Hey Team,

A lot to unpack today. More teenagers are turning to AI chatbots for emotional support and even therapy, but is it actually safe, especially when we know AI still hallucinates and gets critical things wrong? Therefore, we’re launching a new segment highlighting recent AI mistakes so we can all stay alert to what these systems get right and wrong. On the enterprise side, OpenAI just released a study showing massive growth, but the real question is whether that momentum is sustainable as competition becomes stronger. Plus, we’ve got our Jobs Corner, and more key news and trends. Let’s dive in and stay curious.

• What AI got wrong today:
• Is ChatGPT safe for Mental Support?
• 🧰 AI Tools
• OpenAI’s Enterprise Momentum in One Snapshot
• 🛠️ AI Jobs Corner
• 📚 AI Learning Resource
• One Standard for all AI models

Subscribe today and get 60% off for a year, free access to our 1,500+ AI tools database, and a complimentary 30-minute personalized consulting session to help you supercharge your AI strategy. Act now as it expires in 3 days…

Get 60% off for 1 year

📰 AI News and Trends

• U.S. to allow Nvidia H200 sales to China. Nvidia won approval to sell H200 AI chips to “approved customers” in China, subject to a 25% surcharge, while reports suggest Beijing may still curb domestic access, clouding demand prospects.
• Cursor developed an internal AI Help Desk that handles 80% of its employees’ support tickets, says the $29 billion startup’s CEO.
• Warby Parker, Google to launch AI-powered smart glasses in 2026.
• SoftBank and Nvidia reportedly in talks to fund Skild AI at $14B, nearly tripling its value.

Other Tech News

• China Pivots to Domestic Demand as Trade Surplus Hits Record $1 Trillion
• Paramount is launching a hostile $108.4 billion takeover bid for Warner Bros. Discovery after the entertainment conglomerate chose Netflix as its buyer.
• Spotify Wrapped hit a record 250 million engagements in just 65 hours.
• Uber is turning rider data into insights that can be sold to marketers.

What AI got wrong today:

I’ve been building an AEO checker app using Vibe coding with Lovable, and I’m benchmarking it against Cursor and Gemini 3. One core feature sends users an email report with their site analysis. For that, I’m using Resend, which lets you connect your domain via DNS records, perfect for fast MVP testing and deployment.

I ran into DNS issues and asked ChatGPT for help. It confidently told me I didn’t need MX records. That advice was completely wrong. After ChatGPT doubled and tripled down on it, I added the MX records anyway, and everything worked instantly.

Just another real-world reminder:

LLMs still make confident, time-wasting errors that can distort your understanding of systems.

Trust your own thinking, verify with docs, and always cross-check when something doesn’t feel right.

Share

🛠️ AI Jobs Corner

Apply Today — Open Positions.

• Machine Learning Researchers (PhD)
• First-Line Supervisors of Productions and Operating Workers
• Computer and Information Systems Managers
• Database Administrators
• Apply Data Scientists
• Software Technical Writers

OpenAI’s Enterprise Momentum in One Snapshot

OpenAI released its State of Enterprise AI with a few key stats. This comes out when the model seems to be losing shine due to hefty competition from Anthropic and Gemini models. Adoption in Enterprise AI is still early, but the winners are being defined today.

• Usage is exploding: ChatGPT message volume is up 8× YoY, and API reasoning-token usage surged 320×, a clear signal that companies are moving beyond basic prompts.
• Real productivity gains: A new survey of 9,000 employees shows workers save 40–60 minutes per day, with the biggest wins in engineering and data science.
• Enterprise penetration: 36% of U.S. businesses now use ChatGPT Enterprise.
• Custom GPTs are taking off: Usage is up 19×, but most workers still don’t use advanced features like reasoning or search, meaning adoption is still early-stage with major upside left.
• Non-technical workers are coding more: Research and IT employees without technical titles increased coding-related AI messages by 36% in the last six months.
• Pressure is rising: Competition from Gemini, risk to consumer revenue, and $1.4 trillion in infrastructure commitments are real headwinds.
• Pushback on “AI is useless” claims: OpenAI’s data directly contradicts recent MIT/Harvard studies that found zero ROI and “workslop.”

