Is Google positioned to win the AI model race given its full-stack advantages?

Richard Socher · Aug 18, 2023

Google will not make its default experience chat-first overnight because of an innovator's dilemma: it makes about $500M a day from privacy-invading advertisements on a carefully tuned results page

Replacing the ad-laden results page with a chat interface would destroy hundreds of millions of dollars a day of revenue on a page whose every detail has been A/B tested, so they will only move slowly and carefully

Scope: they will still move slowly in the default experience; this opening does not make it easy for startups

28:12 20VC: Does Value Accrue to Incumbents or Startups in the AI Race, Why Model Size Matters More Than Data Size, Why Artificial General Intelligence is Far Away, Why Carpenters Will Be Paid More Than Software Engineers & Future of Jobs with Richard Socher

Harry Stebbings · Nov 15, 2023

Google does not want to build 'Google 2' because doing so cannibalizes its core business

The cost of building the LLM-first product eats the existing revenue engine, like Netflix's self-disruption dilemma

36:27 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Jeff Seibert · Nov 22, 2023

Google is by far the most vulnerable of the big incumbents in AI

Their business model is binary — search is all their revenue, so damage there is a huge problem — and they've been slow to react, merging two ML/AI teams and punting Gemini to Q1, which suggests it isn't going well

38:25 20VC: Why OpenAI Will Become an Infrastructure Play, Why Apple Will Win in an AI World, Why Google is the Most Vulnerable Incumbent, Will LLMs Be Commoditised, Which Startups Are Thin vs Thick Wrappers on Top of LLMs with Jeff Seibert, Founder @ Digits

Elias Torres · Mar 21, 2025

Incumbents are not actually moving fast on AI — they are talking a big game while their shipped products fail

Google invented the transformer paper yet everything it ships doesn't work (NotebookLM already effectively dead); Microsoft got early access via OpenAI but Bing search and Copilot aren't being used

13:27 20VC: Selling Drift for $1.2BN is the Biggest Failure: What No One Tells You About Selling Your Company | Why Incumbents Are Slower & Worse Than Ever | Why the Most Valuable Companies in a World of AI Will Not Have More Than 100 People with Elias Torres

Richard Socher · Apr 18, 2025

Google's problem in AI has never been technical strength but classic innovator's dilemma

They make money showing ads in lists of blue links, so giving users a straightforward useful answer cannibalizes a business printing ~$500m a day; absent competitive pressure they would not change the experience

Scope: they are being forced to experiment now because pressure is too high

18:33 20VC: Foundation Models: Who Wins & Who Loses | How Economies and Labour Markets Need to Change in a World of AI | China vs the US in an AI Race: What You Need to Know | Rich Socher, Founder @ You.com

Edwin Chen · Jul 21, 2025

Google's central difficulty is that building something better requires accepting a short-term hit to advertising revenue, which is extremely hard to do

58:08 20VC: Scaling to $1BN+ in Revenue with No Funding: Surge AI | The Most Insane Scaling Story in Tech |

Zach Lloyd · Oct 17, 2025

Google is at risk because it has a big innovator's dilemma problem and is moving too slowly on AI

Incentive structures rather than talent; he finds himself using ChatGPT more and more

Scope: would not have said this five to ten years ago

7:36 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Zach Lloyd · Oct 17, 2025 · hedged

Google's culture is too slow to compete with fast-moving rivals like OpenAI, though its size and distribution inertia may be enough to carry it

OpenAI's execution on ChatGPT is really good and Google is going up against very aggressive fast-moving companies

Scope: concedes distribution inertia might be sufficient

9:34 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Des Traynor · Nov 15, 2023

Google's real question is whether it will risk everything to win — going LLM-first and taking a massive revenue hit — or accept a long slow decline into irrelevance

An LLM-first Q&A future has no worked-out monetization yet, so self-disruption is a massive existential risk that a CEO answering to Wall Street shareholders is unlikely to volunteer for

36:35 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Harry Stebbings · Nov 15, 2023 · hedged

A company the size of Google cannot pivot that radically without tanking shareholder value

