Video · 2026-07-03 · 42m · 36 moments

Ben Horowitz on the Global Race for Tech, Power, and Influence

✦ AI generated

GUEST Ben Horowitz GUEST Ann GUEST Ragu GUEST Raghu

timeline · colored by role

Claim

America is uniquely the best place in the world for a person with just a good idea for making the world better — even without existing culture, money, or capital resources — to succeed, and a core mission of the firm is preserving that special quality by helping America remain a leader economically, culturally, militarily, and technologically.

Ben Horowitz frames the firm's global mission as rooted in preserving America's unique openness to anyone with a good idea, which requires America to lead economically, culturally, militarily, and technologically.

transcript

Ben Horowitz: if you're drop a person into any country in the world and they don't have the culture and they don't have uh money and capital resources yet, and they just have an idea of how to make the world a better place, um their best shot is in America.

explains mechanism · 1

Claim

Deterrence between nations is no longer just about the size of a military — it's about the pace of innovation in reacting to cheap, software-built technology that adversaries, whether countries, criminals, or terrorist groups, may be using.

Ann explains that national deterrence has shifted from a function of military scale to a function of how fast a country can innovate and respond to cheap, software-driven threats deployed by state and non-state actors alike.

transcript

Ann: deterrence is no longer just the size of a military. I think as we see in the straight of Hermuz or we saw in the Red Sea, it's the pace of innovation in reacting to cheap softwarebuilt tech that adversaries may be using whether they're a country, criminal or a terrorist group.

extends · 1

Claim◆

Deterrence between nations is no longer just about the size of a military — it is about the pace of innovation in reacting to cheap, software-built technology that adversaries use, and those decisive technologies are being built by the private sector, not governments.

Ann describes how deterrence has shifted from raw military size to who innovates fastest with cheap software-based technology, and notes that the private sector, not governments, is building those decisive tools.

transcript

Ann: you know deterrence is no longer just the size of a military. I think as we see in the straight of Hermuz or we saw in the Red Sea, it's the pace of innovation in reacting to cheap softwarebuilt tech that adversaries may be using whether they're a country, criminal or a terrorist group.

Claim◆

Deterrence between nations is no longer just about the size of a military — it's about the pace of innovation in responding to cheap, software-built technology that adversaries (state, criminal, or terrorist) are using, and that technology is now built by the private sector rather than governments.

Ann describes a fundamental shift she witnessed in government: deterrence and national power now hinge on private-sector-built technology and innovation speed rather than traditional military scale.

transcript

Ann: You know deterrence is no longer just the size of a military. I think as we see in the straight of Hermuz or we saw in the Red Sea, it's the pace of innovation in reacting to cheap softwarebuilt tech that adversaries may be using whether they're a country, criminal or a terrorist group. And those technologies today are not being built by governments. They're being built by the private sector.

Claim◆

Deterrence today is no longer just the size of a military—it's the pace of innovation in reacting to cheap, software-built technology that adversaries (countries, criminals, or terrorist groups) may be using, and those technologies are being built by the private sector, not governments.

Ann argues that national deterrence has shifted from being about military scale to being about how fast a country and its private sector can innovate against cheap, software-driven threats.

transcript

Claim◆

Deterrence between nations is no longer determined mainly by the size of a military but by the pace of innovation in responding to cheap, software-built technologies that adversaries — whether states, criminals, or terrorist groups — deploy, and those technologies are now built primarily by the private sector rather than governments.

Ann describes a shift she witnessed from the intelligence community into the White House: technology moved from being a diplomatic tool to the central arena of national power, redefining deterrence around innovation speed rather than military scale.

transcript

rebuts · 1

Claim

Deterrence between nations is no longer just a function of military size — it now depends on the pace of innovation in responding to cheap, software-built technologies wielded by adversaries, and those technologies are built by the private sector, not governments.

