO was really for, whether Jio is the SpaceX of India or its Bell System, and what a landlord with a listed share means for every fintech on its rails.

Key takeaways:
1. The Jio IPO was not a growth raise: nobody sold a share and most of the $3.2 billion repays debt, so what it really bought was a listed share price to spend.
2. Jio and SpaceX run the same playbook: win the allocation, collapse the unit cost, own the stack and let the cash machine fund the moonshot.
3. In structure Jio is closer to the Bell System than to SpaceX: one network, one country, one government, until the state changes its mind.
4. Musk alleges that Ambani is locking Starlink out of India, yet Jio's own satellite service is waiting on the same security clearance.
5. Every Indian fintech is a tenant on Jio's rails, and a landlord with a listed currency does not raise the rent. He buys the tenants.

Questions answered:
- What is the Jio IPO? India's largest: Jio Platforms raised 30,213 crore rupees (about $3.2 billion) in new shares and listed on 28 October 2026.
- Why is Starlink not available in India? It has its licence. What it lacks is a Home Ministry security clearance, the same one Jio's satellite service is waiting for.
- Is Jio the SpaceX of India? In ambition and in its dealings with the state, yes. In structure it is closer to the old Bell System.

Topics covered:
- The listing: 21% up on day one, 70 times subscribed by institutions, under three times by retail
- Why 42 times earnings only works if revenue per customer keeps rising
- How a non-compete that expired on 23 May 2010 let Mukesh Ambani back into telecoms
- Four moves Jio shares with SpaceX, and Zubin's case that it is really the Bell System
- The satellite spectrum fight Jio lost in 2024, and Musk's three posts in October 2026
- Whole or parts, and the bet: Jio or Airtel by November 2027

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.

Chapters:

• (00:00) - Cold open: Musk, X, 8 October 2026

Corrections: Reliance's statement was issued on 16 October 2026, not 9 October. The IPO was 22.19 times subscribed overall and 2.72 times by retail investors, not 21 and 2.76. Dhirubhai Ambani died in 2002; the brothers divided the group in 2005.

Related episodes: S2E25, UPI at 10; S2E17, SpaceX's $2 Trillion IPO vs the AI Listing Race.

Hosted by:
Creators & Guests

For information and entertainment only. Not financial advice.

Transcript:
Click here to view the episode transcript.

Click here to watch a video of this episode.

From Piggly Wiggly in 1916 to Alipay's escrow in 2004, each leap in shopping removed a friction and needed a new trust device. Today AI agents research, recommend and fill the basket, yet they are barely buyers. China runs agent shopping inside super-apps, India builds on public rails, and the West is split between open protocols and walled gardens.

Aman Narain (Singapore; ex Standard Chartered, HSBC and Google Pay) and Zubin Vandrevala (San Francisco; your payments provocateur) break down the future of shopping in Part two of The Future Of, and answer the question that decides it: when AI agents do the buying, who pays when they get it wrong?

Key takeaways:
1. Every shopping revolution looked like a technology story and turned out to be a trust story; AI agents will be no different.
2. Amazon's stand-off with Meta's Muse is about the billboard, not the checkout: an agent never sees the ads.
3. AI is a research engine and barely a buyer today: discovery moves first, checkout follows later.
4. Know your customer becomes know your agent, and whoever underwrites the software's mistakes will own the customer.

Topics covered:
- Four frictions: distance, information, trust and attention
- Three operating systems: China, India and the West
- Amazon versus Walmart on AI agents, and the ad money at stake
- CART, the Delegation Ladder, know your agent and the Money OS

Chapters:

• (00:00) - Cold open: Memphis, 1916, and the first self-service shop

Related episodes: S2E31, The Future of Work; S2E30, Bot Wars; S2E25, UPI at 10.

Hosted by:
Creators & Guests

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.

Click here to watch a video of this episode.

In this first episode of The Future Of, a three-part mini-series on how you bank, how you shop and how you work, Aman Narain and Zubin Vandrevala return to a conversation first recorded late last year. The future of work, they argue, runs through the one-person enterprise: a creator, freelancer or sole trader who looks like a company on paper but has no employees, no payslip and no HR. Platforms such as Uber, Shopify and YouTube now supply the sales, marketing and operations, and increasingly the finance function. The banks underneath have not caught up.