It would be like SVB announcing its own insolvency — announcing the disruption destroys the value

38:02 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Des Traynor · Nov 15, 2023

Staying silent about the disruption is not a viable strategy for Google either; it has to stare irrelevance in the face and choose a different path

38:16 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Jeff Seibert · Nov 22, 2023

Google has no choice but to go all in on AI and cannibalize search, because AI replacing search is existential for them

It is far more effective to kill your own golden goose than to watch someone else kill it, as Apple did by actively killing the iPod Nano when it knew better tech was coming

33:41 20VC: Why OpenAI Will Become an Infrastructure Play, Why Apple Will Win in an AI World, Why Google is the Most Vulnerable Incumbent, Will LLMs Be Commoditised, Which Startups Are Thin vs Thick Wrappers on Top of LLMs with Jeff Seibert, Founder @ Digits

Emad Mostaque · May 17, 2023

Google will be one of the main winners in AI

They have the LLMs and the full stack

Scope: a position he says he took back in February

12:25 20VC: Why the AI Bubble Will Be Bigger Than The Dot Com Bubble, Why AI Will Have a Bigger Impact Than COVID, Why No Models Used Today Will Be Used in a Year, Why All Models are Biased and How AI Kills Traditional Media with Emad Mostaque, Founder & CEO @

Harry Stebbings · Feb 24, 2025

The consensus view that OpenAI threatens Google's search golden goose and leaves Google more endangered than ever is a shallow read

47:12 20VC: Why Google Will Win the AI Arms Race & OpenAI Will Not | NVIDIA vs AMD: Who Wins and Why | The Future of Inference vs Training | The Economics of Compute & Why To Win You Must Have Product, Data & Compute with Steeve Morin @ ZML

Harry Stebbings · Sep 15, 2025

Google is massively underestimated

43:52 20VC: Mercor: From $1M to $500M in 17 Months: The Fastest Growing Company in the World | How to Think About Margins and Revenue Sustainability in AI | Why Evaluation Benchmarks in AI are BS Today with Brendan Foody

Harry Stebbings · Oct 17, 2025

Google is a buy

8:53 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Harry Stebbings · Nov 22, 2023

Google's ownership of compute and ability to control compute pricing is a fundamental advantage over OpenAI, which is at the mercy of Nvidia

OpenAI depends on Jensen Huang and Nvidia for its compute while Google controls its own

Scope: accepts the criticism of Google and Bard

28:25 20VC: Why OpenAI Will Become an Infrastructure Play, Why Apple Will Win in an AI World, Why Google is the Most Vulnerable Incumbent, Will LLMs Be Commoditised, Which Startups Are Thin vs Thick Wrappers on Top of LLMs with Jeff Seibert, Founder @ Digits

Mati Staniszewski · Sep 8, 2025 · hedged

Google has a good future and is firmly in the model race, though he would not call himself super bullish

Their recent model innovations like Genie and Veo are incredible, they are catching up in many areas, and the scale of operation is formidable

Scope: explicitly not 'super bullish'; acknowledges questions over the ads model

67:10 20VC: ElevenLabs Hits $200M ARR: The Untold Story of Europe's Fastest Growing AI Startup | The Real Cost of AI from Talent to Data Centres | How US VCs are in a Different League to Europeans | The Future of Foundation Models with Mati Staniszewski

Aravind Srinivas · Jun 15, 2026

Google has every structural advantage needed to be the lowest-cost token producer but is currently behind the frontier

They underestimated the importance of coding models

Scope: they are capable of catching up

55:15 20VC: Micron Will Be More Valuable Than Meta | How Export Controls Helped Not Hurt China | Power is the Bottleneck to AI | Why Dario Has Done a Disservice to AI with his Labour Replacement Messaging with Aravind Srinivas, Founder @ Perplexity

Harry Stebbings · Apr 18, 2025

Google is among the best positioned players in AI — only Microsoft is comparable — because touch points to end consumers are one of the most important assets

Consumer touch points matter most, and nobody other than Microsoft has better ones

