Good morning. CoreWeave's stock has surged 277% since its IPO, with retail traders driving the AI company from $23 billion to $72 billion while Wall Street shorts 30% of available shares. Even analysts calling the run-up "a bit of a joke" won't dare recommend selling—because apparently the most dangerous game in AI investing is trying to time when gravity kicks in.

— The Deep View Crew

In today’s newsletter:

• 🌧️ AI for Good: AI is helping people prepare for hurricanes

• 🤖 Meta’s AI push could (will) reshape the ad industry

• ⛳ Anthropic is changing the game for PGA golf fans

🌧️ AI for Good: AI is helping people prepare for hurricanes

AI has quietly revolutionized hurricane forecasting, delivering breakthrough accuracy that's saving lives and billions in damage.

What happened: Google's GraphCast AI model correctly predicted Hurricane Beryl's Texas landfall a full week before traditional models, which incorrectly forecasted that the storm would hit Mexico instead. The model outperforms Europe's premier forecasting system on 90% of verification targets while generating 10-day forecasts in under one minute.

The National Hurricane Center has been quietly using AI since 2008, but recent advances are dramatic. The center's AI system now achieves a 70% detection rate for rapid storm intensification, up from much lower rates a decade ago.

• Europe's new AIFS AI model predicted Hurricane Milton's landfall within 13 miles average error, compared to traditional models' 100-mile errors.

• Microsoft's Aurora AI achieved 20-25% improvement in tropical storm tracking over 2-5 days.

• The Hurricane Center's AI translation system reduces forecast translation time by 83% — from one hour to under 10 minutes — critical during rapidly evolving storms.

Why it matters: AI models can now predict hurricane landfalls up to 10 days in advance, compared to 6 days for traditional forecasting. But intensity prediction remains challenging — AI consistently underestimates wind speeds because models avoid making dramatic predictions. Human forecasters still lead the process; AI just gives them better tools faster.

Here Are 15+ AI-Friendly Small Biz Ideas

Starting your own small business is a dream for many of us – a difficult, exhausting, and often deeply rewarding dream. But thanks to the explosive advancement of AI, the barrier of entry to running your own operation is lower than it’s ever been. Which only leaves one question… What should you start?

Our suggestion? Check out this list from Salesforce: 15+ Small Business Ideas To Start Today. 

This free download contains more than a dozen business ideas, ranging from the innovative to the tried and true, to help you get the ball rolling on your next big thing. Plus, as a Deep View reader, you’ll be able to tap into your AI knowledge to automate things like customer service, email marketing, and beyond.

Download 15+ Small Business Ideas To Start Today right here. 

🤖 Meta’s AI push could (will) reshape the ad industry

Meta is building ad tools that could cut out the middleman. By the end of next year, advertisers will be able to use Meta’s AI to generate entire campaigns—from images to targeting—without ever touching an agency. 

What happened: Meta plans to roll out advanced AI tools that allow advertisers to fully generate and target ad campaigns on Facebook, Instagram, and WhatsApp by the end of next year. Brands will be able to upload a product image and a marketing budget, and Meta’s platform will handle the rest—creating visuals, writing copy, producing video, and delivering the ad to the right audience.

The new tools were first reported by the Wall Street Journal and are being built to automate the entire ad creation process. Meta already offers limited AI features, but this is a much deeper push into AI-powered campaign execution.

The announcement spooked global ad firms. Stocks for WPP, Publicis Groupe, and Havas all dropped by around 3 percent following the news. Investors see Meta’s move as a direct challenge to agencies that traditionally control ad planning, creative, and media buying.

Why it matters: Meta is entering a space already crowded with AI ad startups. Companies like AdCreative.ai, Arcads and Icon have been building similar tools for months, while Reelfarm focuses specifically on AI-generated video ads. Meta's entry validates this emerging market but is likely to squeeze out smaller players.

Meta is going after the $160 billion ad market it already dominates with new tools that could cut agencies out of the process. For small businesses, the tools could be a breakthrough, letting them run smart campaigns without needing external help.

With plans to invest up to $72 billion into AI and infrastructure next year, Meta is signaling that this is not a side project.

AI Video Repurposing Tool: Turn One Video Into a Content Engine

Turn your videos into a content engine powerhouse with Goldcast’s Content Lab.

With Content Lab, you can automatically turn your long-form content (think podcasts, YouTube videos, webinars, and events) into a robust library of snackable clips, social posts, blogs, and more.

See why marketing teams at OpenAI, Hootsuite, Workday, and Intercom are using this AI video repurposing tool.

The best part?

It’s free to get started, so try Goldcast’s Content Lab for yourself right here.

