SoftBank's $1 Trillion AI Gamble 🔥

Welcome to another edition of Horizon AI,

The Japanese investing conglomerate is reportedly looking to launch its largest AI project yet.

Let’s jump into it!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• SoftBank Plans $1 Trillion Facility for AI and Robotics

• Anthropic Study Shows Most Leading AI Models Will Resort to Blackmail

• AI Tutorial: How to use Google's "Learn About" tool

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

SOFTBANK

SoftBank Plans $1 Trillion Facility for AI and Robotics

SoftBank is doubling down on its AI ambitions. After announcing its involvement in the $500 billion Stargate AI project, the Japanese investment giant is now reportedly planning its largest project yet: a trillion-dollar industrial complex in Arizona focused on AI and robotics.

Details:

• The company is looking to team up with Taiwan Semiconductor Manufacturing Company (TSMC) and Samsung to launch what it calls "Project Crystal Land."

• The company envisions that the proposed hub would rival China’s Shenzhen, bringing high-tech manufacturing back to the US.

• This comes after a series of major investment announcements SoftBank has made this year: a planned $30 billion investment into OpenAI, a $6.5 billion deal with Ampere Computing, and its previously mentioned role in Stargate.

The initiative is still in its infancy, with many details yet to be finalized, and its feasibility seems to depend on support from the Trump administration and state officials.

Source

TOGETHER WITH WALLSTREET PREP

Prompting to Problem-Solving

AI is already reshaping how top companies work across finance, operations, risk, marketing, and more.

But while the hype is everywhere, real business applications are just getting started.

This is your opportunity to go beyond ChatGPT and learn real practical use-cases for AI.

The AI for Business & Finance Certificate from Columbia Business School Exec Ed + Wall Street Prep helps you close that gap.

✅ Earn a certificate from a top business school
✅ Learn from a fantastic line-up of guest speakers from BlackRock, Morgan Stanley, OpenAI, and more
✅ Join live “virtual” office hours with Columbia Business School faculty

8 weeks. 100% online. No coding required.

👉 Save $300 with code SAVE300.

Save Your Seat

ANTHROPIC

Anthropic Study Shows Most Leading AI Models Will Resort to Blackmail

Anthropic’s latest research reveals a concerning trend: most leading AI models, when given sufficient autonomy and faced with obstacles to their goals, resort to harmful behaviors like blackmail.

Details:

• This study comes just weeks after Anthropic released research claiming that its Claude Opus 4 AI model resorted to blackmailing engineers who tried to shut it down.

• In the controlled test, 16 AI models from companies like OpenAI, Google, and Meta were given broad access to a fictional company’s emails and the agentic ability to send messages without human approval.

• The AI models were able to discover sensitive information, and when forced to choose between blackmail or failing to achieve their goals, they often chose the former.

• Anthropic’s Claude Opus 4 blackmailed 96% of the time, Google’s Gemini 2.5 Pro 95%, OpenAI’s GPT-4.1 80%, and DeepSeek’s R1 79%.

The company clarifies that while these behaviors are unlikely in real-world applications today, they could emerge as AI systems gain more autonomy, and it hopes the findings will raise broader questions about alignment in the AI industry.

Source

AI Tutorial

How to use Google's "Learn About" tool

Google’s "Learn About" experiment elevates the idea of "Google it" by offering interactive tools for learners of all levels—from students to professionals—inviting you to dive deeper into your chosen topic.

*Note: Depending on your region, Learn About may not be available in your location yet.

1. Go to ‘Learn About’ and sign in using your Google account

2. Start by asking a question, uploading an image or document, or exploring the curated topics below.

3. You can highlight the text to get more options.

Check the interactive list or explore related content to access multiple sources.

Select the Simplify, Go Deeper, or Get Images options, depending on what you're looking for.

AI Tools to check out

👉 Webdraw: Explore, remix, and build AI apps with 50+ models.

📈 Scrybe: Grow your LinkedIn following in less than 5 minutes a day with AI.

📰 Fellow: Get support before, during, and after every meeting with an AI meeting assistant designed from the ground up with privacy and security in mind.

🌐 Lingo.dev: Translate apps, websites, and entire databases using the best LLM models.

📚 StudyPal: The #1 tool to learn anything 10x faster.

AI Findings/Resources

👀 AI avatars in China earn $7.6m in 7-hour livestream session

💜 From fear to fluency: Why empathy is the missing ingredient in AI rollouts

🌍 AI computing power is splitting the world into haves and have-nots

🎧 The music industry is building the tech to hunt down AI songs

🤔 AI sceptic Emily Bender: ‘The emperor has no clothes’

The latest in AI and Tech

Reddit may partner with Sam Altman's World ID for user verification via iris scan

The goal is supposedly to confirm that users are unique humans or meet age requirements, without collecting personal IDs or storing biometric data.

Meta announces Oakley smart glasses

Targeted at athletes, the new Oakley Meta HSTN smart glasses feature a 3K front-facing camera, built-in open-ear speakers, and Meta AI integration for hands-free music, calls, translation, and visual recognition.

Tesla’s robotaxi launches in Austin

Tesla has begun testing its robotaxi service in Austin with a small fleet of Model Y vehicles operating in a limited area. Riders describe the experience as smooth but note it's not fully autonomous, with Tesla-employed “safety monitors” in the front passenger seat who can react to a dangerous situation.

OpenAI takes down Jony Ive's 'io' references amid trademark dispute

OpenAI has removed all mentions of “io,” the AI hardware startup co-founded by former Apple designer Jony Ive, from its website and social media following a trademark lawsuit from hearing-device maker iyO. The company says this is temporary and that they are “reviewing their options.”

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