I’m joining Sound Ventures to invest in the tech industry I’ve been covering for the last decade.

Ten years ago, I set out to have a front-row seat to the biggest stories in tech. I’ve sometimes played a part in them through my reporting. Along the way, I’ve gotten to know many of the most influential founders shaping the future. Now I want to back them.

I first met Guy Oseary about a year ago through a mutual friend. Over the months that followed, I also got to know Effie Epstein, who leads Sound with Guy in Los Angeles, where I’m also based, and the rest of the team. Guy and I initially discussed my serving as an advisor to Sound. As he described what he and Effie wanted to build for the firm’s next phase, I found myself asking, “What if I went all in?”

As AI makes software easier and cheaper to build, more companies will compete for the same slice of finite attention. Having a great product is increasingly just the starting point. Knowing what you stand for, how to tell your story, and how to reach people will matter even more. That’s central to what Guy and Effie are building at Sound, and it’s a key part of why I’m joining as a partner on the investing team.

Guy and Effie also understand why it’s important that I keep building Sources. I’ll continue to own and run it independently of my work at Sound. The newsletter is continuing, and I’m investing more in the Sources podcast, which I just launched with back-to-back conversations with Sam Altman and Mark Zuckerberg. I have an incredible roster of guests lined up for the coming months. Those conversations will continue to reflect my curiosity across tech, with founders in and beyond Sound’s portfolio.

What drew me to Sound starts with the firm's approach to talent. Guy’s background is in finding generational music artists and helping them reach an audience. He brings that same approach to founders: recognizing something exceptional, then helping it connect with the world. Sound’s track record made the opportunity even more compelling. The firm backed OpenAI, Anthropic, and Hugging Face in 2023, and its record spans dozens of exits over the last decade. We’re a non-traditional firm that’s structured well to collaborate with other investors and support the best companies as they scale.

Venture capital has interested me for a long time. I’ve reported on funding rounds, watched startups become big companies over the years, and learned a lot from founders and investors who were generous with their time. Much of what I’ve enjoyed about journalism translates to the VC world: spotting a trend early, connecting the dots that others haven’t, and figuring out which founders have something worth paying attention to.

That’s the perspective you’ll hear from me going forward. I’ll keep writing about what interests me, sharing what I think, and having the candid, insightful conversations with tech leaders that you’ve come to expect. The world of tech media is as fast-moving and relevant as ever, and I’m excited to continue to play my part from my new vantage point at Sound.

Speaking of which, it’s almost the one-year anniversary of Sources, which has far exceeded my expectations and become a healthy, growing business. To everyone who has read, subscribed to, sponsored, listened to, or shared my work: thank you. I’m excited to keep going.

I got into covering Silicon Valley because I’m insatiably curious about what technology makes possible. It’s time to follow my curiosity again.

A MESSAGE FROM MY SPONSOR

Moving America Forward

Constellation is America’s largest producer of reliable, clean energy, driving progress and innovation coast to coast. From local neighborhoods to iconic cities, we bring the energy that serves industries, strengthens communities, and moves America forward.

Learn More

Lately on the Sources podcast

Sources is a newsletter by Alex Heath about the AI race and the people shaping it. Every week, Sources is read by thousands of decision-makers in tech, finance, policy, and media.

Subscribe now

When Mark Zuckerberg laid out his “personal superintelligence” vision over a year ago, he was thinking about what Meta released today: Muse, an AI agent with its own computer that works for you around the clock. His ambition, he tells me, is to “give every person in the world a very capable personal agent that can understand their goals and can just work on their behalf 24/7.”

I sat down with Zuckerberg at Meta headquarters last Friday, and I’ve been trying Muse for the last several days. I’m impressed. It’s essentially OpenClaw for normies, with a serious focus on security that I think will make this kind of product accessible to a lot more people. You can also connect Muse to your Instagram and Facebook accounts, including the tools for running a business and buying ads.

Zuckerberg’s ambition is to build something he can hand to anyone in his family, regardless of their technical ability. “I don’t think there are going to be billions of people who are going to buy a Mac Studio and configure it and run it at their home,” he tells me.

Available now everywhere you get podcasts.

Getting people comfortable giving an agent access to their lives is another challenge. Zuckerberg tells me he personally helped recruit the founder of Signal to work on a confidential virtual machine designed to keep even Meta from seeing what Muse is doing on a user's behalf. That’s an important promise from a company asking people to connect their email, messages, and other personal information to an AI that can act on their behalf. We get into how Meta is approaching that problem and why Zuckerberg sees privacy as a reason people will choose Muse.