• Researchers think DeepSeek quietly trained its latest model using Google’s Gemini

• Kuwaiti sovereign wealth fund joins Blackrock, MGX for infrastructure project

• You can now generate AI videos for free with Microsoft Bing’s new Sora-powered tool

• OpenAI’s boardroom meltdown is getting the Hollywood treatment

• The FDA’s new AI tool for vetting medical devices is already struggling with basic tasks

• Meta just locked in 20 years of nuclear energy to keep its AI data centers powered.

• AI might wipe out entry-level jobs before anyone’s ready

• Boston Consulting Group: Data Scientist

• Meta: GenAI Project Manager

• Perplexity Labs: AI-powered research assistant that delivers cited answers, real-time web results, and contextual follow-ups in a chat-style interface

• Veed.io: AI-powered video creation for teams (with all the bells and whistles)

• Typewise: An enterprise-grade AI customer service platform

⛳ Anthropic is changing the game for PGA golf fans

The PGA TOUR is rewriting the playbook for sports coverage, but the real story isn't about golf—it's about the future of how we experience live sports.

Working with AWS Bedrock and Anthropic's Claude 3.5 Sonnet, the TOUR added AI-generated commentary to over 30,000 shots during THE PLAYERS Championship (just realizing how much the PGA likes to leave their caps lock on...), bringing contextual analysis to every player for the first time. While other leagues struggle with AI accuracy issues, the TOUR achieved 96% precision through strict validation checks and a custom rules engine that filters out common errors.

The technical breakthrough: The TOUR built a two-tiered system that separates facts from context. First, AWS Bedrock generates factual commentary from ShotLink data (distance, club selection, course conditions). Then, a second AI pass adds strategic context—why a shot matters for the player's round, tournament position, and career trajectory. The system processes data from 120-150 cameras and 36 radars capturing roughly 2 million data points during each tournament.

But the innovation extends beyond text generation. The TOUR uses what AWS calls "LLM-as-a-Judge"—a second AI model that evaluates the first model's output for factual accuracy, golf-specific language, and narrative coherence. Content flagged three or more times gets rejected automatically. This peer-review approach solves a major problem plaguing AI sports applications: how to validate subjective content at scale.

The business case: Golf's sprawling format makes human coverage economically impossible. A typical tournament spans four days across 200 acres with up to 156 players. TV broadcasts cover maybe 2,000 shots—leaving 28,000 others in relative obscurity. The TOUR produces nearly 22,000 hours of golf content annually, and AI commentary will eventually extend to all of it.

The competitive landscape: While the NFL focuses AI on injury prevention and the NBA deploys it for automated highlight generation, the TOUR is pioneering real-time content creation. The global AI in sports market is projected to grow from $915 million in 2024 to $3.4 billion by 2032, driven largely by enhanced fan engagement applications like this.

The roadmap reveals the TOUR's true ambition. Phase two adds synthetic voice commentary, enabling multilingual audio for all 48 Every Shot Live streams. Phase three introduces hyper-personalization—custom recaps for favorite players, "Golf 101" explanations for newcomers and even betting-specific analysis.

The TOUR has solved the core economic challenge of modern sports media: how to scale premium content creation without proportional cost increases.

The model-agnostic architecture built on AWS Bedrock is crucial here. By avoiding vendor lock-in, the TOUR can swap between AI models as they improve, optimizing for cost, speed, and quality on a per-use basis.

But the deeper implications concern competitive dynamics across sports. The TOUR's success will pressure other leagues to match this level of coverage granularity. Imagine AI-generated commentary for every NBA player's off-ball movement, or contextual analysis of every NFL snap for all 22 players. The technology exists—it's an economic and strategic choice.

AI bias in sports analytics could inadvertently favor certain playing styles or demographics. Data privacy concerns grow as leagues collect more granular performance metrics. And there's the fundamental question of whether algorithmic commentary can capture the nuanced storytelling that makes sports compelling.

By starting with factual commentary and building toward more subjective analysis, they're learning systematically about AI's limitations and fan preferences. The emphasis on validation and accuracy suggests they understand the reputational risks of hallucinating sports content.

For other leagues, the message is clear: AI-powered content creation isn't coming—it's here. The question isn't whether to adopt these technologies, but how quickly you can implement them before competitors gain an insurmountable content advantage. The TOUR has fired the starting gun on an arms race that will reshape how we experience live sports.

🤔 Your thought process:

Selected Image 1 (Left):

• “I didn’t think anyone writing a prompt could be this simple and reserved. Beautiful Turner watercolor”

• “Sorry to break my recent streak of picking the fake — this time I went with the "real" banding of the sky, and apparently you left that out of your AI prompt :-)”

Selected Image 2 (Right):

• “Should have gone with my hunch that the prettier photo was AI”

• “Details of grass blades and disturbed sand look random enough to be real”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

P.S. Enjoyed reading? Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning. Subscribe here!

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.