Aman Narain and Zubin Vandrevala break down what the future of work demands from the financial system: real-time visibility of income, instant and programmable payments, credit that reads cash-flow futures rather than payslips, portable pensions, and insurance for algorithmic risk.

Key takeaways:
1. Platforms such as Uber, Shopify and YouTube now act as the CFO layer for independent workers, running payroll, treasury, working capital and tax, while banks become the regulated infrastructure underneath them.
2. Credit built on payslips cannot see a creator with 400,000 subscribers and about $180,000 a year, so banks must underwrite cash-flow futures from platform data rather than income history.
3. Four forces explain why work feels broken: walls, human agency, automation and modularity, which the hosts spell as WHAM.
4. A new Money OS for the future of work needs five primitives: see, move, borrow, grow and protect your money.
5. Both hosts make a dated bet: by the end of 2027 a major platform will auto-save into a portable pension by default, and a top-ten Asian bank will approve home loans on platform earnings data, not a payslip.

Topics covered:
- The 1985 job interview against the 2025 AI screen
- Carlota Perez, the vibecession and the chaotic middle between two eras
- WHAM: walls, human agency, automation and modularity
- Why the future of work looks like a playlist, not an album
- Why US business formation hit a record after 2020, and India's 63 million MSMEs
- The one-person enterprise at the heart of the future of work, and platforms as the new C-suite
- Why platforms are becoming your CFO, with banks as the regulated layer beneath
- Five primitives of a new Money OS: see, move, borrow, grow, protect
- Five risks and three upsides of the independent economy
- Closing wagers: platform auto-pensions or payslip-free home loans by the end of 2027

Chapters:

• (00:00) - Cold open: a job interview, 1985

Referenced in this episode: Carlota Perez on technological revolutions; the WHAM framework; an MIT research report on automation and job polarisation; US Census Bureau Business Formation Statistics; India's MSME sector; Uber, Shopify, YouTube, Stripe and Amazon FBA as financial operating systems; Brazil's Pix and India's UPI; platform-worker pension policy in Singapore, France and the UK; PinBox; platform cooperatives.

Related episodes: S2E18, The Stablecoin Endgame Nobody's Talking About; The Future Of: Banking and Shopping [TBD: add links when published].

Hosted by:
Creators & Guests

Aman Narain writes at amanwhoblogs.substack.com. Zubin Vandrevala is your payments provocateur.

Disclosure: Aman Narain is a board adviser to PinBox, which is mentioned in this episode.

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.

Click here to watch a video of this episode.

In seven weeks, Meta launched Muse (8 September), SpaceXAI put Grok Bot into beta, OpenAI unveiled dots (29 September) and Instinct raised $1 billion at a $10 billion valuation. Three of those AI agents can spend today, with Stripe's Link wallet behind Muse, Grok Bot and Instinct. The twist is in the contracts: on paper the user pays, the bot never does, and under US rules the bank may end up writing the refund.

Aman Narain and Zubin Vandrevala break down who pays when AI agents make mistakes, who moves the money behind them, and why Meta, OpenAI and Instinct gave their AI agents a face while Google and Anthropic did not.

Key takeaways:
1. On the contract, you pay: Meta and Instinct both put liability for an AI agent's purchases on the user, and in no version does the bot pay.
2. A bot's silence is the new pending: the failure that nearly cost Aman a meeting was a confirmation that never came, not a wrong charge.
3. Muse, Grok Bot and Instinct all pay through Stripe's Link wallet, yet the power sits at the two ends: the bot you trust and the shop with the goods.
4. Amazon blocked Muse twelve days after launch, and Sierra and Meta's answer is a standard that says who is knocking but not who pays.
5. Before an AI agent gets your card, ask three things: does it ask first, what is the cap, and who do you call.

