Apple built an empire on doing everything in-house. Now they're asking an AI startup for help with coding. What went so wrong that the tech giant had to swallow its pride? The answer reveals cracks in Apple's perfect facade – and hints at a major shift in Silicon Valley's power dynamics.
Apple's CEO just thanked Trump for Biden's biggest tech victory. But behind Tim Cook's peculiar case of selective memory lies $1 billion in tariffs, a high-stakes political gambit, and perhaps the most expensive thank-you note in corporate history. The real story? Even more fascinating.
What if everything we thought about AI training was wrong? Researchers just discovered that artificial intelligence can master complex tasks by studying a single example—matching the performance of models trained on thousands. The secret lies in finding that one perfect example.
Good Morning from San Francisco,
Silicon Valley's AI giants are squabbling like schoolchildren - over chips. 🍿
Anthropic wants to block China's access to advanced processors. Their evidence? Smugglers stuffing chips into fake pregnancy bellies and lobster shipments. Yes, really. 🦞
Nvidia fired back 🔥, telling Anthropic to stop spinning spy novels and compete fairly. The real battle? Computing power. Block China now, Anthropic argues, and by 2027 their AI training costs could soar tenfold. 💸
Meanwhile, Chinese firms aren't waiting around. They're stockpiling chips faster than a squirrel before winter. 🐿️
Stay curious,
Marcus Schuler
A fight over AI chips has sparked a Silicon Valley showdown. Anthropic wants tighter controls on China's access to advanced processors. Nvidia calls this misguided.
Anthropic's policy paper reads like a thriller. It describes Chinese smugglers hiding chips in fake pregnancy bellies and live lobster shipments. Nvidia's response? Stop spinning tales and compete on merit.
The clash centers on computing power. Anthropic argues America's edge will vanish without stricter rules. They want lower purchase limits and tougher screening. Nvidia sees these moves as thinly veiled attempts to hurt its overseas sales.
Time matters. Trump's team plans to update Biden's export rules by May 15. Chinese firms are stockpiling chips fast. Anthropic claims computing power doubles every two years - block China now, and by 2027, their AI training could cost 10 times more than U.S. efforts.
Nvidia disagrees. They point to China's AI talent, arguing trade barriers won't stop progress. "America cannot manipulate regulators to capture victory in AI," they said.
The numbers tell a story. America's share of global chip production dropped from 40% to 12% since 1990. Today, 90% of cutting-edge semiconductors come from overseas.
Why this matters:
Read on, my dear:
Prompt:
1990s color photograph of a large, shaggy creature with big, floppy ears and droopy eyes, napping beside a child on a grassy hill, under a cloudy sky
Google is bringing AI search to mainstream users for the first time. The company will add an AI Mode tab to Google Search for some US users in the coming weeks, moving the tool beyond its experimental Labs platform.
The new AI Mode works differently than regular search. Instead of listing webpage links, it gives AI-generated answers pulled from Google's search index. It sits in a new tab to the left of "All," "Images," and other search options.
The update puts Google in direct competition with AI search engines like Perplexity and ChatGPT. These specialized tools tap into real-time web data more effectively than standard chatbots like Gemini.
Google has added new features too. Users can now save past searches in a side panel for quick reference. The tool also shows cards with business details and product information, including real-time prices and shipping options.
US users no longer need to join a waitlist to try AI Mode in Labs before its wider release.
Why this matters:
Read on, my dear:
Recommend one evidence-based micro-habit (under 5 minutes daily) that creates compound benefits across my financial stability, wealth building, productivity, and personal development.
Core Requirements:
For Each Life Domain, Explain:
Implementation Guide:
Apple's stock tumbled 4% after hours despite beating earnings expectations, as investors digested Tim Cook's admission that predicting tariff costs beyond June is "very difficult" - a masterclass in CEO understatement. The tech giant is already sourcing half of its U.S.-bound iPhones from India and most other products from Vietnam, suggesting that Apple's long-term relationship status with China has officially changed to "it's complicated."
Apple plans to buy 19 billion chips from U.S. factories this year, with TSMC's Arizona plant already making processors for iPads and Apple Watches. CEO Tim Cook announced the move while detailing how Apple's shifting production from China to dodge Trump's tariffs - though he left unsaid whether 19 billion chips is a lot or a little for a company that sells hundreds of millions of devices yearly.
Amazon's stock dipped 2% after hours as the e-commerce giant joined Apple in the "we can't predict tariffs" club, with CEO Andy Jassy admitting it's "hard to tell what's going to happen" with Trump's 145% China levy. The company tried to soften the blow by noting that some third-party sellers might eat the tariff costs themselves - a comfort about as reassuring as a participation trophy.