Ann describes how her time on the intelligence committee and in the White House showed her that tech shifted from a diplomatic tool to the central arena of national power, redefining deterrence around innovation speed rather than military mass.

transcript

Claim◆

AI models are not objective—they carry strong opinions on history, culture, ethics, and values that differ depending on where they were built, so it matters whether the world's technology is built on models reflecting American/Western values or some other country's values.

Ben Horowitz argues that because AI will underpin every future tool and system, and because models embed the values and worldview of their creators, it matters greatly whose values—American or otherwise—get projected globally through them.

transcript

Ben Horowitz: every tool we have every technology we use every kind of system we have from your car to your refrigerator to your education system is going to be built on these models and the models um are not objective. They have opinions um and they they have strong opinions on things like history, on culture, on um you know, ethics, on values. And by the way, those are very different depending where you are.

Claim◆

AI models are not objective — they carry strong opinions on history, culture, ethics, and values that differ depending on where they are built, so it matters whether the world's foundational models are built on American/western values or someone else's.

Ben Horowitz argues AI models embed the values of their creators rather than being neutral, making it critical that the world's dominant models reflect American/western values rather than authoritarian ones.

transcript

Ben Horowitz: well you know when AI becomes the interface for everything um so every every tool we have every technology we use every kind of system we have from your car to your refrigerator to your education system is going to be built on these models and the models um are not objective. They have opinions um and they they have strong opinions on things like history, on culture, on um you know, ethics, on values.

provides context · 1

Claim◆

AI models are not objective — they carry strong opinions on history, culture, ethics, and values that vary depending on where they're built, and since AI is becoming the interface for every tool and system (cars, refrigerators, education), whose values get embedded in the models matters enormously.

Ben Horowitz argues AI models are inherently value-laden, not neutral, and since they're becoming the interface for every technology and system, it matters whose values — American/Western or otherwise — get built into them.

transcript

rebuts · 1

Claim◆

AI models are not objective — they carry strong opinions on history, culture, and values, and those opinions differ depending on where the model comes from, which is why it matters whose values get built into and projected through the models everything else is built on.

Ben Horowitz argues AI models are inherently value-laden rather than objective, and since every future tool and system will be built on them, whose values they encode (American/Western vs. others) is a matter of real consequence.

transcript

Claim◆

AI models are not objective — they embed strong opinions on history, culture, ethics, and values, and those values differ depending on where the model is built, so it matters whose values get projected onto the world through AI.

Ben Horowitz argues AI models embed the cultural and political values of their creators, making it critical that American/Western values be the ones projected globally as AI becomes the interface for everything.

transcript

provides context · 1

Claim◆

AI models are not objective — they embed strong opinions on history, culture, and values, and those opinions differ depending on where and by whom the model was built, so as AI becomes the interface for everything, whoever's values are baked into the dominant models effectively projects those values onto the world.

Ben Horowitz argues AI models carry embedded cultural and political viewpoints, not objective neutrality, making the question of whose values are baked into the dominant models a matter of global power and influence.

transcript

Ben Horowitz: every technology we use every kind of system we have from your car to your refrigerator to your education system is going to be built on these models and the models um are not objective. They have opinions um and they they have strong opinions on things like history, on culture, on um you know, ethics, on values. And by the way, those are very different depending where you are.

Mechanism

Because great products and technology now spread instantly worldwide via the internet and APIs, startups must go international much sooner in their lifecycle than companies did in the past, when firms waited until they had hundreds of millions in revenue.

Ragu explains that products now travel faster than companies can, forcing startups to internationalize much earlier than in prior eras, even though the underlying work of building regional trust and relationships hasn't gotten any faster.

transcript

Ragu: the reality is that products and technology travel faster than companies can... so what happens is for startup companies, usually in the olden days, you didn't care about going international until you had a few hundred million dollars in revenue etc. Now these companies have to go sooner and sooner internationally.

gives example · 1

Mechanism

Because great products now spread globally almost instantaneously via APIs, startups can no longer wait until they have hundreds of millions in revenue to expand internationally — they must go international much sooner in their lifecycle than in the past.

