AI Daily Briefing — 2026-07-02

Wayve, a British autonomous driving startup, has launched an $85 million tender offer allowing employees to sell a portion of their vested equity. The program is backed by existing and new investors at the company’s current $8.5 billion valuation, which was set during a $1.2 billion Series D round in February. This marks Wayve’s second such liquidity event, following a similar offer tied to its $1.05 billion Series C in May 2024.

TechCrunch Disrupt 2026 has announced the lineup for its Builders Stage, a dedicated track focused on practical strategies for scaling startups. Scheduled for October 13-15 at San Francisco's Moscone Center, the event will host over 10,000 founders, investors, and operators. The stage features speakers like Grant Lee of Gamma, Leah Solivan of Precedent.vc, and Google's Robby Stein, who will share actionable insights on fundraising, hiring, go-to-market tactics, and AI adoption through real-world case studies and live Q&A sessions.

Key sessions address pressing startup challenges. One panel, "How to Win When You’re Not Building AI," explores how companies can thrive by focusing on fundamentals like retention and disciplined execution rather than AI hype. Another, "What Happens When OpenAI Ships Your Roadmap," discusses defensibility against competition from AI giants. "Winning Pre-Seed Without a Product" offers guidance on securing investment based on founder-market fit and vision, while Stein's talk examines how product decisions must evolve from MVP to billion-user scale.

Designed for founders ready to scale, the Builders Stage is one of six industry-focused tracks at Disrupt 2026. It aims to help startups navigate growth hurdles, from raising capital to transitioning from seed to Series A. Attendees can save up to $330 on tickets before prices increase. More speakers and sessions will be announced as the event approaches.

Meta is reportedly developing a cloud infrastructure business to sell access to its AI computing power and models, according to Bloomberg. The move follows massive investments in data centers and aims to generate revenue from excess capacity, pitting Meta against major cloud providers like Amazon Web Services, Google Cloud, and Microsoft Azure. The initiative, internally called Meta Compute, is led by infrastructure head Santosh Janardhan, AI lab leader Daniel Gross, and president Dina Powell McCormick.

This strategy mirrors recent moves by SpaceX, which signed a deal with Anthropic to lease all compute capacity at its Colossus 1 data center. Both companies are capitalizing on the growing demand for AI compute, suggesting that owning data centers may be more valuable than providing the best AI models or services. However, skeptics warn that the AI infrastructure boom could create a bubble, with rapidly depreciating chips and questions about whether AI companies can generate enough end-user revenue to justify trillion-dollar bets.

Meta has committed $182.9 billion to AI infrastructure through early 2026, including massive projects in Louisiana and Ohio—the latter described by CEO Mark Zuckerberg as the size of Manhattan. Unlike Google and OpenAI, Meta has not seen significant demand for its own AI models, such as Llama or Meta AI, which do not yet represent a material revenue line. To recoup costs, Meta may follow CoreWeave’s model of selling raw compute capacity or offer hosted access to its closed-weight model, Muse Spark.

Zuckerberg previously indicated that a cloud computing business was “definitely on the table” as a way to return on Meta’s AI superintelligence investments. TechCrunch AI has reached out to Meta for comment.

The Trump administration has removed export controls on Anthropic's Mythos and Fable AI models, ending a policy that effectively blocked public access to what many consider the most advanced AI systems released. The restrictions, imposed on June 12, required a special license to share the models abroad—a rule Anthropic found impossible to follow at scale, leading it to halt public access entirely. Starting July 1, the company will restore availability.

Commerce Secretary Howard Lutnick announced that Anthropic agreed to proactively detect security risks, collaborate on release protocols, and report malicious activity. However, the company had already made similar voluntary commitments months earlier, leading cybersecurity experts to view the original ban as punitive leverage rather than genuine security policy. Critics noted the restrictions followed Anthropic executives' public criticism of how the government and political opponents might use AI.

The reversal came amid growing pressure from Asian competitors like Fugu and Tulongfeng, which are approaching Mythos-level capabilities. To keep American AI globally competitive, Lutnick last week cleared Mythos for select White House-approved customers. OpenAI's latest models faced similar controlled releases. The administration's shifting AI policies have left the industry uncertain about future regulations, with an executive order in June calling for pre-release reviews drawing sharp criticism from analysts like Dean W. Ball.

Cloudflare announced a major policy shift on Wednesday, giving AI companies until September 15, 2026, to separate their web crawlers used for search from those used for AI training and agent services. Starting that date, Cloudflare's default settings will block so-called "mixed-use" crawlers from accessing any pages that display ads. The change applies automatically to new customers, new sites from existing customers, and all existing free users, though site owners can adjust the settings if they choose.

The move targets a growing tension in the digital ecosystem: publishers want their content discoverable through search and AI tools, but they also want to protect their intellectual property from being used without compensation. Cloudflare specifically highlighted that the world's largest search engine—widely understood to be Google—has access to roughly twice as much data as other AI firms because it bundles search crawling with AI data collection. Google has countered that its Google Extended bot allows site owners to opt out of AI training without affecting search rankings.

Cloudflare CEO Matthew Prince noted that non-human traffic now makes up the majority of internet activity, a milestone that arrived earlier than expected. "We must go further and act faster so a sustainable ecosystem can emerge," he said. To support that goal, Cloudflare is expanding its "Pay Per Crawl" marketplace into a broader "Pay Per Use" model, allowing publishers to charge AI companies when their content generates value, not just when it is fetched. The company estimates that over half of AI crawl traffic involves re-fetching unchanged pages, wasting bandwidth and compute resources.

Cloudflare is launching this initiative with two partners: Ceramic.ai and You.com. Publishers who opt in will receive payment when their content appears in Ceramic's AI search results or when You.com accesses premium material. Other AI companies can adapt this model to their own needs.