Topics covered:
- Muse, Grok Bot, dots and Instinct: which AI agents can spend, and which cannot yet
- The two clauses that decide liability: Meta's help page and Instinct's section three
- Why silence, not a wrong charge, is the failure mode
- How an agent purchase works: request, approve, issue, pay
- Regulation E and the bank as the unwitting insurer
- Amazon's block of Muse, and the Personal Agent Protocol from Sierra and Meta
- India's NPCI and its Unified Agent Protocol
- Why Meta, OpenAI and Instinct gave agents a face, and why Google and Anthropic did not
- Ask, cap, call: three questions before an AI agent gets your card

Chapters:

• (00:00) - Cold open: Maya, Roxy and a $1,200 room, Austin

Referenced in this episode: Meta Muse launch and help page; Stripe Link wallet for agents; Instinct Series C; Grok Bot beta; OpenAI DevDay and dots; Amazon's block of Muse; Sierra and Meta Personal Agent Protocol; NPCI Unified Agent Protocol; Regulation E; White House executive order on "Super Intelligence"; Microsoft Copilot's Mico.

Related episodes: Google I/O 2026 (Gemini Spark); Stripe and the Money OS [TBD: title and link]; S2E28, Revolut [TBD: link].

Hosted by:
Creators & Guests

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.

Click here to watch a video of this episode.

On 30 September, a16z's Growth team, led by David George, published State of Markets II. The deck makes a serious case: tech delivered about 76% of the S&P 500's earnings growth this year, and valuations fell while profits rose. The twist is in the financing. Meta's Hyperion data centre sits in a vehicle 80% owned by Blue Owl's funds, the Chicago Fed puts large US banks' committed AI-related lines at about a quarter of their core capital, and the AI bubble debate has quietly moved from shareholders to banks, insurers and pension funds.

Aman Narain and Zubin Vandrevala boil the deck down to five mega trends and ten takeaways for financial services, score every argument, and walk through, day by day, what happens if the AI bubble bursts.

Key takeaways:
1. a16z is right about the companies and silent on who pays for them: in one quarter Alphabet and Amazon booked about $150 billion of paper gains on private AI stakes, roughly 12% of S&P 500 earnings on Goldman's estimate.
2. Finance has become the balance sheet of AI: Meta's Hyperion data centre sits in a vehicle 80% owned by Blue Owl's funds, with about $27 billion of bonds behind it and Meta as the tenant.
3. The Chicago Fed puts large US banks' drawn AI-related loans at about 9% of Tier 1 capital and committed lines at about 25%. The gap is what nobody sees until a bad week.
4. AI efficiency is real and gets competed away. The saving goes to the customer, and part of the AI budget comes straight out of headcount.
5. If the AI bubble bursts, pensions and insurers face a markdown rather than a wipe-out. The danger is that every loss arrives in the same week.

Topics covered:
- Why a16z's deck deserves both respect and a disclaimer: the umbrella seller's weather forecast
- Five mega trends, argued both ways and scored: earnings, capex, atoms, cheap intelligence, private markets
- The merry-go-round between cloud giants, AI labs and lenders
- Linda in Loudoun County, and what data centres do to a household electricity bill
- How Meta rents its own Hyperion data centre, and why Steve Eisman reaches for Enron
- Graham's pension: one capex cycle in four wrappers
- Chip-backed loans, data-centre securitisations and a CDO question from London in 2000
- Fintech funding splitting in two: Stripe and Revolut, then everyone else
- Agentic payments: machines paying machines before consumers do
- What happens to pensions and life insurance if the AI bubble bursts, and the likelier plateau
- Four dated predictions, to be checked on 6 October 2027

Chapters:

• (00:00) - Cold open: Pokémon cards beat the Nasdaq

Related episodes: Why the AI Bubble Won't Burst (Even If They Say It Will); S2E28 — Revolut-ion; Could a Kitchen Table in Ohio Decide AI's Future in November?; S2E17 — SpaceX's $2 Trillion IPO vs the AI Listing Race; S2E16 — Polymarket, Kalshi and the $1B Bet on Who Knew First.

Hosted by:
Creators & Guests

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.

Click here to watch a video of this episode.

In September Nik Storonsky won conditional approval from the OCC to form Revolut Bank US, sat down with the Banque de France in a black suit and a red tie, and watched two data breaches reach Revolut customers inside a fortnight. The founder who spent years being publicly rude about regulators is now dressing for them, and for the investors who will price a Revolut listing at up to $200 billion.