Raghu explains that because technology and product awareness now spread instantly worldwide via APIs, startups face pressure to internationalize far earlier than the old model of waiting for hundreds of millions in revenue.

transcript

Raghu: because if you got a great product everybody learns about it very quickly instantaneously you can access it through APIs wherever in the world you are you can build on top of those things so what happens is for startup companies usually in the olden days you didn't care about going international until you had a few hundred million dollars in revenue etc now these companies have to go sooner and sooner sooner internationally.

Mechanism

Because products and technology now spread virally and can be accessed instantly via APIs anywhere in the world, startups no longer wait until they have hundreds of millions in revenue to expand internationally — they must go global much earlier in their life cycle than companies did in the past.

Ragu explains that because technology and products spread instantly and globally, portfolio companies are internationalizing far earlier in their life cycle than startups historically did.

transcript

Ragu: the reality is that products and technology travel faster than companies can right because if you got a great product everybody learns about it very quickly instantaneously you can access it through APIs wherever in the world you are you can build on top of those things so what happens is for startup companies usually in the olden days you didn't care about going international until you had a few hundred million dollars in revenue etc now these companies have to go sooner and sooner internationally.

Claim

Products and technology now spread across borders faster than companies can organizationally follow, forcing startups to go international much earlier in their lifecycle than they used to, rather than waiting until they have hundreds of millions in revenue.

Ragu explains that because great products spread globally almost instantly via APIs and word of mouth, startups today must pursue international expansion far sooner than companies did in the past.

transcript

Ragu: the reality is that products and technology travel faster than companies can. Because if you got a great product everybody learns about it very quickly, instantaneously you can access it through APIs wherever in the world you are, you can build on top of those things. So what happens is for startup companies, usually in the olden days, you didn't care about going international until you had a few hundred million dollars in revenue, etc. Now these companies have to go sooner and sooner internationally.

gives example · 1

Claim

Because a great product spreads globally almost instantly through APIs and software distribution, startups now must pursue international expansion much earlier than in the past, when companies waited until they had hundreds of millions in revenue.

Ragu explains that products and technology now travel faster than the companies behind them, forcing startups to go international far sooner in their lifecycle than in prior eras.

transcript

Ragu: products and technology travel faster than companies can right because if you got a great product everybody learns about it very quickly instantaneously you can access it through APIs wherever in the world you are you can build on top of those things so what happens is for startup companies usually in the olden days you didn't care about going international until you had a few hundred million dollars in revenue etc now these companies have to go sooner and sooner internationally.

Mechanism

Because products and technology now travel internationally faster than companies themselves can expand, startups are forced to go international much earlier in their life cycle than they used to, instead of waiting until they had hundreds of millions in revenue.

Ragu explains that instantaneous global access to products via APIs and software means startups can no longer delay international expansion until they're large, unlike in the past.

transcript

Ragu: the reality is that products and technology travel faster than companies can because if you got a great product everybody learns about it very quickly instantaneously you can access it through APIs wherever in the world you are you can build on top of those things so what happens is for startup companies usually in the olden days you didn't care about going international until you had a few hundred million dollars in revenue now these companies have to go sooner and sooner internationally.

gives example · 2

Example

By pairing portfolio company ElevenLabs with Mexican media company Televisa Univision to solve accent and language localization, a16z helped a culturally local Spanish-language product become globally distributable, leading to a major Netflix deal.

Ben Horowitz gives a concrete example of a16z connecting a Mexican media company with ElevenLabs' voice AI to dub content authentically across Spanish dialects, unlocking global expansion and a Netflix deal.

transcript

Ben Horowitz: we've got this great company, 11 Labs. They can help you with it. We put them together in a couple of months. He was in, you know, France and Poland and all over the world with this incredible technology that had his actors basically acting exactly as they normally do with their right kind of intonations, their right laugh, everything but in different languages.