19:19 20VC: Foundation Models: Who Wins & Who Loses | How Economies and Labour Markets Need to Change in a World of AI | China vs the US in an AI Race: What You Need to Know | Rich Socher, Founder @ You.com

Zach Lloyd · Oct 17, 2025

Google is not beyond hope: it has enormous distribution, brand, and infrastructure

Distribution, brand, and infrastructure advantages

Scope: stated as the bull case

8:55 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Harry Stebbings · Oct 17, 2025

Google is attractive because Gemini integration can improve over time and it owns YouTube, Waymo, and 10% of SpaceX

The breadth of assets beyond search and Gemini

9:09 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Aidan Gomez · Aug 19, 2024

Google was not technologically or research-wise behind in language models; what it lacked was product vision and the organizational ability to let people with that vision execute internally

Brain started training language models within two weeks of the transformer paper, so the gap was not research capability but the vision of what the technology could do and whether the org enabled execution on it

42:06 20VC: Chips, Models or Applications; Where is the Value in AI | Is Compute the Answer to All Model Performance Questions | Why Open AI Shelved AGI & Is There Any Value in Models with OpenAI Price Dumping with Aidan, Gomez, Co-Founder @ Cohere

Elias Torres · Mar 21, 2025

Google's distribution and data are genuine assets, but on speed they have failed — several years post-GPT they still haven't delivered

The question was about speed, not distribution; years have passed since GPT and nothing works

Scope: concedes Google wins on distribution and data

15:19 20VC: Selling Drift for $1.2BN is the Biggest Failure: What No One Tells You About Selling Your Company | Why Incumbents Are Slower & Worse Than Ever | Why the Most Valuable Companies in a World of AI Will Not Have More Than 100 People with Elias Torres

Nick Frosst · Sep 1, 2025 · hedged

Google failed to commercialize or scale the transformer quickly despite inventing it, and the fact that everyone who worked on it left to keep working on it elsewhere points to something in Google's internal systems

The transformer was published at Google Brain in 2017 but the scaling and commercialization work had to be done elsewhere years later

Scope: says he doesn't know what systems caused this; notes DeepMind still has brilliant people making good products

6:19 20VC: Cohere Founder on How Cohere Compete with OpenAI and Anthropic $BNs | Why Counties Should Fund Their Own Models & the Need for Model Sovereignty | How Sam Altman Has Done a Disservice to AI with Nick Frosst

Harry Stebbings · Mar 19, 2025

Google is one of the most underappreciated companies right now because distribution, not productization or brand, is the durable advantage in AI

Google's consumer distribution endpoints let it deploy AI to enormous existing audiences

Scope: he holds both Microsoft and Google on the same distribution logic

61:55 20VC: The 10 Question Framework a $217BN Manager Uses to Make Investment Decisions | Lessons from Turning Down Stripe, Coinbase and Losing Money on Northvault | The Bull Case for Bytedance | How Anduril Could Be a $200BN Company with Peter Singlehurst

Harry Stebbings · Mar 21, 2025

Google will win in AI because of its distribution and endpoints to billions of consumers, even though the product isn't there today

Access to consumers and distribution is the fundamental thing, and Google has the most endpoints

Scope: concedes the product is not there today

14:58 20VC: Selling Drift for $1.2BN is the Biggest Failure: What No One Tells You About Selling Your Company | Why Incumbents Are Slower & Worse Than Ever | Why the Most Valuable Companies in a World of AI Will Not Have More Than 100 People with Elias Torres

Richard Socher · Apr 18, 2025

Google's default distribution is a lock-in so powerful that degrading the product actually increased revenue, and it remains very hard to dislodge

Most users never change defaults — 80% of iPhone users never change a single setting — which is why Google pays Apple $20B a year; adding ads while organic results decayed through SEO made ads relatively more relevant and more lucrative over a decade of monopoly

Scope: concedes there is some potential opening for challengers

11:43 20VC: Foundation Models: Who Wins & Who Loses | How Economies and Labour Markets Need to Change in a World of AI | China vs the US in an AI Race: What You Need to Know | Rich Socher, Founder @ You.com

