David Sacks: AI, Crypto, China, Dems, and SF

Europe treats AI leadership as writing the rulebook in Brussels, then calling that victory.

It can feel like a game show: choke startups early, and if any survive years of punishment, throw money at them to scale.

It’s that old pattern—tax it, regulate it, and when it stalls, subsidize it—which is where Europe is now.

AI and crypto now sit at the heart of the global tech and economic race; today David Sacks joins Marc and Ben to unpack the Trump administration’s strategy, from open source to avoiding overregulation—let’s get into it.

Why bundle AI and crypto, and what’s the plan for each under Trump?

They’re new and misunderstood, but the fixes diverge: crypto needs clear rules after years of enforcement-by-ambush that pushed builders offshore, while AI needs room to run after heavy-handed moves; our policy is to win the AI race by being pro-innovation, pro-infrastructure, pro-energy, and pro-exports, and my job is to bridge Silicon Valley and Washington so the culture of building is protected.

Did people really grasp how rough it got for crypto founders?

Founders and even their personal bank accounts were debanked while the SEC prosecuted without clear rules; hosting a White House crypto summit was a turning point and we’ve shut that down.

In AI, some incumbents are pushing rules to lock in their leads; where does that go?

It’s a serious problem—Anthropic’s policy rhetoric and Q&A made plain that stoking fear to get pre-approvals is the goal, which kills permissionless startups and hands the field to giants; we reversed a late Biden rule that effectively required a license for every GPU sale on Earth, and we’re pushing back on software schemes that morph reporting into pre-release permission.

The same firms warning of monsters hoard GPUs and run sloppy security, which tells you what they really believe.

Apocalypse talk also flatters them as noble gatekeepers, but the bigger danger is a flood of state laws like algorithmic discrimination that make toolmakers liable and force ideological filters, drifting toward an Orwellian AI that censors, surveils, and rewrites facts.

Where are we on AGI and claims like automated researchers by 2028?

The imminent-AGI drumbeat has cooled; progress is fast but uneven, and the healthy view is many specialized systems that amplify people, with AI doing the middle steps and humans setting goals and validating work.

We’re seeing specialization win—no single model dominates, and intelligence isn’t life.

What’s striking is how quickly AI has democratized—my wife spun up an entrepreneurship curriculum for our ten-year-old in a morning—which shows this is empowering individuals, not just a few firms.

Decentralization and fierce competition are healthy, and no one has run away with it; I’m skeptical a true virtual AI researcher arrives before AGI, not the other way around.

Those timelines often sound like recruiting pitches, not forecasts.

Others are selling the same idea, but the sequence still looks backwards to me.

You’ve also backed open source; what have you done and why?

Open source is software freedom—run models on your own hardware and keep control—and while the strongest open models today are Chinese, we want more Western efforts; open choices also check future consolidation and blunt any government-corporate squeeze we saw in social media.

We’re funding a wave of model companies, including new open projects, so more is coming.

That’s good, because open source is the one area where China currently leads; otherwise, the U.S. is ahead.

Walk us through the China race, chips, and exports—what matters most?

Winning comes from unleashing builders, not micromanaging them, which is why a single federal standard should preempt the fifty-state thicket; we also need permits and power for data centers, and we should export to allies to grow our ecosystem instead of pushing them to Huawei by blocking access to American tech.

What’s it take on energy and infrastructure to meet demand?

Nuclear will help later, but near term it’s gas while we clear turbine backlogs, and we can unlock a lot of grid capacity by shifting a few dozen peak hours to backup if we fix rules that block smart load-shedding.

America is basically doing the opposite of the EU’s regulate-first model.

In Europe, writing rules is the brand of leadership, and when growth stalls they subsidize.

It’s good to see the U.S. re-center on core American values.

Our goal is to lead by letting builders build, not by drowning them in paperwork.

You said climate doom is giving way to AI doom; what did you mean?

The left swapped one crisis narrative for another to justify economic and speech controls, and effective altruists drove an x-risk agenda inside the last administration—talking bans on open source and picking a few winners—yet DeepSeek and Huawei’s systems exposed the folly of self-handicapping while rivals surge, and scary compute thresholds we were told to fear have now been crossed without catastrophe.

Officials told us outright they’d ban open algorithms, even invoking Cold War physics as precedent.

And the person who said that now works at the company pushing those policies.

They framed AI like nuclear and wanted an IAEA-style regime, but facts on the ground have undercut that case.

Back to crypto—stablecoin law is working; what’s the importance of the Clarity Act?

It sets durable rules for the rest of the tokens so founders have long-term certainty beyond a friendly SEC chair; the House passed it with broad bipartisan support, we’re working toward sixty in the Senate, and Trump’s win and direct engagement made the stablecoin law possible and set the tone for making America the crypto capital.

Where do you see the Democrats heading?

Their energy seems tied to woke socialism and soft-on-crime ideas that fail in practice, which puts them on the wrong side of common-sense issues and raises the stakes in places they control.

Is San Francisco salvageable, and what about the National Guard?

We have the best mayor in decades, but a weak-mayor system, a powerful board, and lenient judges make it hard to tackle repeat offenders, so the president is holding off while the city tries its plan.

Thank you, David—great discussion, and we appreciate the work to fix what was broken and set up a better path.

Thanks to you as well—I appreciate the support and everything you’re doing.

Thanks for listening to the a16z Podcast; follow, rate, and share, and remember this is for information only and not legal, business, tax, or investment advice—see our site for disclosures.