Good morning, it’s Tuesday. China bypasses U.S. chip bans through offshore data centers and cloud leasing. The smartest models hallucinate more—even as their reasoning improves. Meanwhile, OpenAI drops its investor return cap but keeps its nonprofit at the helm.

In today’s Forward Future Original, we unpack the two beating hearts of AI—training and inference—and why understanding that split is key to decoding cost, speed, and innovation.

Read on!

P.S. Don’t miss this week’s prompt: the “Invisible Forces Audit.” It’s part self-reflection, part system reboot.

🗞️ YOUR DAILY ROLLUP

Top Stories of the Day

💰️ OpenAI Sticks with Nonprofit, Drops Profit Cap
OpenAI is ditching its plan to restructure under a for-profit board, keeping its nonprofit in control and converting its commercial arm into a Public Benefit Corporation. The move removes a cap on investor returns to attract more funding, reflecting the rise of multiple AGI players. CEO Sam Altman says the change better aligns with OpenAI’s mission to make AI benefit all of humanity.

🦜 Lightning-Fast, Open-Source Speech Recognition Model
NVIDIA has released Parakeet-TDT-0.6B-v2, a fully open-source automatic speech recognition (ASR) model, now available on Hugging Face. This model boasts the capability to transcribe an hour of audio in just one second when utilizing NVIDIA's GPU-accelerated hardware, achieving a Real-Time Factor (RTFx) of 3386.02 with a batch size of 128. With a Word Error Rate (WER) of 6.05%, it ranks at the top of Hugging Face's Open ASR Leaderboard.

📈 Microsoft Surges Ahead of Amazon in Cloud & AI
Microsoft outshined Amazon in recent earnings, with Azure's 33% YoY growth—half driven by AI—highlighting its growing dominance in enterprise tech. Amazon's AWS remains the cloud leader, but its 17% growth lagged behind Azure’s momentum. Microsoft impressed Wall Street with a clear AI vision, while Amazon's sprawling ambitions left its narrative less focused in a high-stakes AI race.

♻️ AI Robots Boost Recycling Power in Seattle
Amazon-backed startup Glacier is rolling out AI-powered robots to sort recyclables at Recology’s Seattle facility, helping recover plastics with up to 90% accuracy. Trained on billions of images, these robots improve sustainability outcomes. Glacier just raised $16M to expand its operations, including testing at Amazon’s own sustainability lab. It's a high-tech push to keep more waste out of landfills.

👩‍🏫 CEOs Urge AI, Coding in K-12 Classrooms
Over 250 top U.S. CEOs, including leaders from Microsoft and Uber, signed a public letter pushing for AI and computer science to be core subjects in K-12 education. They argue this is vital for U.S. competitiveness, citing countries like China and Singapore that already mandate such training. The plea aligns with Trump’s new executive order promoting AI education, despite looming budget cuts.

Enjoying Tuesday’s Newsletter? Forward it to a colleague—
it’s a great way to help us grow.

🗺️ GEOPOLITICS

How China Keeps Feeding Its AI Ambitions—Despite U.S. Chip Sanctions

Despite sweeping U.S. export restrictions, China continues to acquire NVIDIA’s top-tier AI chips through a web of smuggling, cloud leasing, and offshore data centers. New regulatory efforts target not just the hardware but the cloud providers enabling access, yet loopholes—like Malaysian data hubs and third-country intermediaries—undermine enforcement.

Singapore has become a major transshipment point, with suspicious trade patterns and arrests underscoring the scale of the grey market. As one expert put it, “the chip has sailed”—suggesting that containment may be futile unless the U.S. out-innovates rather than just out-regulates. → Read the full article here.

👾 FORWARD FUTURE ORIGINAL

Inference vs. Training

When ChatGPT took the world by storm at the end of 2022, millions of people experienced the impressive ability of artificial intelligence to generate human-like texts for the first time. What most users didn't see, however, was the massive infrastructure and complex process that made this digital miracle possible in the first place.

We interact with AI systems every day - be it through voice assistants, recommendation algorithms or automatic translations - but what we experience is just the tip of a technological iceberg. → Read the full article here.

🤖 MODELS

As AI Grows Smarter, Its Hallucinations Are Getting Harder to Ignore

AI systems touted for their reasoning abilities—like OpenAI’s latest models—are hallucinating more often, not less, even as their math skills improve. In tests, OpenAI’s top-tier models fabricated information in up to 79% of answers, a marked increase from prior versions. Experts say the very mechanisms that enhance complex reasoning—like multi-step thinking and reinforcement learning—also increase the chance of compounding errors.

Companies are scrambling to understand and control the problem, but even their own researchers admit: they still don’t know exactly how these models work. When it comes to AI truth-telling, the frontier remains murky. → Read the full article here.

🛰️ NEWS

What Else is Happening

💸 Rork Scores $2.8M After Viral Tweet: Founders were broke until one post praising their no-code app exploded—now backed by a16z Speedrun.

🧪 Anthropic Funds AI-Powered Science: New program offers $20K in credits to researchers using Claude models to tackle big biology and life sciences questions.

💰 Cursor Hits $9B Valuation: AI coding app maker Anysphere lands $900M from Thrive, a16z as its tool nears $200M ARR—despite investor jitters over AI hype.

😵‍💫 Meta’s AI Feed Feels Chaotic: The new public prompt stream is more bizarre than useful—showcasing AI's weirdest instincts, not its strengths.

👗 Pinterest Upgrades Visual Search: New tools let users refine fashion picks by vibe, color, or fit—no keywords needed, just tap and shop the aesthetic.

🚀 FORWARD FUTURE PRO

Our Most Popular Premium Posts

Practical guides, immersive tutorials, and sharp breakdowns crafted to elevate your AI game.

📟 PROMPT OF THE WEEK

The "Invisible Forces Audit" Prompt

This prompt helps you identify and leverage the hidden forces that shape your life, decisions, and outcomes - forces that most people overlook but have enormous influence.

I'd like to conduct a Weekly Invisible Forces Audit. Please help me examine:

1. Environmental factors: Identify 2-3 aspects of my physical environment that are subtly influencing my behavior, focus, or mood this week. What small environmental adjustments might create disproportionate positive effects?

2. Information diet: Analyze what information sources I'm regularly consuming. Which are adding genuine value versus creating noise? What information gaps exist that, if filled, would significantly improve my decision-making?

3. Social dynamics: What unspoken social scripts or expectations am I following that I haven't consciously chosen? Which relationships are energizing versus depleting my resources? What small social boundary adjustments might be beneficial?

4. Default choices: What decisions am I making on autopilot that could benefit from intentional reconsideration? What defaults have I inherited that no longer serve me?

5. Based on this analysis, recommend one "invisible lever" I can adjust this week that might create outsized positive effects across multiple areas of my life.

🧰 TOOLBOX

Visual Thinking, Voice Editing, and Conversational Data Analytics

🧠 GitMind: Free AI tool for collaborative mind mapping, diagrams, and brainstorming with templates for creative visual thinking.

🗣️ Lipsync: Edit speech, clone voices, and dub videos in any language with seamless lipsync—no training needed, up to 4K via API.

📊 Julius: Chat with your data to visualize, clean, and forecast insights—ideal for analysis, modeling, and expert summaries across sectors.

That’s a Wrap

Thanks for reading today’s newsletter—see you next time!

The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