Des Traynor · Nov 15, 2023 · speculative

Google could give away Android phones for free, treating hardware cost as customer acquisition cost and monetizing by controlling the intent layer and routing requests to the highest-paying service provider

It's the same model as giving away a free search engine and making it back on sponsored placement — just a different rendering

40:42 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Kieran Flanagan · Jul 11, 2025

Google should go all in on building an ads model integrated into the AI conversation and tailored to the individual, because today's AI ad experiences are poor and its old moat is gone

Google has the history and skill set for personalized advertising, but PageRank — the thing it built its business on — no longer really matters, and there is nothing stopping OpenAI from building an equivalent or better model

Scope: notes shipping fast is hard at Google's scale

43:11 20Growth: The Death of Growth Teams? | How Hubspot Use AI to Triple Email Conversion | The Future of AI SEO | Why Prompt Engineering is the New Coding | What Every CMO Needs to Know About AI in 2025

Des Traynor · Nov 15, 2023 · hedged

Bard felt like a reactive release driven by ChatGPT's traction, and Google needs a reintroduction moment

It felt like they had to release it because ChatGPT was getting a lot of traction

Scope: 'I feel like'

0:00 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Des Traynor · Nov 15, 2023

Google has snatched defeat from the jaws of victory in AI — the problem is not Bard's quality but the absence of authoritative leadership and a stated vision, which Sam Altman and Satya Nadella have provided

Google authored the original transformers paper yet Bard felt like a reactive release to ChatGPT's traction rather than a reinvention of search; leadership means walking on stage and declaring the mission and end goal

Scope: not unimpressed by Bard the product itself

35:17 20VC: How to Survive and Thrive in a World of OpenAI, Are LLMs Being Commoditised, Where Does the Value Lie; Infrastructure or Application Layer, How Apple Could Win in a World of AI, How Amazon Could Threaten OpenAI and Why Google Struggle with Des Trayn

Jonathan Ross · Sep 29, 2025

Gemini has been a success for Google

The adoption numbers have been great

57:23 20VC: OpenAI and Anthropic Will Build Their Own Chips | NVIDIA Will Be Worth $10TRN | How to Solve the Energy Required for AI... Nuclear | Why China is Behind the US in the Race for AGI with Jonathan Ross, Groq Founder

Arvind Jain · Jul 11, 2026

Google has adopted AI better than any other large incumbent, both internally and in shipped products

They rate higher than anyone on embracing AI internally and launching AI products

Scope: notes it may be unfair to classify Google as a legacy company since it is an AI company

47:50 20VC: Why OpenAI and Anthropic Won't Win the App Layer | Why Teams Will Get Bigger Not Smaller in a World of AI | Why AI Removes Incumbents Advantage of Bundling | China vs America: Who Wins the AI War with Arvind Jain, Co-Founder @ Glean

Emad Mostaque · May 17, 2023

Google will continuously improve its models from here because it has regained a shared narrative and psychological safety, and the walls between DeepMind and Brain have come down

People at Google were scared during a period of weird monoculture; now everyone shares the goal of building the best language models and can speak up again

Scope: organizational change takes time to move the ship

12:39 20VC: Why the AI Bubble Will Be Bigger Than The Dot Com Bubble, Why AI Will Have a Bigger Impact Than COVID, Why No Models Used Today Will Be Used in a Year, Why All Models are Biased and How AI Kills Traditional Media with Emad Mostaque, Founder & CEO @

Jonathan Ross · Sep 29, 2025

Google has executed the biggest turnaround among incumbents, and its engineer-driven culture is a systemic advantage

Google depends on engineers rather than management for good ideas, so as long as management gets out of the way great things happen

57:00 20VC: OpenAI and Anthropic Will Build Their Own Chips | NVIDIA Will Be Worth $10TRN | How to Solve the Energy Required for AI... Nuclear | Why China is Behind the US in the Race for AGI with Jonathan Ross, Groq Founder