He’s also making it very hard to compete on price. Muse launches with up to 100 million tokens per week for free, enough to give people plenty of room to put an agent to work. Zuckerberg thinks Muse will eventually pay for itself by helping people earn and save money, with Meta taking a small cut of transactions and offering a paid subscription for power users. We discuss how he uses Muse himself, from organizing baking projects with his daughter to critiquing his MMA training, and what happens when Muse agents start interacting with one another.

We also get into the backlash against data centers, why he thinks speculative developers are making it worse, Meta’s youth safety settlement, and the privacy concerns surrounding its smart glasses. He explains what went wrong with Llama 4, how he rebuilt Meta’s AI lab around him, and why he disagrees with rivals keeping their most powerful models to themselves.

A MESSAGE FROM MY SPONSOR

Reliable Clean Energy

Constellation is America’s largest producer of reliable, clean energy, driving progress and innovation coast to coast. From local neighborhoods to iconic cities, we bring the energy that serves industries, strengthens communities, and moves America forward.

Learn More

Episode highlights

• Zuckerberg is giving away a lot of compute to win the personal agent market. Muse launches with up to 100 million tokens a week for free, plus a virtual machine to work on your behalf. With startups like Instinct and Town chasing the same market, Meta is setting an aggressive expectation for how much a personal assistant should do before asking you to pay. Those startups don’t have Meta’s existing business to subsidize that usage. Zuckerberg’s plan is to eventually “take a very small cut of whatever the transaction is,” potentially paid by the businesses users transact with. “We think that this thing is actually going to make you money and save you money, and that is how it’s going to pay for itself.”

• Zuckerberg personally helped recruit Signal’s founder to make Muse’s virtual computer private. He says he and Nat Friedman recruited Moxie Marlinspike to work on the confidential virtual machine behind Muse. The ambition is to give people a computer in the cloud with the confidentiality they’d expect from a machine under their own desk: “Even Meta cannot see the content that is in there.” Zuckerberg says he isn’t aware of anyone offering anything close to that system. He also describes separate “sentinel” agents that monitor what Muse receives and sends, flag possible prompt injections, and require user approval for sensitive actions.

• He thinks speculative data center developers are helping fuel the backlash against AI infrastructure. Zuckerberg draws a distinction between companies like Meta, which he says plan to invest in communities for decades, and developers trying to secure a site and flip it. “They don’t really care as much about the local community, and they’re not invested for the long term,” he says. He argues that Meta’s projects face less opposition because the company has a reason to invest in local jobs and training.

• He responds to Meta’s landmark youth-safety settlement and to privacy concerns over the company’s smart glasses. Zuckerberg says he wants the youth safety settlement to become an industry framework, with YouTube and TikTok agreeing to the same restrictions so that tightening Instagram’s rules doesn’t simply send teens elsewhere. On glasses, he defends the recording light and says Meta needs to explain its safeguards more clearly as the devices go mainstream.

ICYMI: Sources ep. 1 with Sam Altman

Available everywhere you get podcasts.

Sources is a newsletter and podcast by Alex Heath about the AI race, featuring conversations with leading founders and technologists. Every week, Sources reaches the inboxes of thousands of decision-makers in tech, finance, policy, and media.

After an unreleased OpenAI model recently escaped its sandbox and hacked Hugging Face, Sam Altman tells me why the company slowed parts of its frontier research and delayed a major reinforcement-learning run: “Getting AI safety right is more important than any company’s momentum.”

We also discuss Astra, OpenAI’s next-generation family of models, which is…

Watch or listen to the full conversation on Tuesday, September 1st, available everywhere you can get podcasts.

Yesterday, TIME and I published a cover story called “Inside OpenAI’s Reboot.” I spent two weeks at OpenAI’s San Francisco headquarters and spoke with more than 20 executives, investors, customers, and rivals.

I went into the reporting process wanting to understand OpenAI’s last year, what allowed Anthropic to gain so much ground, and the internal mood a…

Read more

My social feeds have recently been dominated by buzz for two new AI products: Elon Musk’s Grok Bot and an invite-only, text-based AI assistant called Instinct.