Apple finally surrendered in its App Store payment battle, updating guidelines to allow external payment links in U.S. apps after Judge Gonzalez Rogers ruled the company "willfully chose not to comply" with previous orders. Spotify wasted no time submitting an app update with external payment options, while Apple muttered about appeals - a bit like insisting the chess match isn't over after checkmate.
Anthropic opened Claude's door to third-party apps today, letting users connect tools from Atlassian to PayPal to their AI chatbot. The company also supercharged Claude's research abilities, though with a $34.5 billion revenue target by 2027 and current earnings at $1.4 billion, they'll need more than just new features to catch up to ChatGPT's head start.
Temu yanked all China-shipped products from its U.S. store this week, leaving bargain hunters staring at empty shopping carts and sellers scrambling to adapt. With Trump's 145% tariffs about to kill the $800 duty-free loophole that fueled Temu's rise, the company's attempting a hasty transformation into Amazon-style local warehousing - though explaining that to angry customers whose wish lists just vanished might be a harder sell.
Twilio beat earnings estimates with $1.14 per share against expected $0.92, as revenue hit $1.17 billion this quarter. But Wall Street seems unimpressed - the stock has dropped 10.5% this year while the broader market only fell 5.3%, suggesting even strong profits may not be enough to win back investors.
A Microsoft-led research team with collaborators from multiple universities has shown that AI can master complex mathematics using just one training example.
Their paper "Reinforcement Learning for Reasoning in Large Language Models with One Training Example" demonstrates how reinforcement learning boosted AI math accuracy from 36% to 74% on advanced tests.
The findings upend traditional AI training methods. While most approaches use thousands of examples, the team - led by Yiping Wang at the University of Washington and Microsoft researchers - proved that one carefully chosen problem does the job. The AI continues improving even after mastering the training example, suggesting deeper learning at work.
The method works across different AI models and mathematical domains. An algebra problem helps with geometry and number theory. The AI also develops better self-checking habits, using terms like "recheck" and "recalculate" more frequently.
The research uncovered surprising patterns. Even when the AI's answers on training problems turned into gibberish, it still performed well on new ones. Minor errors in training answers (12.8 instead of 12.7) barely affected results.
Why this matters:
Read on, my dear:
Scale AI fuels the AI gold rush as Silicon Valley's premier data training powerhouse. Founded in 2016 by dropout wunderkind Alexandr Wang, the company transforms raw information into meticulously labeled datasets that feed hungry machine learning models.
• The Founders 📊 Alexandr Wang (19-year-old MIT dropout) and Lucy Guo (ex-Quora) launched Scale AI in 2016 in San Francisco. Now employs 300+ staff with thousands of contractors globally. The pair spotted a critical bottleneck: AI development was starving for quality-labeled data.
• The Product 💻 Scale combines human expertise and machine intelligence to deliver precision-labeled data. Their platform annotates everything from self-driving car footage to medical imagery. Strengths include autonomous vehicle data processing, military-grade security clearance, and reinforcement learning from human feedback. Their hybrid "human-in-the-loop" approach outperforms purely automated solutions.
• The Competition 🥊 Scale battles Labelbox, Appen, and Hive AI for data annotation dominance. Big Tech's in-house solutions (AWS SageMaker Ground Truth) threaten from one side while startups like Snorkel AI attack with automation. Scale's edge? Enterprise-grade quality and defense contracts that competitors can't match.
• Financing 💰 Scale's valuation rocketed to $14 billion after raising $1 billion in 2024, with backing from elite VCs (Accel, Tiger Global, Founders Fund) and tech giants (Amazon, Meta). The company has secured $1.6 billion total funding and reportedly targets a $25 billion valuation in 2025.
• The Future ⭐⭐⭐⭐ Scale AI sits at the control panel of the AI revolution, expanding beyond pure data labeling into model evaluation and safety testing. 🚀 As AI becomes mission-critical across industries, demand for high-quality training data will explode. But automation threatens their labor-intensive model.
☀️ Good Morning from San Francisco, Silicon Valley's elite can't decide if AI will replace humans or
Good Morning from San Francisco, Microsoft dropped a mic: AI writes 30% of their code now! 💻 Google matches this pace,
Good Morning from San Francisco, The tech world's most intriguing power couple hits turbulent waters. Microsoft and OpenAI,
Good Morning from San Francisco, Meta's AI chatbots slipped into dangerous territory. 🚨 The Wall Street Journal busted them
No jargon. No nonsense. Smart cookies subscribe. Just saying.