Enterprise AI is delivering measurable time savings and accelerating into deeper technical work, but most companies are still only scratching the surface. Altman, knows the urgency and instructed employees to boost ChatGPT through the better use of user signals. OpenAI is set to release a new model this week, and it also plans to release another model in January with better images, improved speed, and a better personality, after which it will end the ‘code red’.

Share

Is ChatGPT safe for Mental Support?

A new UK study of 11,000+ youths shows that 1 in 4 teenagers (13–17) and nearly 40% of those affected by youth violence now turn to AI chatbots like ChatGPT for mental health support, driven by long NHS waiting lists, privacy concerns, and 24/7 access.

Usage is twice as high among Black teens, and victims and perpetrators of violence are significantly more likely to rely on AI than their peers. Teens describe chatbots as non-judgmental, always available, and safer than adults, especially when fearing school or police involvement. But experts warn this creates serious risks, citing lawsuits linked to suicide cases and weak regulation, with youth leaders stressing that “children need a human, not a bot.” The findings expose a growing gap between mental health demand and real-world access, which AI is rapidly filling by default.

🧰 AI Tools of The Day

• Search our large AI tools Database of over 1000 tools for free.

What We are Reading

• Book Browse — Helps you discover Great Books
• Opening Bell Daily — Industry-leading data newsletter covering the story of the stock market. Trusted by 190,000+ investors every single morning.
• The Sales Wagon — Proven sales tactics, psychology, and automation strategies that help you close more deals and build predictable revenue.

One Standard for all AI models

Major AI players, including Anthropic, OpenAI, Google, and Microsoft, are set to form a new open-source standards group called the Agentic Artificial Intelligence Foundation, organized by the Linux Foundation, to standardize how AI agents connect to enterprise software.

The goal is to make AI agents interoperable across apps — similar to how banks standardized electronic payments, by aligning on tools like Anthropic’s Model Context Protocol (MCP), OpenAI’s Agents.md, and Block’s Goose local AI agent. MCP is already gaining real traction across products like ChatGPT and Google Workspace, enabling agents to connect tools like Slack to automate workplace tasks. However, CIOs warn that security risks like prompt-injection attacks remain unresolved, especially as companies plug agents into sensitive systems like PagerDuty and internal financial platforms. Despite the risks and the tech industry’s long history of open-standards disputes, the formation of this group signals a major push toward shared infrastructure for enterprise AI automation.

🛟Is ChatGPT safe for Mental Support? was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.

Greetings team,
I’ll be speaking at AI Edge in two weeks. Come join us at what’s shaping up to be a standout AI event.

Today’s edition covers a wild week in the AI world: Kimi K2 just outperformed every major model, signaling how fast Chinese open-source systems are advancing while keeping costs dramatically lower. Anthropic is now on track to hit profitability years ahead of OpenAI, and a new Silicon Valley-backed startup is raising the question none of us expected so soon: Would you edit your future baby’s DNA?

Let’s dive in — and stay curious.

• Anthropic to be Profitable before OpenAI
• AI Tools
• Kimi K2 AI Model Beats all other LLMS
• AI Guides
• Would You Edit Your Future Baby’s DNA?

Subscribe today and get 50% off for life, free access to our 1,500+ AI tools database, and a complimentary 30-minute personalized consulting session to help you supercharge your AI strategy.