Example

AI voice technology from ElevenLabs let Televisa Univision authentically dub its Spanish-language content and actors into other languages and accents, letting culturally-local content go global within months and leading to a major Netflix deal.

Ben Horowitz recounts how a16z connected Televisa Univision with ElevenLabs to dub its Spanish-language shows into other languages with the actors' authentic intonations, quickly opening international markets and a Netflix deal.

transcript

Ben Horowitz: we've got this great company, 11 Labs. They can help you with it. We put them together in a couple of months. He was in, you know, France and Poland and all over the world with uh this incredible technology that had his actors basically acting exactly as they normal do with their right kind of intonations, their right laugh, everything but in different languages. And he recently um based on that success got a big deal with Netflix.

Anecdote

By pairing Televisa Univision with a16z portfolio company ElevenLabs, a Mexican Spanish-language media company was able to dub its content into multiple languages while preserving actors' original intonation and delivery, letting it go global within months and subsequently land a major deal with Netflix.

Ben Horowitz gives a concrete anecdote of a16z's value-add model: connecting Televisa Univision with ElevenLabs let the company's Spanish-language content go global in months, preserving actors' authentic performance, and led to a Netflix deal.

transcript

Ben Horowitz: We put them together in a couple of months. He was in, you know, France and Poland and all over the world with uh this incredible technology that had his actors basically acting exactly as they normal do with their right kind of intonations, their right laugh, everything but in different languages. And he recently um based on that success got a big deal with Netflix.

Anecdote

Pairing Televisa Univision, a culturally-local but language-constrained media company, with the AI dubbing startup ElevenLabs let its Spanish-language content be redistributed globally in different languages and accents within a couple of months, culminating in a major deal with Netflix.

Ben recounts how a16z connected Televisa Univision's CEO with ElevenLabs to dub his Spanish-language content into other languages while preserving actors' authentic performances, quickly expanding his global reach and leading to a Netflix deal.

transcript

Ben Horowitz: we've got this great company, 11 Labs. They can help you with it. We put them together in a couple of months. He was in France and Poland and all over the world with this incredible technology that had his actors basically acting exactly as they normally do with their right kind of intonations, their right laugh, everything but in different languages. And he recently based on that success got a big deal with Netflix.

Mechanism

AI is fundamentally dual-use in cybersecurity: the same capability that lets defenders find and patch vulnerabilities also lets attackers find and exploit them, and models make the step of finding vulnerabilities in code far easier for both sides.

Ann explains that AI models cut both ways in cybersecurity, making vulnerability discovery easier for attackers and defenders alike, though she believes defense benefits more since attackers only need one way in while defenders must cover everything.

transcript

Ann: fundamentally as many technologies are AI is dual use right so the first step whether you are an attacker or a defender is finding vulnerabilities in systems and then if you're a defender you patch them if you're an attacker you exploit them so models make the step of finding vulnerabilities in code far easier.

Mechanism

AI is fundamentally dual-use in cybersecurity: the same capability that lets a defender find and patch vulnerabilities in code lets an attacker find and exploit them, since finding vulnerabilities is the identical first step for both attacker and defender.

Ann explains the core dual-use dilemma of AI in cybersecurity: models make finding vulnerabilities easier for everyone, and the first step is identical whether the goal is to patch (defense) or exploit (attack).

transcript

Ann: fundamentally as many technologies are AI is dual use right so the first step let's use cyber as an example right the first step whether you are an attacker or a defender is finding vulnerabilities in systems and then if you're a defender you patch them if you're an attacker you exploit them

explains mechanism · 1rebuts · 1

Mechanism

AI is inherently dual-use in cybersecurity: the same models that let attackers find vulnerabilities also let defenders find and patch them first, and because defense means protecting a broad expanse while attackers only need one way in, models ultimately help defenders more even though the near-term attack surface looks daunting.

