YouTube redefines the 'view', XPeng's $900M robot bet, and a laptop battery myth

From August 24, 2026, YouTube counts a public 'view' the moment a video starts playing — no minimum watch time — across long-form videos, livestreams and podcasts, bringing them in line with how Shorts already work. Public view counts will jump overnight. The catch: monetization still runs on 'engaged views' and watch hours behind the scenes, so your numbers grow but your payout doesn't. It's a bigger flex for sponsor decks, not a bigger paycheck — and in 2027 the Partner Program bar rises to 8,000 watch hours or 20M Shorts views.

The problem

For years a YouTube 'view' required real watch time, so public counts lagged behind actual plays and a creator's reach looked smaller than it really was.

What changed

From Aug 24, 2026, YouTube registers a public view the instant a video starts — no minimum watch time — across long-form, livestreams and podcasts, matching how Shorts already work. Public totals will jump overnight.

The catch

Monetization and Partner Program eligibility still run on 'engaged views' and engaged watch hours, so a bigger public number does not mean a bigger payout. And from 2027 the YPP bar rises to 8,000 watch hours or 20M Shorts views.

What it means for you

Treat the inflated public count as a sponsorship-negotiation number, not an earnings signal, and keep optimising for watch time and retention — that's still what actually pays.

XPeng's robotics unit has raised over $900M in its first funding round at a $6.3B post-money valuation — a record for China's embodied-AI sector — with IDG and Gaorong leading and Tencent and Alibaba joining as strategic investors. The money funds IRON, a humanoid with 76 degrees of freedom, 21 joints per hand and around 2,250 TOPS of onboard AI, targeting mass production by the end of 2026 and commercial deliveries in 2027. The robot race is quietly moving from demo reels to a production line.

The problem

Humanoid robots have been stuck in viral demo videos — impressive on stage, but nowhere near real production or everyday use.

What changed

XPeng's robotics unit raised over $900M in its first round at a $6.3B post-money valuation — a record for China's embodied-AI sector — backed by IDG, Gaorong, Tencent and Alibaba. It funds the IRON humanoid: 76 degrees of freedom, 21 joints per hand, ~2,250 TOPS onboard, targeting mass production by end-2026 and deliveries in 2027.

The catch

A record raise and a spec sheet aren't shipped units; robotics 'mass production' timelines routinely slip, and early deployment is limited to XPeng's own stores and campuses.

What it means for you

Read this as the moment humanoid robotics shifts from demo-reel to a funded production race — real household or workplace robots are still a couple of years out, not this year.

iOS hides a setting that instantly upgrades typing: Haptic keyboard feedback. Go to Settings > Sounds & Haptics > Keyboard Feedback and turn Haptic on. Now every keypress gives a subtle vibration, so your thumbs 'feel' each tap and you make fewer typos. It's off by default, which is why most people never find it — it costs a sliver of battery, but it's worth it. Try it for a day and you won't switch it back.

The problem

Typing on glass gives no physical feedback, so it's easy to miss keys and rack up typos without noticing.

What changed

iOS has a built-in Haptic keyboard feedback setting that gives a subtle vibration on every keypress — it's off by default and buried under Settings > Sounds & Haptics > Keyboard Feedback.

The catch

Apple notes it can slightly affect battery life, though in practice the drain is minimal.

What it means for you

Open Settings > Sounds & Haptics > Keyboard Feedback and switch Haptic on; give it a day and your typing accuracy noticeably improves.

When an AI keeps giving mediocre answers, the fix is usually the prompt, not the model. Copy-paste this: 'Here's the prompt I've been using: [paste]. Here's the disappointing output: [paste]. Rewrite my prompt so it gets a better result, and tell me the one thing I got wrong.' It turns the model into your prompt coach and quietly kills most 'bad AI' moments. Save it — you'll reuse it weekly.

The problem

When AI gives a weak answer, most people blame the model and give up — when the real issue is usually a vague prompt.

What changed

A simple meta-prompt turns the AI into its own prompt coach: paste the prompt you used and the disappointing output, then ask it to rewrite your prompt for a better result and name the one thing you got wrong.

The catch

It only works if you paste the real prompt and output; a vague 'make my prompts better' won't get you far.

What it means for you

Save it and run it whenever a chatbot underdelivers — you'll fix the ask instead of fighting the tool.