Get 50% off forever

📰 AI News and Trends

• xAI works on Grok Code Remote to rival OpenAI’s Codex
• Lovable has surged to nearly 8 million users and sees 100,000 new projects built daily.
• OpenAI to sell compute capacity to other companies, putting the company in direct competition with Amazon, Microsoft, and Google. Being a cloud provider would help fund big AI spending.
• Altman and Masayoshi Son Back 27-Year-Old Louis Andre’s Bid to Build “A New Bell Labs” With Episteme, a San Francisco Research Lab Aiming to Free Top Scientists From Bureaucracy and Turn Breakthrough Ideas Into World-Changing Products
• Wikipedia urges AI companies to use its paid API and stop scraping
• OpenAI is weighing a move into consumer health apps

Other Tech News

• SoftBank sold 32.1 million shares of Nvidia in October. It also sold part of its stake in T-Mobile for $9.17 billion to finance OpenAI
• With $19 billion in quarterly sales, TikTok Shop now rivals eBay
• Apple Plans Major New Satellite-Powered Features for iPhones
• China-Backed $23B Simandou Mine Launches, Promising Cheaper Steel and a Potential Guinea Boom Amid Growing Environmental Fears
• Tesla Breaks Ground on Massive Giga Texas Plant to Build Millions of Optimus Humanoid Robots

Anthropic to be Profitable before OpenAI

The Information

Anthropic and OpenAI are on wildly different financial paths. New documents show Anthropic is set to break even by 2028, while OpenAI expects $74B in operating losses that same year, about 75% of revenue driven by massive chip and data-center spending.

OpenAI plans to burn 14× more cash than Anthropic before reaching profitability in 2030, fueled by Sam Altman’s push for scale and $1.4T in long-term compute commitments. Anthropic, valued at $183B vs. OpenAI’s $500B, is growing its business more efficiently by focusing on corporate customers (80% of revenue) and avoiding compute-heavy ventures like video generation. In 2024, both companies burned roughly 70% of revenue, but Anthropic’s burn rate drops to 9% by 2027, while OpenAI’s remains high. OpenAI is betting big that demand will justify its infrastructure buildout, spending nearly $100B on backup data-center capacity alone, while Anthropic opts for steadier, revenue-aligned growth.

Courtesy of Sacra

Leave a comment

One small AI shift can 10× your results

Big breakthroughs rarely come from doing more.
They come from making one small shift that changes everything.

That’s what the AI Edge Digital Summit (Nov 18–20) is all about — discovering the smartest ways to use AI so you can:
⚡ Automate 80% of the work that eats your time
⚡ Turn a single idea into an entire marketing system
⚡ Scale faster — without hiring a bigger team

It’s not about becoming an “AI expert.”
It’s about knowing which 5 minutes of AI work replace 5 hours of manual grind.

Ready to find your leverage?
🎟️Grab your free ticket now and see what one smart shift can do for your business.

What we are reading:

• Learn How To Achieve Financial Freedom With a Business Model Called Instagram Automation
• Join 280,000+ entrepreneurs as we dive into wealth-building strategies, cash flow management, tax strategy, investing, and lifestyle design.

Share Yaro on AI and Tech Trends | Your Top AI Newsletter

Kimi K2 AI Model Beats all other LLMS

Kimi 2 Thinking Benchmarl

Moonshot’s new Kimi K2 Thinking, just changed the AI landscape. The fully open-source model is now outperforming GPT-5, Claude Sonnet 4.5, and Grok-4 on major benchmarks in reasoning, coding, retrieval, and agentic tool use.

Built as a 1T-parameter MoE (32B active), K2 Thinking scores 44.9% on HLE, 60.2% on BrowseComp, and 71.3% on SWE-Bench. Verified all frontier-level results. It supports 256k context, native INT4 inference, and long autonomous tool chains (200–300+ calls), while staying far cheaper than closed models. Moonshot released it under a modified MIT license, making it one of the most permissive high-end models available. The breakthrough marks a turning point: open-weight systems are now matching and in many cases beating proprietary giants like OpenAI and Anthropic. At a time when U.S. labs face scrutiny over trillion-dollar compute spending, K2 Thinking shows that frontier AI doesn’t require massive capital infrastructure, just efficient architectures.

