PLUS: Huawei's chip workaround, AI's productivity paradox, and an AI that sold a house
Happy reading
Meta is officially entering the subscription business, launching new paid tiers for its massive user base across Instagram, Facebook, and WhatsApp. These new plans are designed to sell access to more powerful AI features, moving beyond the company's traditional ad-supported model.
The strategy marks a significant pivot for the social media giant, turning its free platforms into a marketplace where advanced AI is the premium product. But as Meta makes this bet, will users be willing to open their wallets for AI tools on platforms they've always used for free?
In today’s Next in AI:
Meta's new paid AI subscriptions
Huawei’s new chip workaround
The AI productivity paradox in coding
An AI chatbot sells a house
Meta's AI Money Moves
Next in AI: Meta is rolling out new subscription plans for Instagram, Facebook, and WhatsApp, introducing premium tiers under the "Meta One" brand that unlock enhanced AI capabilities for a monthly fee.
Explained:
The AI-focused plans include Meta One Plus ($7.99/mo) and Meta One Premium ($19.99/mo), with the premium tier offering more capacity for complex tasks, deeper reasoning, and increased video and image generation.
These new consumer plans are a separate offering from the existing Meta Verified service, which remains focused on verification badges and impersonation protection.
This strategy allows Meta to diversify its revenue streams beyond advertising by creating a multi-tiered service model on top of its globally saturated social platforms.
Why It Matters: The move signals a major shift, turning Meta's free, ad-supported ecosystem into a premium marketplace where advanced AI is the main product. For professionals, this could unlock powerful, integrated AI tools for content creation and analysis, directly within the apps they use daily.
Huawei's Chip Workaround
Next in AI: Huawei has unveiled a new chip architecture, the 'Tau Scaling Law,' designed to help China produce advanced semiconductors and sidestep US sanctions on key chipmaking technology.
Explained:
The new approach aims to achieve a transistor density equivalent to a 1.4-nanometre process in high-end chips by 2031, which would significantly close the gap with global leaders.
This strategy works by bypassing the need for the most advanced extreme ultraviolet (EUV) lithography machines, which China currently cannot access due to export controls.
Analysts are calling it "another DeepSeek moment," suggesting it could dramatically boost investment and confidence in China’s domestic semiconductor ecosystem.
Why It Matters: If successful, this development could reduce the effectiveness of US tech sanctions and accelerate China's goal of semiconductor self-sufficiency. This puts immense pressure on Western competitors and reshapes the global race for AI computing power.
AI's Productivity Paradox
Next in AI: A new report analyzing 22,000 developers reveals a surprising trend: while AI coding assistants make individual developers faster, they are slowing down entire teams and increasing bugs.
Explained:
AI tools are creating more work downstream by generating a higher volume of lower-quality code, increasing the time teams spend on pull request reviews and fixes.
The study found that teams using AI see their overall system throughput drop by 70-80% and experience a 50% increase in defect rates per developer.
While some argue this is a temporary deployment problem, the data shows even high-performing engineering organizations are experiencing the same negative effects.
Why It Matters: This highlights the danger of measuring AI's impact on isolated tasks instead of the entire workflow. The findings feed into a broader discussion about the AI productivity paradox, where individual speed boosts aren't yet translating into measurable gains for businesses.
Your Next Realtor is an AI
Next in AI: A New York Times technology reporter put AI to the ultimate test by using a chatbot to sell his house. The bot handled everything from pricing and marketing to negotiation strategy, resulting in a sale that saved him over $90,000.
Explained:
Reporter Stuart Thompson tasked Google's Gemini with the entire home-selling process for his upstate New York property. The AI analyzed the market to set a price, wrote a listing so convincing a seasoned agent mistook it for a professional's, and scheduled all 20 showings.
The experiment was a financial success, securing three offers above the asking price and a final sale of $605,000. This outcome demonstrates how AI is already reshaping real estate, empowering individuals to bypass traditional processes and save on hefty commission fees.
Beyond automating tasks, the AI provided effective negotiation strategies that led to the buyer covering their own agent's commission. This shows AI's potential to not just execute tasks but also to provide valuable, strategic guidance in complex transactions.
Why It Matters: This real-world example shows AI's power to democratize access to specialized knowledge that was once locked behind professional paywalls. Industries that rely on agents or middlemen may face significant disruption as powerful AI tools become more accessible to the public.
AI Pulse
Amazon scrapped its internal AI leaderboard, KiroRank, after employees engaged in “tokenmaxxing” to inflate their usage metrics, telling staff to use AI to solve business problems, not just for the sake of using AI.
Sergey Brin told Google's Gemini team in a memo that working 60 hours a week is the “sweet spot” for productivity, warning that the final race to AGI is underway and requires a turbocharged effort.
Zig banned all AI-assisted code contributions to its open-source programming language after its president called the submissions “invariably garbage” that have negative value and waste reviewers' time.
Researchers warned that thousands of "abliterated" open-weight AI models are proliferating online, allowing anyone with a consumer laptop to easily remove safety guardrails and generate dangerous content with no oversight.