Tomasz Tunguz · Apr 21, 2023

Google's failure to capitalize on DeepMind stems from the golden-goose problem: leadership teams almost never have the discipline to destabilize a great business model themselves

When you have an incredible business model you must choose between disrupting yourself and waiting for someone else to do it, and self-disruption is extraordinarily hard for a leadership team

40:21 20VC: Who Wins in AI; Startup vs Incumbent, Infrastructure vs Application Layer, Bundled vs Unbundled Providers | From 150 LP Meetings to Closing $230M for Fund I; The Fundraising Process, What Worked, What Didn't and Lessons Learned with Tomasz Tunguz

Alex Lebrun · Jun 19, 2023

Google and Meta missed the LLM wave not because of cost or cannibalisation but because nobody could predict LLMs would be so powerful before training one at scale, and nobody internally had the incentive to spend $500m with no business benefit

The usefulness of LLMs was unforeseeable pre-training; no one in Google's org chart was incentivised to fund it, and legal departments were unhappy about releasing chatbots that could say anything — whereas OpenAI as a private company needed to show something and had curiosity plus financial power

20:49 20VC: Why No Models Today Will Be Used in a Year, Why Open Will Always Beat Closed in AI, Why Proprietary Data is Less Important Than Ever And Why EU AI Regulation is a Disaster with Alex Lebrun, Founder & CEO @ Nabla

Zach Lloyd · Oct 17, 2025

Google is a very risk-averse place that retains smart but risk-avoidant people because compensation makes it cushy to stay

The respected people he knows stay because they're extremely well compensated, which selects for people who want to take fewer risks

7:56 Cushy compensation breeds risk averse culture

Brendan Foody · Sep 15, 2025

DeepMind's smaller Gemini Flash models are the most underappreciated models, and are extraordinary on evals

OpenAI gets the consumer brand attention, Anthropic the code attention, and xAI consumer attention, leaving the small Gemini models overlooked

43:18 Gemini flash models are the most underappreciated in the market

Steeve Morin · Feb 24, 2025

Winning in AI requires holding all three of product, data and compute, and Google is the only player with all three, which makes it the sleeping giant

Google owns its own compute and can sprinkle AI across Android, Google Docs and its other products; Amazon has AWS but no real consumer products; OpenAI's compute is Microsoft's, and if you don't own your compute you don't own your margin

46:01 Owning product data and compute together makes google the sleeping giant

Aaron Levie · May 22, 2024

Google has now got religion on AI and Google I/O was evidence of that top-down prioritization working

Sundar set a message that AI is the number one priority, ten times above everything else — the new way to search, new cloud products, in Workspace — and Cloud Next and I/O are the manifestations of that message paying off

33:00 Googles top down ai prioritization is now working as shown by io

Emad Mostaque · May 17, 2023

The DeepMind–Brain merger does create real friction and duplicated roles and cannot be integrated instantly, but the fusion of ideas is already visible in the models

PaLM's 540B-parameter scale plus DeepMind's Chinchilla insight (train more rather than add parameters) combine in PaLM 2, which achieves comparable results at a fraction of the size on far better data

13:40 Merger friction real but technical fusion of ideas already visible in models

Jonathan Ross · Sep 29, 2025

Gemini's integration into Google's consumer products is half-thought-through and often unusable, but it shouldn't be judged yet because the exposure teaches Google what to build

Getting real usage data lets them iterate — Google TV flopped and was iterated into Google Chrome; companies willingly take the darts to build a better product

57:31 Too early to judge messy integration exposure teaches iteration

Jonathan Ross · Sep 29, 2025 · hedged

Google may be too late, given OpenAI has already closed the distribution gap

58:20 Openais distribution lead may leave google too late

Harry Stebbings · Apr 21, 2023 · hedged

Google's inaction is explained by the fact that the move from search to chat still isn't obvious, unlike Netflix's move from mail-order DVDs to streaming which was obviously right despite short-term revenue loss

Netflix could see the shift was inevitable; search-to-chat is potential but not obvious, so self-disruption was harder to justify

40:39 Search to chat shift isnt obvious unlike netflixs streaming pivot explaining googles inaction