☕️ Memory shortage hits the MacBook Air

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

💻 Memory shortage hits the MacBook Air

📈 Trump sells early post access for $100K

🐉 Alibaba unveils its most powerful AI model

🐋 DeepSeek launches world's cheapest AI model

👓 Apple glasses to track your health

🇪🇺 EU can fine AI makers 3% of revenue

Plus: 🎁 11 other news you might like, 🧰 6 tools, and 📚 5 papers.

💻 Memory shortage hits the MacBook Air LINK

A global memory shortage is now squeezing supply of the MacBook Air, making Apple's most popular laptop harder to find before the end of August as the company hurries to restock.

Orders on Apple.com carry a roughly two-week backlog, and Mark Gurman reports Apple is buying RAM from China while pushing shoppers toward the $1,999 base MacBook Pro over the $1,299 Air.

Tim Cook called the situation "a hundred-year flood on the memory pricing," and relief looks unlikely before 2028, since the M6 Air due in spring 2027 will use the same LPDDR5X memory.

📈 Trump sells early post access for $100K LINK

Trump Media and Technology Group launched Truth API, a paid service that gives trading firms and other subscribers early access to Truth Social posts from Trump and other high-profile accounts, for as much as $100,000 a month.

The feed promises a real-time stream of the platform's "most market-moving Truths," and it stands out because Trump, its biggest user with 13 million followers, is also the largest shareholder of the parent company and could gain from it personally.

Democratic senators Adam Schiff and Elizabeth Warren asked the SEC to investigate whether the service breaks the law, calling it insider trading and market manipulation since Trump often reveals policy moves on tariffs and Iran there first.

🐉 Alibaba unveils its most powerful AI model LINK

Alibaba revealed Qwen3.8-Max, which it calls its most powerful AI model yet, as Chinese firms push to catch up with U.S. companies in the race to build stronger artificial intelligence systems.

Due out next week, the model has 2.4 trillion parameters and a context window of 1 million tokens, letting it read and work with thousands of pages of text at once, Alibaba said.

Alibaba said the model can code on its own for weeks with little human help, spending 16 days in one test building a coding tool, and can review legal papers, run financial research, and grasp long videos.

🐋 DeepSeek launches world's cheapest AI model LINK

DeepSeek released V4-Flash, a version of its flagship AI model that a research firm ranks as the cheapest well-known model to run on benchmark tests, costing over 100 times less than Anthropic's Claude Fable 5.

Artificial Analysis pegged V4-Flash at about 3 cents per test, versus 86 cents for Moonshot's Kimi K3, $1.86 for OpenAI's GPT-5.6 Sol and $3.15 for Claude Fable 5, charging $0.14 per million input tokens.

V4-Flash scored 50 out of 100 on the firm's Intelligence Index, matching Google's Gemini 3.6 Flash but trailing Kimi K3 at 57 and top models from Anthropic and OpenAI by nine or more points.

👓 Apple glasses to track your health LINK

Apple is developing health and fitness tracking for its planned non-AR smart glasses, though these features probably won't arrive with the first model expected in the coming year, according to Mark Gurman's Power On newsletter.

The idea revisits an abandoned plan for Vision Pro, where a version of Fitness+ would have let people follow workout classes while the headset analyzed their movements, but the project failed because the headset was too heavy.

Since the glasses lack displays, the approach must differ, and it could pair a heart-rate sensor like the one in AirPods Pro 3 with cameras that watch activities and suggest ways to improve workouts.

🇪🇺 EU can fine AI makers 3% of revenue LINK

The European Union can now inspect AI models before they launch, block them from its market, and fine providers up to 15 million euros or 3% of yearly turnover, whichever is higher.

The powers, part of the 2024 EU AI Act and enforced by the EU AI Office, took effect on Sunday and reach any company offering a general-purpose model in Europe, no matter where it is based.

Refusing an information request, giving misleading answers, or blocking a model evaluation is fineable on its own, exposing U.S. labs like Anthropic, OpenAI, and Google, which must appoint an EU-based representative.

Other news you might like

• Apple still plans a camera-free AirPods Pro update for 2026LINK

• Judge denies xAI’s request to block Minnesota ban on ‘nudify’ appsLINK

• How bitcoin cold wallets lost $70 million in an attack that never touched the devicesLINK

• A real macOS flaw worth $200K went unreported because Apple's bug bounty inbox was full of AI slopLINK

• Xbox consoles are now much more expensive in the EU and UKLINK

• EchoStar's Hughes filed for bankruptcy as its satellite internet subscribers fled to StarlinkLINK

• Google’s AI is digging up Chrome bugs that humans missed for yearsLINK

• Hugging Face Doesn’t Want to Sue OpenAI. It Does Want $100 MillionLINK

• AI's talent wars have a loyalty problemLINK

• SpaceX's post-IPO plunge sets tense backdrop for first earnings reportLINK

🧰 Trending tools

Plethora: a curated hub of tiny interactive experiences-games, art, music, puzzles, and toys-for quick creative breaks or discovery.LINK

Inventory: an AI-powered platform that tracks stock levels, forecasts demand, and automates reordering to help businesses optimize inventory decisions efficiently.LINK

yapyap: a macOS menu bar app that lets you post to LinkedIn instantly via keyboard shortcut, skipping the browser and distracting timeline.LINK

Ctruh Studio: a no-code, in-browser platform for building AR try-ons, 3D product visualizers, configurators, and virtual stores without hiring developers.LINK

Doxy: a browser-based editor that lets you write and format documents in Markdown and HTML, with instant preview and no compile times.LINK

mpai: joins a teammate's live Codex or Claude Code terminal session over Tailscale, preserving full conversation context so nobody re-explains progress.LINK

📚 Trending papers & reports

On-device personalization lets edge gadgets like wearables or hearing aids learn and adapt to a specific user in real time, without cloud retraining, hitting 96.8% accuracy on a benchmark recognition task.LINK

Continuous speech tokens at a lower frame rate keep richer voice detail while resisting the error buildup that normally derails long streaming audio generation, enabling more stable, higher-fidelity synthetic speech.LINK

Synthetic contrast MRI generates the dye-enhanced tumor images breast cancer scans need without injecting gadolinium contrast agents, improving tumor-boundary detection accuracy by ~22% and cutting boundary errors by over 39%.LINK

Infrared AI cameras can be tricked by a QR-code-shaped heat pattern placed in a scene, silently steering the system's captions and answers toward whatever false label an attacker picks, without looking suspicious.LINK

Sparse-reward AI training gets a bonus system that rewards agents for pushing past familiar territory, helping them learn faster and outperform existing methods when feedback is rare or delayed.LINK

See you tomorrow for a new dose of ☕️ Techpresso!