Ann explains that AI's dual-use nature in cybersecurity means models help both attackers and defenders find code vulnerabilities, but she believes defense benefits more since attackers only need one opening while defenders must cover everything.

transcript

extends · 1supports · 1

Mechanism

AI models help cyber defenders more than attackers because defense means protecting a broad expanse while an attacker only needs to find one way in, and models let organizations find and patch their own vulnerabilities before adversaries exploit them.

Ann argues that although AI is dual-use in cybersecurity, it ultimately favors defenders more because defense is structurally harder, and models now make it economically feasible to find and fix vulnerabilities before attackers exploit them.

transcript

Ann: models make the step of finding vulnerabilities in code far easier... I personally believe... they help far more in defense because defense is far harder. You have to be defending a broad expanse and attacker has to find one way in.

rebuts · 2extends · 1supports · 2

Mechanism

AI models help both attackers and defenders find software vulnerabilities, but defenders benefit more because defense requires covering a broad attack surface while an attacker only needs to find one way in, meaning AI could ultimately push the world toward much more secure systems.

Ann argues that although AI is dual-use in cybersecurity, it disproportionately helps defenders because defense is inherently harder than attack, pointing toward a future of more secure systems despite today's dangerous transition period.

transcript

Ann: they help far more in defense because defense is far harder. You have to be you know defending a broad expanse and attacker has to find one way in. Um so as a result I actually think we are in a difficult space at this moment because there's such a broad attack surface there's so many systems that are vulnerable but I think we will see the opportunity for much more secure space.

extends · 1rebuts · 4supports · 1

Mechanism

Governments cannot simply ban AI models from having cyber hacking capabilities, because the same capability that lets a defender find and patch vulnerabilities in its own code is inherently the capability to find and exploit vulnerabilities elsewhere — you can't have one without enabling the other.

Ben Horowitz points out the dual-use trap in AI cybersecurity policy: restricting models' ability to find vulnerabilities for defense also removes the ability to defend, since the underlying capability is identical to what attackers need.

transcript

Ben Horowitz: you'd like to say okay don't release your model with cyber hacking capabilities. But if you do, as an say and you want to find the vulnerabilities in your own code and patch them, you can't do that without also enabling cyber attacks because as soon as it can find the vulnerabilities, guess what?

Context

Cyber threats have evolved through distinct eras — from vandalism, to theft, to war — and much of today's technology infrastructure, including systems Microsoft built by putting Windows on the internet before it was ready, was designed for the pre-AI era and isn't secure enough for the AI world.

Ben Horowitz traces cyber threats evolving from vandalism to theft to war, using Microsoft's early internet-era Windows as an example of infrastructure built before it was ready — and argues today's legacy tech is similarly unprepared for the AI era.

transcript

Ben Horowitz: originally cyber hacking was vandalism, right? Like, okay, I'm going to mess up your PC. Then, you know, with the internet, it became theft and now it's war. And the great thing when it went from vandalism to theft, the vandalism company, Microsoft, is one who created all the theft opportunities by putting Windows on the internet when it wasn't ready to be on the internet.

provides context · 1supports · 1

Claim

Silicon Valley cannot simply be replicated elsewhere just because the internet lets products and technology reach anywhere — countries differ too much in their laws, policies, and governance, which is why America and its Silicon Valley ecosystem remain uniquely special rather than being effectively duplicated worldwide.

Ben Horowitz rejects the idea that the internet has already made Silicon Valley 'global' by default, arguing that deep differences in national laws, policy, and governance still prevent other countries from replicating it.

transcript

Ben Horowitz: I don't think that's true, by the way. And like they think that's a happy talk. If you you know like one thing you learn when you go to different countries is they're very different. Like the laws are different, the policies are different, the governance are different. And that's why you know America has continued to be special.

Claim

Recreating Silicon Valley elsewhere requires three ingredients—technical talent, pro-entrepreneurship laws and policies (including protection of ownership of illiquid assets), and a culture that confers social status on founders—not merely internet access, and it's easy to destroy this combination once built.