While the latter has yet to officially emerge from stealth, the rumor circulating is that Instinct’s small team, which includes an early Sierra researcher named Noah Shinn, is currently raising a Series A round of funding to buy more compute and open its waitlist. Like Poke before it, which was recently sold to Cognition, Instinct seems to be orchestrating off-the-shelf models with some specialized fine-tuning and tool use. It’s unusually capable and proactive at web-related tasks. There are a lot of little nice touches that help onboard a user more quickly, like the agent automatically researching your background while you’re setting it up. Early user takeaways often describe it as “OpenClaw for normal people.”

Read more

There are several ways to interpret the implications of OpenAI’s moves this week for Anthropic.

OpenAI’s largest frontier reinforcement-learning run remains on hold while the company studies what Sam Altman described to me as “various degrees of misalignment.” OpenAI says its expanded monitoring adds roughly 20% to the inference compute being monitored. That isn’t 20% of all internal inference, but it is a substantial compute tax. If the company made this move as a public stock, it’s hard not to imagine a strong market reaction.

Anthropic’s alignment work, meanwhile, could simply be ahead of OpenAI’s. Its internal frontier models could be less capable, meaning they aren’t showing the same problems yet. Or Anthropic could be seeing warning signs it hasn’t discussed, with a model jump forcing the issue soon.

Read more

“I think it is a good time to slow down,” OpenAI CEO Sam Altman told me last week, describing the company’s decision to halt the development of its most powerful, unreleased models.

The company announced Tuesday that it’s implementing new safeguards that will slow its future AI development. The company recently paused training on its next set of models, codenamed Astra, for a little more than two weeks, according to executives, and its largest planned frontier training run remains on hold while the new guardrails are put in place.

It is the first time OpenAI has made such a move. The extraordinary decision comes as OpenAI gears up for an anticipated IPO amid a highly competitive race with arch-rival Anthropic, and as researchers grapple with rapid advancements in AI capabilities that have left industry leaders worried about their ability to control them.

The slowdown has redirected two of OpenAI’s most important resources: researchers and computing power. Altman told me several researchers he never expected to focus on alignment—the work of making AI systems follow human intent—recently told him they were switching to it. “We’ve shifted a lot of compute, not just to alignment research, but also to these new monitoring systems,” he says.

Read more

It has been awhile since I’ve done a subscriber survey. If you have a moment, consider answering a few questions about yourself. Results are anonymized, and this data helps me build a better media business. It’s coming up on a year since I launched Sources, and I’m so grateful for all your support.

The trick to raising $5 billion, according to Ali Ghodsi, is not to plan on raising $5 billion.

Databricks was heads down preparing for its Data + AI Summit in June when The Information called to say it was about to report that the company was raising money at a $175 billion valuation. “We were just scratching our heads,” Ghodsi told me Thursday. Databricks asked the publication not to print the story. It did anyway.

Then Ghodsi’s phone blew up.

Read more

Over the weekend, Mark Zuckerberg called me through his Meta glasses while speeding across the water toward a fishing spot. The noise cancellation was holding up against what sounded like a wind tunnel. Before he got there, he wanted to walk me through the 6,500-word pro-AI manifesto he had spent the past month or two writing.

Read more

Happy Friday. Thanks to all of you, especially those of you at Google, who wrote in to say you agreed with my last issue analyzing the implications of the company’s AI exec shuffle.

Today, I’m sharing a pre-weekend link roundup that I hope you find interesting. I’m in San Francisco next week for a special project I’m excited to share more on soon.

Feed check

• OpenAI paused training for Astra. “Our latest internal evaluations of Astra, one of our upcoming models, over the past few days indicate significant advancements in agentic coding and cybersecurity. These results, in addition to expert assessments, have led us to conclude last night that we cannot rule out critical cyber capabilities under our Preparedness Framework⁠.” Per the framework, this means OpenAI has to halt development on the model until appropriate security controls are in place. This is a big deal for any frontier lab. (So much for yesterday’s X rumors that Astra was shipping next week.) OpenAI also confirmed Astra is not the same model that hacked Hugging Face.

Read more

Inside Google DeepMind, today’s executive shakeup landed with more or less a shrug. Demis Hassabis handing day-to-day control of the lab to Koray Kavukcuoglu was closer to a formality than a bombshell, people there tell me, and while Jeff Dean’s exit stings, it’s being processed internally as more of a spiritual departure than a strategic one.

That’s not how the outside world took it. Alphabet shed 4 percent on the news that Demis is moving up to GDM chairman and Alphabet chief scientist, that Koray is stepping up as an SVP reporting to Sundar Pichai, and that Jeff is leaving after 27 years, along with Sanjay Ghemawat, Oriol Vinyals, and Quoc Le, to start Discovery Loop, a public-benefit corporation that wants to automate scientific discovery itself. The take writes itself: Google is losing the pillars of its AI efforts at the worst possible moment.