For enterprises, it raises a blunt question: why pay for closed APIs when a free open model now leads the benchmarks?

🧰 AI Tools of The Day

Search our large AI tools Database of over 1000 tools for free.

Would You Edit Your Future Baby’s DNA?

A Silicon Valley startup called Preventive, backed by Sam Altman and Coinbase CEO Brian Armstrong, is researching whether gene-editing human embryos can safely prevent hereditary diseases, sparking fierce ethical and legal debate.

The company has raised $30 million and is studying CRISPR-based embryo editing in San Francisco, even though implanting a gene-edited embryo is illegal in the US and banned in most countries. Preventive says it isn’t trying to create a baby now, only to prove the technology can be made safe, but critics warn it risks sliding into “designer baby” territory and eugenics. Some insiders say the startup has explored doing future trials in countries with looser rules, while supporters argue it could one day eliminate devastating genetic disorders like cystic fibrosis or sickle-cell disease.

💰Anthropic to be Profitable before OpenAI was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.

💥Microsoft’s $33B Compute Gambit Turns Cloud Partners into AI War Machines

Brought to you by Delve:

The AI race is quickly becoming a battle for cloud power. Tech giants now sit on more GPUs than the electric grid can support, turning compute and energy into the new oil of the digital era. China is countering with massive energy subsidies that could tilt the balance in its favor, while Microsoft and others play chess-like moves to lock up cloud access and outsource power capacity.

For the rest of us, the best strategy is to stay ahead of the curve, keep learning, and experimenting. So today, we’re sharing 3 powerful AI-augmented tools to explore and a set of free AI courses to level up your skills. Let’s dive in and stay curious.

• China’s Cheap Power Push To Tilt the AI Race
• AI Tools — AI-augmented Workflows Tools
• Microsoft’s $33 B Compute Gambit Turns Cloud Partners into AI War Machines
• AI Guides — Three New Free AI Courses You Can’t Miss.

Subscribe today and get 50% off for life, free access to our 1,500+ AI tools database, and a complimentary 30-minute personalized consulting session to help you supercharge your AI strategy.

Get 50% off forever

📰 AI News and Trends

• AI Giants Face Power Crunch as OpenAI and Microsoft Admit Energy Shortages Now Limit Growth
• Apple will rely on Google Gemini models for the new Siri experience, moving away from competing with existing AI chatbots.
• Microsoft signed one multibillion-dollar data center deal with Australia’s IREN, and another with Lambda.
• Anthropic Projects $70 Billion in Revenue, $17 Billion in Cash Flow in 2028

Other Tech News

• Tesla’s European Sales Collapse as Chinese EV Rivals Erode Market Share, but
• Tesla has reportedly signed a deal with South Korea’s Samsung SDI to purchase over $2.1 billion in energy storage system (ESS) batteries across three years.
• Dating Apps are still hot, Facebook Dating has reached 21M+ daily users since its 2019 launch
• Kimberly-Clark makes $49B bet that Tylenol risk is overblown.
• Starbucks to sell 60% of China business to Boyu.

Turn RAG into riches with Delve’s AI for security questionnaires

Hate those endless security questionnaires? We do too. So we hired the sharpest AI brains from Stanford, MIT, and beyond — and made them do your grunt work. Delve uses state-of-the-art agentic RAG, not your average “retrieve + answer” hack.

Our agents think ahead: pull evidence, resolve conflicts, reason across your policy graph, interrogate your infrastructure, and draft bullet-proof responses. We’ve helped Lovable, Bland, Micro1, and a ton of the fastest-growing AI companies navigate and close review with almost every F50 — and saved dozens of hours with our AI-native compliance platform.

Book a demo with Delve and get $1,000 off SOC 2, HIPAA, GDPR, and more — and automate security questionnaires away forever.