Ben lays out talent, favorable entrepreneurship policy, and a status-rewarding culture as the three necessary ingredients for a Silicon Valley-like ecosystem, rejecting the idea that internet access alone can replicate it, and warns that this combination is fragile and easily destroyed by bad policy.

transcript

Ben Horowitz: The first is you do need the talent... The second is do you have a set of laws and policies that facilitate entrepreneurship, are essentially good for business? Can you own an illiquid asset that has value or will it get seized?... And then the third one is the culture such that young people, the most capable young people willing to make the biggest contribution, get status, social status reward—is it cool to either start or join a new company.

Mechanism

A second essential ingredient for a Silicon-Valley-like ecosystem is having laws and policies that genuinely facilitate entrepreneurship: easy company formation, easy hiring and firing, non-onerous tax structures, and secure ownership of illiquid assets rather than exposure to seizure via things like unrealized capital gains taxes.

Ben Horowitz lays out the second of three ingredients needed to replicate Silicon Valley: business-friendly laws and policies, including tax treatment and secure property rights, without which people won't take entrepreneurial risk.

transcript

Ben Horowitz: do you have a set of laws and policies that uh facilitate entrepreneurship are essentially good for business? Like can you create a company easily? Can you hire people easily? Um can you fire people easily? Is there a tax incentive structure that's not ownorous? Can you um own and this is a big one lately which uh you know California is kind of moving against but can you own an illquid asset that has value um or will it get seized

Claim◆

Recreating Silicon Valley elsewhere requires a culture where the most capable young people get real social status and reward for starting or joining a new company — and almost no country besides the US (and, to some extent, China) has that cultural ingredient, even when they have talent and decent laws.

Ben Horowitz argues that beyond talent and business-friendly laws, the missing ingredient in most countries trying to build their own Silicon Valley is a culture that rewards and confers status on entrepreneurial risk-taking.

transcript

Ben Horowitz: is is the culture such that young people the most capable young people willing to make the biggest contribution is do they get status social status reward is it cool to either start or join a new company um and almost no very few countries have that uh I would say that it's really really good in the US.

Claim◆

Recreating Silicon Valley elsewhere requires three ingredients together — technical talent, business-friendly laws and policies (including secure property rights and no punitive taxes like unrealized capital gains taxes), and a culture that confers social status on entrepreneurship — not just internet access, since many countries have talent but still lack a Silicon Valley.

Ben Horowitz lays out why 'the internet is just distributed Silicon Valley' is false, arguing that talent, pro-business laws/property rights, and a status-rewarding entrepreneurial culture must all combine, and that culture is the hardest and rarest ingredient.

transcript

Claim◆

The hardest ingredient to replicate in building a Silicon-Valley-like ecosystem is not talent or favorable laws but a culture that grants social status and prestige to people who start or join new companies — something almost no country besides the US, and to some degree China, has.

Ben Horowitz argues that beyond talent and good policy, the decisive and hardest-to-copy ingredient of Silicon Valley is a culture that confers social status on entrepreneurs, which is rare, fragile, and easy to destroy.

transcript

Ben Horowitz: And then the third one which is I think kind of a combination of everything in a way is is the culture such that young people the most capable young people willing to make the biggest contribution is do they get status social status reward is it cool to either start or join a new company um and almost no very few countries have that uh I would say that it's really really good in the US.

supports · 1

Highlight slides

Related episodes

Audio · All-In Podcast

SpaceX-Cursor Deal, SaaS Debt Bomb, New Apple CEO, SPLC Indictment, Colon Cancer Spike

→

Video · Lenny's Podcast

The rise of taste, human authenticity and judgment in an AI world | Adam Mosseri (Head of IG)

→

Audio · All-In Podcast

OpenAI Misses Targets, Codex vs Claude, Elon vs Sam Trial, Big Hyperscaler Beats, Peptide Craze

→