People I’ve talked to inside Google say Koray has been the main guy on the models for at least a year (Demis’s his own memo to employees essentially says this: “the Gemini models are in good hands with Koray and the leads, as they have been for a while”), and Demis has become increasingly less interested in shipping chatbot features and refereeing Google’s endless internal compute allocation fights.

Where his head has been instead:

Read more

This week, Jay Parikh, the EVP who runs Microsoft’s sprawling CoreAI org, sent his teams a memo that’s designed to end tokenmaxxing, tells engineers to review their individual AI spend, and makes OpenAI’s GPT-5.6 Sol the default model for internal use. The GitHub Copilot economics underneath the memo help explain it.

Read more

Read more

Follow ACCESS on Instagram:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.instagram.com/accesspodcast/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Follow Alex:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://sources.news/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://x.com/alexeheath/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Follow Ellis:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://meaning.company/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://x.com/hamburger⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ 

ACCESS is produced in partnership with the Vox Media Podcast Network.

Learn more about your ad choices. Visit podcastchoices.com/adchoices

Robin Rombach rarely does interviews, so when I got the Black Forest Labs CEO on a video call the other day, I started with the question hanging over his corner of the industry: are open models at risk?

“The state of open innovation might be at risk,” he told me. “It’s extremely crucial that innovation and information can flow across all kinds of borders... It might be easy to lose sight of that inside of massive commercial opportunities.”

Read more

I’m back from vacation. Did you miss me? The AI news cycle certainly did not pause while I was trying to unplug in Greece. Oh well.

The first time I interviewed Tom Alison, about a decade ago, he ran engineering for News Feed and drew me a diagram on a whiteboard to explain how ranking worked. He runs all of Facebook at Meta now. When we sat down in Los Angeles on Friday, I asked where AI ends up in what employees still call the blue app.

“We talk about this all the time,” he said. “Should Facebook be a human-only network forever or not? And we haven’t made a strong stance on that.”

I haven’t heard anyone at Meta say that publicly, and it’s telling that Allison left it open in the same week Facebook began rolling out a free verification badge built to prove a person is real. Mark Zuckerberg has spent the past couple of years describing AI-generated content as a third era for social media, predicting that AIs and humans will co-exist in feeds together. Alison’s version is looser.

Read more

Follow ACCESS on Instagram:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.instagram.com/accesspodcast/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Follow Alex:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://sources.news/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://x.com/alexeheath/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

Follow Ellis:

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://meaning.company/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://x.com/hamburger⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ 

ACCESS is produced in partnership with the Vox Media Podcast Network.

Learn more about your ad choices. Visit podcastchoices.com/adchoices

Schedule note: I’m taking the next two weeks off for some much-needed R&R. I trust (not) that no major AI news will break while I’m gone.

Anyone closely following OpenAI’s palace intrigue wasn’t surprised when Fidji Simo announced yesterday that she isn’t returning to her job as the company’s number two. Greg Brockman absorbed most of her remit in mid-May on a non-interim basis, all but confirming that she wasn’t coming back. Simo is staying on as an advisor while she continues to work on her health, and I’m told she has talked with Brockman almost daily throughout her medical leave.

Simo originally hoped that her leave would last a month or two at most, according to my sources. The AI race doesn’t idle, though, and while some companies can maintain velocity on autopilot when a senior leader steps away, OpenAI clearly couldn’t.

Read more

When I got Alexandr Wang on the phone ahead of Meta’s Muse Spark 1.1 release, I wanted to know how serious Meta was about its new API business for selling access to the model, which now puts it in more compeitition with OpenAI, Anthropic, and Google. Wang said that Meta sees the API as a real business, not just a learning exercise. Tokens are “gigantic and growing very, very quickly,” he told me, and capturing even a slice of that “can be a very meaningful business even for Meta at our scale.”

Every new API account comes with $20 in free credits, then $1.25 per million input tokens and $4.25 per million output, making it “one of the most attractive pricing levels in the market,” as Wang put it. The API also plugs into the OpenAI SDK, so developers can point their existing stack at Meta’s model. Wang said that was deliberate: “We wanted adoption of our model to be very efficient and easy for developers starting out.” A Meta-native stack will come with time. Claude SDK compatibility isn’t part of this release, but he didn’t rule it out either.

Read more