Aman Narain and Zubin Vandrevala trace Storonsky from Dolgoprudny and the Lehman trading floor to the Revolut of today, and ask whether he has built anything that works when he is not in the room.

Key takeaways:
1. Revolut's scar is Lehman, not Moscow: it lends under 5p of every deposit pound, where Nubank lends roughly 58p.
2. Nubank, Revolut and SoFi run the same flywheel but answer one question differently: a deposit is for credit, for float, or for software.
3. Removing credit risk bought rate risk. When rates fall the float shrinks, as Wise has already shown.
4. Citi, HSBC and Standard Chartered retreating as Revolut advances is one spreadsheet read from opposite ends.
5. Storonsky took judgment out of lending, hiring and investing, and kept all of it in one room: his own.

Topics covered:
- Dolgoprudny, the 1998 default, Phystech and the Lehman floor on 15 September 2008
- A sub-5% loan-to-deposit ratio, the contested 40-50% return on equity, and the ECB's 4.5% capital add-on
- Nubank, Revolut and SoFi: one flywheel, three ladders
- Mexico's 15% savings rate, and the rate risk inside the float
- Forty direct reports, Karma scores and five risk chiefs in seven years
- Revolut Pay, Spanish ATMs, a £500,000 private bank, Revolut People and AIR
- John Reed's Citi, success transfer, and the incumbents' retreat from retail
- Ports versus platforms, and why Durbin may tax the US charter
- The case against: fines, a secret ECB freeze, two breaches and sanctions
- The IPO test, and two closing wagers

Chapters:

• (00:00) - Cold open: Nik Storonsky's suit, Washington, April 2026

Related episodes: Nubank and the Purple Revolution; Wedge, Ladder, Driver; S2E20 — The Real Price of PayPal.

Hosted by:
Creators & Guests

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.

Click here to watch a video of this episode.

On Saturday 12 September, Anthropic chief executive Dario Amodei published We Must Pace the Frontier, a call for an AI slowdown in how fast frontier models gain capability. Within hours Elon Musk replied "Dario is right" and Sam Altman told Fortune there would be no OpenAI IPO this year. By Monday, President Trump had called AI fears a hoax and phoned Jensen Huang live at the All-In Summit. The twist: the call for an AI slowdown came in the same quarter Anthropic booked $11.5 billion of revenue and positive adjusted operating profit.

Aman Narain, solo this week while Zubin Vandrevala is in London, breaks down who actually wants an AI slowdown, who only says so, and who gets paid whichever way it goes.

Key takeaways:
1. Almost nobody changed position this week. What changed is which kind of doom each player chose to say out loud.
2. Read the capex, the PAC cheques and the IPO calendar: what AI leaders say moves weekly, what they spend does not.
3. The call for an AI slowdown arrived in the quarter Anthropic proved the frontier pays.
4. Pacing is not stopping. In a boom Nvidia sells the training, in a slowdown it sells the testing, and inference carries on regardless.
5. The doom that decides the US midterms is not extinction. It is the electricity bill of a voter who lives near a data centre.

Topics covered:
- Five things in fourteen days: escaped agents, a resignation, Nvidia buying Hugging Face, the essay, the speakerphone
- A four-box map: AGI believers and sceptics, accelerators and restrainers, and the two people who move
- Three dooms and two booms: extinction, displacement and deflation against capability and capital
- Say versus pay: Sam Altman's four positions in eleven years
- David Sacks's cartel charge, Jensen Huang's inspectors, and 100 gigawatts of physics
- The circular trade: Nvidia's equity in its customers, and the Lucent precedent
- Why an AI slowdown does not shrink Nvidia's order book
- Data centres, power bills and the midterms
- China: invoked everywhere, included nowhere
- Four calls for 2026, including where Anthropic's IPO prices
- Closing wagers: prophet or cynic?

Chapters:

• (00:00) - Cold open: $2 trillion changes its mind

Related episodes: SpaceX, Anthropic, OpenAI: The $2tn IPO Boom; Google I/O 2026: Sundar Pichai's 25-Year AI Long Game; S2E24 on Google's AI leadership reshuffle.

Hosted by:
Creators & Guests

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 watch a video of this episode.

