Billy Joel's Long-Lost Tape, a Chipmaker Overtakes Tencent, and Other Ripples

When a Lost Interview Meets AI

Some stories just wait. A 1971 recording of a 22-year-old Billy Joel—before he was the Piano Man—sat in a drawer until director Adam Ripp stumbled on it. On the tape, Joel talks about being homeless, moving to LA, planning his next record. Ripp turned it into a 20-minute short, 1971: Billy Joel’s Lost Interview, premiering September 25 at the Long Island Music and Entertainment Hall of Fame Film Festival.

But the face you see isn't real. Using authorized photos from that era, the filmmakers trained a generative model to make a young Joel appear to speak the recorded words. The voice is authentic, but the performance is synthetic. Joel didn't participate or endorse it. It's a strange hybrid—a documentary that feels like a deepfake but isn't malicious. It makes you wonder: when we resurrect the past with algorithms, are we preserving history or inventing a new version of it?

I keep thinking about that tape. It’s not just about Billy Joel; it’s about how we’ll remember everything. Ten years ago, this would have been a simple audio clip. Now it’s a full visual experience, pixel by pixel. That’s a historical activity in itself—one that forces us to ask what “real” means when the past gets a digital facelift.

A Chipmaker Surpasses Tencent—For a Day

On August 13, ChangXin Memory Technologies (CXMT) closed with a market value of 3.54 trillion yuan, just edging past Tencent’s Hong Kong-listed value of roughly 3.44 trillion yuan (after currency conversion). That made CXMT the most valuable company traded on both A-shares and H-shares—a moment that felt monumental for China’s semiconductor ambitions.

Tencent’s shares fell 4.46% that day, which helped tip the scales. But the story isn’t just one day’s trading. CXMT has been riding a wave of domestic chip demand, while Tencent faces regulatory pressures and a slowing consumer market. Still, rankings like this are fleeting. They change with every tick, every exchange rate, every rumor. A week later, it might be Tencent again.

What sticks with me isn’t the exact number—it’s the broader shift. Memory chips, once a footnote, are now central to national tech strategy. For business historians, this crossover is a reminder that “biggest” is a moving target. The record books can flip overnight.

Measuring the Past: IPOs, Revenue, and the Numbers Game

This week also brought news that Anthropic’s investors are betting on a fall IPO that could value the AI company at over $2 trillion, based on projected revenue of $100–120 billion by 2026. That’s a bet on the future, but it’s also a lesson in how we measure historical significance—through numbers that may or may not hold up. I’ve seen too many projections that looked great on paper and evaporated within a year.

Lenovo, meanwhile, reported a record quarter: revenue hit $26.94 billion, up 43% year over year. Its infrastructure solutions group saw revenue nearly double, and AI-related revenue grew 60%. Those figures will get cited in annual reports and history books, but they’re snapshots. The next earnings call could revise them.

And then there’s JD.com, which posted its first quarterly revenue decline in over a decade, down 2.9% to 346.4 billion yuan. The drop is blamed on weak consumer spending—a reminder that even giants can stumble when the economy cools. For historians, these ups and downs are the raw material of economic narrative.

Rewriting the Rules: Open Source and Price Hikes

History isn’t just about what happened—it’s about who controls the record. This week, X (formerly Twitter) open-sourced its “For You” recommendation algorithm under an Apache 2.0 license. Developers can now download the code, run the ranking system outside the company, and even submit changes. It’s a transparency move, but with limits: some Grok-based moderation tools were kept closed to prevent abuse. I guess they’re not fully ready to let go.

In a different kind of record-keeping, Apple is reportedly planning to raise iPhone prices by $100 across the board, including upcoming models. That would be a historic shift for a company that has kept US prices stable for years. Meanwhile, Google released Gemini 3.7 Flash, a new AI model aimed at coding and agentic tasks, just three weeks after the previous version. The pace is dizzying. Each release rewrites the technical history of AI.

Everyday History: How We Live, Ride, and Remember

History isn’t all corporate battles and AI models. Sometimes it’s about how we get around. BYD announced that over 3.52 million of its vehicles now have driver-assistance systems, generating 220 million kilometers of driving data daily. That’s a lot of data—but it’s also a record of how quickly autonomous tech is becoming mainstream.

Didi’s second-quarter orders hit 5.052 billion, up 13.2% year over year, with international orders growing 29%. In Brazil, its food delivery service 99 Food now covers over 100 cities. These numbers tell a story of urban mobility evolving in real time. I’m not sure what it means for the future, but it’s clear that ride-hailing is no longer just about cars.

And for those who prefer two wheels, Zhang Xue Motorcycles received an investment from Sequoia China, with monthly deliveries exceeding 10,000 units. The company’s revenue is projected to jump from 670 million yuan last year to 2–2.5 billion this year—a small but telling sign of how niche markets can boom. Who knew motorcycles could be a tech story?

Preserving the Present for the Future

What does it mean to document history as it happens? This week’s news offers a mix of deliberate archiving and accidental record-keeping. Apple is in talks to pay publishers for real-time news to feed an AI-powered Siri, which would create a new kind of historical record—one that’s constantly updated and queried. Imagine asking Siri in 2050 what happened today, and it knows because it was there.

At the same time, DeepSeek released an open-source agent harness that lets developers mix and match plugins. It’s a tool for building AI systems, but it’s also a way to document how those systems are constructed—a blueprint for future historians of technology.

Even coffee gets a historical angle: Tencent’s WorkBuddy is hosting a coffee festival with five brands, blending AI tasks with latte art. It’s a small event, but it shows how even daily rituals become part of our collective memory.

In the end, history isn’t just about the past—it’s about the choices we make now about what to remember and how. From Billy Joel’s resurfaced voice to a chipmaker’s market cap, we’re constantly writing and rewriting the record. The question is whether we’ll have the tools—and the wisdom—to read it accurately.