Use YAROCELIS1KOFF for 1K off

China’s Cheap Power Push To Tilt the AI Race

China is cutting electricity costs by up to 50% for major data centers using domestic AI chips from Huawei and Cambricon, according to the Financial Times. Local governments in regions like Guizhou, Gansu, and Inner Mongolia are offering massive energy subsidies to tech giants, including ByteDance, Alibaba, and Tencent, a move to offset U.S. chip bans and boost homegrown computing power.

With industrial electricity dropping to ~$0.056/kWh (half U.S. rates), China could rapidly expand its AI infrastructure, training more models and deploying AI at scale. US AI companies are struggling to power GPUs and are sitting on more hardware than they can power. Cheaper energy may help Chinese firms narrow the compute gap with the West, even if their chips are less efficient. Lower-cost AI experimentation, faster scaling, and a stronger domestic ecosystem all powered by state-backed energy policy.

WSJ — Data Centers are being built in rural America at record paces, but the energy to power them may not be reliable.

The AI race has become an energy race, and those with access to more and cheaper energy will have the upper hand.

Leave a comment

What we are reading:

• Learn How To Achieve Financial Freedom With a Business Model Called Instagram Automation
• Join 280,000+ entrepreneurs as we dive into wealth-building strategies, cash flow management, tax strategy, investing, and lifestyle design.

Share Yaro on AI and Tech Trends | Your Top AI Newsletter

🧰 AI Tools of The Day

AI-augmented Workflows Tools

• Workstreams.ai — If you’re looking to embed AI workflows inside your team communication and task management stack, which is aligned with your fractional CTO and automation role, this tool is a practical pick.
• AflowMind — Helps in the planning/ideation phase of workflows. Not full agent orchestration, but strong for mapping complex processes, startup idea planning, etc.
• Langflow — A visual drag-and-drop workflow builder for LLM chains, agents, memory, and vector stores. It’s positioned as a direct alternative to other AI workflow builders like Flowise.

Search our large AI tools Database of over 1000 tools for free.

Microsoft’s $33 B Compute War

Courtesy of Reuters.

Microsoft has quietly turned the cloud race into a power play. Its latest $9.7 billion deal with IREN converts a former bitcoin miner into a GPU fortress for AI workloads, following a $17.4 billion contract with Nebius and earlier multi-billion-dollar pacts with CoreWeave. These “NeoCloud” alliances give Microsoft instant access to hundreds of thousands of Nvidia GB300 chips without waiting years to build data centers. In effect, Microsoft is outsourcing infrastructure risk while securing a near-monopoly on high-end compute capacity.

These NeoClouds act as leveraged shock troops, borrowing billions against Microsoft contracts to rapidly deploy liquid-cooled GPU farms. It’s a new kind of industrial policy funded by debt and fueled by AI hype. This leads to computing that comes online faster than regulators or power grids can respond. Microsoft gains agility and scale; its partners gain survival-level revenue; Wall Street gains another asset class built on AI infrastructure.

Competitors like Amazon and Google are countering with their own massive compute commitments, while China is cutting energy costs by half to power domestic AI chips. Microsoft’s strategy resembles a chess match for compute dominance, trading capex for strategic positioning and locking up GPU supply before anyone else. The risks are over-leveraged partners, soaring energy costs, and a race that may outpace real AI demand.

🧰 AI Guides

Three New AI Courses You Can’t Miss.

1. University of South Florida Free Micro-Course on AI Skill Building — A new self-paced micro-course launched in early November 2025 for non-technical professionals and educators. No coding background required; focuses on how to incorporate AI into work and life.
2. Microsoft Fabric Data Days: 50+ days of immersive Data & AI learning — Kicked off November 4, 2025. A series of workshops, sessions, and global community events aimed at data professionals at all levels to build AI + data skills.
3. Google & Kaggle 5-Day AI Agents Intensive (Nov 10–14, 2025) — A crash-course on AI agents (autonomous systems, tool-calling, planning) by Google/Kaggle

Explore our AI Guides — from coding to photography and beyond, find step-by-step tips to put AI to work for you.

Brought to you by Delve: was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.