On Wednesday 2 September, Anthropic open sourced two reference agents for agentic commerce: one facing the shopper, one facing the merchant. They place no orders, touch no card credentials, and hand the cart back to the merchant's own checkout. Shopify, Visa, Mastercard and Accenture were partners on day one. The twist is that OpenAI had already tried to own that checkout, charged 4% on completed orders for the privilege, and watched merchants leave. Meanwhile Google turned the same in-chat checkout on inside Gemini via Google Pay, with Gap live, and can price it at zero for as long as it likes.

Aman Narain, solo from Hong Kong this week, breaks down why the agentic commerce fight was never about whether AI can shop but about whether the AI shows up as a supplier or a landlord, and why the agentic commerce blueprint Anthropic gave away is really a distribution channel for embedded lending.

Key takeaways:
1. A percentage of revenue is rent: it scales with your success and cannot be engineered down, which is why merchants walked away from OpenAI's 4% fee.
2. Compute is an input like electricity: it can be cached, shrunk and negotiated, which is why Anthropic priced itself as a supplier.
3. Walmart's 3x conversion gap did not prove agents cannot shop. It proved the cash register was in the wrong place.
4. Google runs agentic commerce checkout at 0% because it makes its money one layer up in advertising, and that is precisely its trap.
5. The merchant agent is an underwriting file assembling itself in real time, which is why Shopify, Stripe and Square, all lenders, integrated on day one.

Topics covered:
- Walmart's ChatGPT experiment and OpenAI's retreat to discovery
- The mechanics of Anthropic's commerce agents: no orders, no cards, no cut
- Rent versus compute: why a 4% take rate wipes a direct-to-consumer margin
- Three lessons Anthropic welded into the plumbing: price, till, guardrails
- Google's Universal Commerce Protocol, Gap on Gemini, and the advertising trap
- The 35% larger cart claim, and why vendor-reported ceilings are not forecasts
- Shopify switching agentic storefronts on by default
- The merchant agent as a real-time underwriting file: Square, Shopify Capital, and the Tuesday stockout
- Three companies, three postures, and the closing wager: research lab or underwriter?

Chapters:
Referenced in this episode: Anthropic Claude Commerce Agents blueprint (2 September 2026); OpenAI Instant Checkout wind-down (March 2026); Walmart's Shoptalk conversion disclosure; Google Universal Commerce Protocol and Gap's Gemini checkout; Google Universal Cart at I/O 2026; Shopify Agentic Storefronts; Square Loans 2024 originations; Shopify Capital; Fast Company reporting on Gemini checkout.

Related episodes: SpaceX, Anthropic, OpenAI: The $2tn IPO Boom; Google I/O 2026: Sundar Pichai's 25-Year AI Long Game; Mastercard's $1.8B Power Move: Why They Outbid Coinbase for BVNK.

Hosted by:
Creators & Guests

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.

Click here to watch a video of this episode.

On 25 August 2016 the Unified Payments Interface went live in Indian app stores, four months after a Mumbai pilot with 21 banks. The architecture was deliberate: a regulator, a non-profit switch, volunteer architects who insisted the centre stay dumb, and private apps free to fight each other on top. The twist is what that fight produced. Google Pay held 62% of UPI volume in 2018 and has since fallen to around 35%, beaten not by a better product but by a field army of QR stands and soundboxes.

Aman Narain and Zubin Vandrevala break down how UPI went from 373 transactions to 241 billion a year, why the app that won the ground war now carries the largest regulatory overhang in consumer fintech, and who actually pays for a payment rail that charges merchants nothing. Both were inside Google Pay during the Tez years, so this is an operator's account rather than an anniversary brochure.

Key takeaways:
1. UPI cleared 241 billion transactions in FY26 against roughly two crore in FY17, a twelve-thousand-fold increase in ten years.
2. Google built the better app and PhonePe built the better army; on open rails the army took the crown.
3. NPCI's 30% per-app cap, mandated in 2020 and deferred ever since, makes Google Pay's decline and PhonePe's success the same regulatory problem seen from opposite ends.
4. India chose a free rail funded by public subsidy; Brazil chose a priced rail at 22 to 23 basis points that funds its own servers.
5. UPI carries 85% of India's retail digital payments and 49% of global real-time volume, on infrastructure no company owns.

