Presented by
Good morning, AI enthusiasts. Google just invested $75M in A24 and opened a DeepMind research lab inside one of Hollywood's most celebrated studios. For a company that has spent years on the defensive against copyright lawsuits from film studios, this marks a different approach
In today's recap:
Google and A24's first-of-its-kind AI filmmaking partnership
SpaceX's $6.3B compute deal with open-source Reflection AI
Prepare to benchmark Seedance 2.5 before its early July launch
Oracle's 21,000 AI-driven layoffs in fiscal 2026
4 new AI tools, prompts, and more
GOOGLE & A24
Google bets $75M on A24's AI future
Recaply: Google just announced a $75M investment in A24 alongside a DeepMind research partnership to build AI-powered filmmaking tools, marking the first time Google has taken a stake in a film studio.
Key details:
DeepMind researchers will work directly inside A24's production pipeline, giving filmmakers the chance to test and shape AI tools as they develop new workflows.
The $75M stake is Google's first investment in any film studio, with the deal non-exclusive and spanning multiple projects, but not granting Google access to A24's film or television library.
The Wall Street Journal reported the deal does not give Google access to A24's film or TV library, a detail absent from Google's own announcement.
The partnership has no fixed end date, with both teams expected to begin hands-on collaboration right away following the June 22 announcement.
Why it matters: There's been plenty of skepticism about AI in creative industries. Copyright fights and artist backlash have been the norm. Google and A24 are trying a different approach. They're giving artists direct access to the tools before those tools go public, so the creative community shapes the output rather than reacts to it. Whether that goodwill survives the first film is another question, but the structural bet is sound.
PRESENTED BY WISPR FLOW
Your prompts are leaving out 80% of what you're thinking.
When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning — constraints, edge cases, examples, tone — and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Try Wispr Flow free — works on Mac, Windows, and iPhone.
SPACEX & REFLECTION
SpaceX signs $6.3B AI compute deal with Reflection
Reuters
Recaply: SpaceX just locked in a deal with Reflection AI, granting the open-source startup access to Nvidia GB300 chips at Colossus 2 for $150M per month through 2029, with total payments reaching $6.3B.
Key details:
Reflection gets immediate access to Nvidia GB300 chips at SpaceX's Colossus 2 data center, with either company able to exit with 90 days' notice after the first three months.
The $150M monthly payment totals $6.3B if the deal runs through 2029, with Reflection last valued at $25B following its most recent funding round.
Reflection has not yet released a public frontier open-source model but is working with the Department of Energy's Genesis Mission and broader Pentagon AI efforts, according to CNBC.
Reflection's $150M monthly payments to SpaceX begin July 1, 2026, with no fixed milestone tied to a public model release.
Why it matters: SpaceX went public, acquired Cursor, and now signed a $6.3B compute deal with an open-source AI lab in the span of weeks. That's not a rocket company running sideline AI experiments. It's a company with a massive data center build-out deciding to monetize capacity the way cloud providers do. Reflection is also a meaningful signal. Open-source frontier labs now need, and can pay for, infrastructure at the scale only a handful of operators can provide.
GUIDES
Prepare to benchmark Seedance 2.5 before its early July launch
Recaply: In this tutorial, you will learn how to set up a benchmarking framework for Seedance 2.5 before it launches in early July, so you can evaluate it against your current video tools on day one and publish an informed comparison.
Step-by-step:
List your current AI video generation tools (Kling, Hailuo, Sora, or whichever you use) and document their output on 3 specific test prompts: a motion test, a character consistency test, and a scene transition test.
Review Seedance 2.0's known limitations (15-second clip cap, limited camera control, cross-scene character drift) to understand the gaps 2.5 is designed to close. ByteDance has confirmed 4K output and near-real-time generation as the two primary improvements.
Build a scoring rubric with 5 categories: resolution quality, prompt adherence, motion realism, generation speed, and character consistency. Score each existing tool on your rubric now to establish a baseline.
Join the ByteDance Seed waitlist at seed.bytedance.com and, if you plan to test HappyHorse 1.1 in parallel, follow Alibaba Cloud's Model Studio for API access, giving you a two-tool comparison on launch day.
On launch day, run the same 3 prompts through Seedance 2.5 and score them against your rubric. Publish the comparison with scores, not just impressions, to give readers something specific to evaluate.
Pro tip: Use identical prompt phrasing across all tools to isolate each model's output rather than your prompting differences. Save your test prompts in a shared doc so you can reuse them for every major model launch.
TOGETHER WITH PULLEY
Better cap table management starts here
Cap table management doesn’t have to be frustrating. From issuing grants to 409A valuations or ASC 718 reporting Pulley can make it simple.
Just ask Linear. They knew they needed a partner who could handle the complexity of their equity management. That’s why they migrated to Pulley.
TOOLS
Trending AI Tools
🎥 Seedance 2.5 - ByteDance's upcoming AI video model targeting 4K resolution and near-real-time generation, launching early July
⚙️ Grok Build - xAI's terminal coding agent for professional software engineering, in early beta for SuperGrok and X Premium Plus subscribers
🤖 HappyHorse 1.1 - Alibaba Cloud's production-ready video synthesis model, now live on Model Studio with full API access for enterprise and developers
🛡️ GPT-5.5-Cyber - OpenAI's cybersecurity model with top CyberGym scores, available to enterprise defenders through the Daybreak Cyber Partner Program
NEWS
What Matters in AI Right Now?
OpenAI unveiled GPT-5.5-Cyber, a cybersecurity model with top-ranked CyberGym benchmark scores, now available to enterprise defenders through the Daybreak Cyber Partner Program alongside a Codex Security plugin and a Patch the Planet open-source initiative.
Dell launched the PowerEdge XE8812, an AI server built on Nvidia's Vera Rubin NVL4 architecture, scaling to 144 GPUs per rack with 50% more memory per socket compared to the prior generation.
IBM joined the OpenAI Daybreak Cyber Partner Program, deploying GPT-5.5-Cyber inside enterprise workflows through a new application security service that identifies and validates software vulnerabilities with read-only code repository access.
Nvidia launched the Vera Rubin platform at ISC 2026, delivering 7 exaflops of AI and 5 petaflops of native FP64 performance in a single rack, with systems from Dell, HPE, GIGABYTE, Bull, and Supermicro available in Q4 2026.
Stripe introduced Stripe Directory, a new search interface that lets users and AI agents find and connect with businesses on the Stripe network directly.
Oracle revealed its workforce shrank by 21,000 employees in fiscal 2026, dropping from 162,000 to 141,000, with the company citing AI adoption and restructuring alongside $1.84B in severance costs.
🧡 Enjoyed this issue?
🤝 Recommend our newsletter or leave a feedback.
How'd you like today's newsletter?
Cheers, Jason