Topics covered:
- The India Stack: Aadhaar, payments, data, and the KYC cost that fell from about twenty-three dollars to ten cents
- Why abstracting real time, not inventing it, was UPI's actual innovation
- Three fights nobody remembers: USSD against IP, banks against third-party apps, SMS OTPs against device binding
- How the baton survived a change of government, from the UPA to the NDA
- Demonetisation as fire starter rather than cause
- Inside Tez: scratch cards, ultrasonic pairing, and a multinational betting on rails it did not own
- PhonePe's ground war, and why the last metre in India is physical
- The 30% cap, the sovereignty subtext, and the crown that eats
- Seven hundred million QR codes in a country that never installed nine million card terminals
- Brazil's Pix: the tidier system, and what India's messier bet bought instead

Referenced in this episode: UPI pilot, Mumbai, 11 April 2016; public launch 25 August 2016; NPCI incorporated December 2008; IMPS, November 2010; iSPIRT, February 2013; demonetisation, November 2016; PhonePe founded December 2015; RBI action on Paytm Payments Bank, January 2024; NPCI's 30% per-app volume cap; zero MDR since 2020; NPCI International and the acceptance corridors; BIS Project Nexus; Brazil's Pix, November 2020; World Bank on Indian financial inclusion; the 1930 cybercrime helpline.

Related episodes: S2E10 on Mastercard and BVNK; S2E20 on the PayPal board fight; The Stablecoin Endgame Nobody's Talking About.

Hosted by:
Creators & Guests

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.

Chapters:
Transcript:
Click here to view the episode transcript.

Click here to watch a video of this episode.

On 5 August, Google DeepMind announced that Demis Hassabis would step back from day to day management to become Chair of Google DeepMind and Chief Scientist of Alphabet. Koray Kavukcuoglu took over as senior vice president reporting to Sundar Pichai. The same morning, chief scientist Jeff Dean left after 27 years, taking Sanjay Ghemawat, Oriol Vinyals and Quoc Le with him to found Discovery Loop. Alphabet is a founding investor and cloud partner in the company its own researchers left to build.

Aman Narain and Zubin Vandrevala break down why the Google DeepMind reshuffle was not a talent story at all. It was an argument over TPU allocation between Google DeepMind's researchers and Google Cloud's enterprise arm, and it was settled in Cloud's favour.

Key takeaways:
1. The reshuffle was not a talent crisis. It was a compute allocation decision, settled in Google Cloud's favour.
2. Berkshire's $37.8 billion Alphabet position was built during the second quarter and only disclosed on 14 August. It is not a dip purchase.
3. Sergey Brin holds no operating title at Alphabet and shaped the outcome anyway.
4. Google DeepMind's independence, protected since the 2014 acquisition, ended at an all-hands on 6 August.
5. The $200 billion that came off Alphabet's market value is roughly a full year of its own AI infrastructure budget.

Topics covered:
- What actually moved between 5 and 16 August, and what else was moving in the same window
- Berkshire's 13F, the $10 billion private placement, and why the timing matters
- Sergey Brin's April town hall and the limits of a co-founder with no operating title
- Why Google Cloud's leadership wanted Koray Kavukcuoglu in the seat
- The all-hands that moved non-technical teams into corporate Google
- Gemini 3.5 Pro, two months late, and the Flash release that arrived instead
- Jeff Dean's 4,400 people and Discovery Loop's reported $10 billion ask
- Why Alphabet is bankrolling the company its own legends left to build

Chapters:
Referenced in this episode: Alphabet leadership announcement, 5 August 2026; Google DeepMind all-hands, 6 August 2026; Reuters exclusive on the executive moves, 12 August 2026; Gemini 3.7 Flash release, 13 August 2026; Berkshire Hathaway Q2 2026 13F, filed 14 August 2026; Alphabet's $80 billion equity raise, 1 June 2026; Alphabet 2026 capex guidance of $195bn to $205bn.

Related episodes: Google I/O 2026: Sundar Pichai's 25-Year AI Long Game; SpaceX, Anthropic, OpenAI: The $2tn IPO Boom; S2E20 The PayPal Deal.

Hosted by:
Creators & Guests

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: