No Image Available

The Many AI Twists and Turns Accelerate. ARD #121

Today’s theme: the many twists and turns ahead in this AI Tech Wave — and they’re coming fast and furious. Three items today — Google, Apple and DoorDash — each with its own unexpected twist, each with my Take, then my Overall Take. Plus a Gadget AI on how ‘RAMageddon’ is hitting Windows PCs on 8GB of RAM far harder than Apple, and two questions. Let’s get started.


(1) Google Falls Back Again in the AI Coding Race

MP TAKE: Google has had quite the twists and turns with Gemini lately. From its ‘Code Red’ moment when OpenAI’s ChatGPT launched over three years ago, to returning that ‘Code Red’ to OpenAI with a resurgent Gemini AI, to now — another turn back. This time it’s in the AI coding race against Anthropic’s Fable 5 and OpenAI’s GPT-5.6 — that layer of AI agents, AI coding and AI productivity on top — where Google’s plan runs on Gemini 3.5 Pro, now months behind in development, its launch delayed for falling short of internal goals.

The industry is full of these twists & turns. The freshest example: China’s Moonshot AI — dealt a blow a year ago by DeepSeek’s open-source progress, and yesterday back strongly with Kimi K3, the largest open source AI model out of China at 2.8 trillion parameters. Going toe to toe with the best of the closed US frontier models from Anthropic & OpenAI. Both would have been down 10% or more today, if they were already public (NOT STOCK ADVICE). Google has a similar opportunity for another turn back with Gemini — it’ll just take more time. In the meantime, Google remains best positioned globally on consumer AI, side by side with Apple.

Sources, in narrative order: BloombergGoogle Gemini Launch Delayed as Tech Falls Short of Internal Goals. For longtime readers: ‘OpenAI’s “Code Red” vs Google Gemini’ in AI-RTZ #924; and ‘Google takes a victory lap with Gemini 3’ in AI-RTZ #911.


(2) Apple Twists Differently — Apple Intelligence Approved in China with Alibaba and Baidu

MP TAKE: It’s good to see Apple navigating the twists and turns of US-China geopolitics — now delivering Siri AI and Apple Intelligence in China on its iPhones soon, through local partners, just as it does with Google Gemini and Nvidia in the US. Apple Intelligence has been approved for launch in China with Alibaba (Qwen) and Baidu — and Alibaba and Baidu shares jumped in Hong Kong on the partnership.

The moves with Alibaba and Baidu now — and potentially ByteDance, DeepSeek and others later — also highlight the open-source focus on AI models, especially smaller ones that can run on-device, as well as on Apple’s privacy-protected Apple Intelligence servers. Apple continues to execute on its AI opportunities with alacrity, worldwide.

Sources, in narrative order: TechCrunchApple Intelligence Approved for Launch in China with Alibaba’s Qwen AI. CNBCAlibaba and Baidu Shares Jump in Hong Kong on Apple AI Partnership. South China Morning PostChina Approves Apple Intelligence for iPhones, with Alibaba, Baidu Emerging as Partners. For longtime readers: ‘Apple Intelligence & Siri AI hum with “Google & Nvidia” inside’ (in the US) in AI-RTZ #1113; ‘US hamstrung on memory chip supply via China’ in AI-RTZ #1128; and ‘How Nvidia and Apple can be the Global, US Open Source AI Champions vs China’ in AI-RTZ #1089.


(3) The Internet Keeps Getting Rebuilt for AI — DoorDash’s Command-Line Ordering

MP TAKE: Another example of the internet being rebuilt, piece by piece, for AI agentsno more blanket ‘no’s to bots. This follows yesterday’s ARD #120 discussion on 1Password accommodating AI agents without revealing passwords. Today’s example: DoorDash now lets you order from the command line — and lets AI agents place real food orders.

A bit funny, yes — but it points to exactly the kind of experiments needed to find a better way for agents to interact with e-commerce sites in particular, and to talk to them in a ‘machine-to-machine’ (m2m) manner. Every small technical step — experiments included — counts in these early days of the AI Tech Wave. This unexpected CLI (command-line interface) twist by DoorDash is a perfect case in point of the turns ahead.

Sources, in narrative order: TechCrunchYes, You Can Now Order DoorDash from the Command Line. Yahoo! FinanceDoorDash to Let AI Agents Place Real Food Orders. For longtime readers: ‘Re-aligning Incentives in an AI-Agent-driven world’ in AI-RTZ #669; ‘AI Agents increasingly need an internet of their own’ in AI-RTZ #1023; ‘A better internet for both AI and humans ahead’ in AI-RTZ #823; and ‘Not “AI”, but “AIs” — the coming of “m2m” AI Agents’ in AI-RTZ #347.


MP OVERALL TAKE

All three of today’s events highlight the accelerating twists and turns in the fortunes of tech companies in this AI Tech Wave — negative or positive, seemingly big or small. They all matter right now, because AI technologies and their applications are still so far from true global ‘product-market fit’ in both enterprise and consumer markets. A lot of the underlying technology is still being researched and invented — much less deployed and distributed at scale.

Yesterday’s launch of Kimi K3 — the 2.8-trillion-parameter open-source model from China’s Moonshot — is another twist coming fast from an unexpected quarter, and it has totally upset the global lead assumed by the two closed frontier labs, Anthropic and OpenAI. The Apple partnerships in China, Microsoft weighing DeepSeek’s open source for Copilot, and Google’s latest Gemini setback all show how US-China opportunities are getting bigger even as geopolitics tries to make them smaller. It’s not out of the question to see Google — or other US companies — work with China’s AI firms, potentially pushing both governments to thread the needle on their differences rather than balkanize AI further. It also means more US big-tech names — Microsoft, Google, Meta — may join Nvidia, Apple and Elon’s Tesla/SpaceXAI in more proactive measures on tech and AI in China. So take the ups and downs for each company with a bit of patience — on the low notes and the high. A lot more can twist and turn in this AI Tech Wave than we expect.


GADGET AI

‘RAMageddon’s’ 8GB Squeeze Hits Windows PCs Far Harder Than Apple

MP TAKE — Gadget AI (take-first, Ep 103+)

Apple remains better positioned than Microsoft Windows PCs and Google Android devices at leveraging lower-RAM machines — The Verge found that even Microsoft couldn’t make Windows 11 run well on 8GB of RAM. The key difference is Apple’s hardware-and-software integration around its ‘Unified Memory’ architecture — which pools the various types of memory on the device together with the CPUs and the GPU/neural chips. Windows PCs and Android phones are well behind on that capability — and it’ll take a while for Microsoft, Google and their OEMs to catch up, right in the middle of an expanding global ‘RAMageddon.’

Sources, in narrative order: The VergeEven Microsoft Couldn’t Make Windows 11 Work Well on 8GB of RAM. For longtime readers: ‘”RAMageddon” really here to stay’ in AI-RTZ #1145; and ‘Apple’s MacBook Neo — a new arrow in its AI quiver’ in AI-RTZ #1017.


QUESTIONS

Q1 — What’s Michael’s current long-term view on the Apple MacBook Neo with its 8GB of RAM?

Answer: It’s more than adequate for most tasks — even regular AI activities. I still recommend it as the best value in laptops, even with the $100 price bump, now at $699. Comparable Windows laptops slow down under Windows 11, as The Verge’s review underscores.

Q2 — What’s the one thing Michael wishes the Neo had?

Answer: More RAM, obviously. But again, its absence isn’t felt as much, given the ‘Unified Memory’ architecture of the Mac OS. It remains ‘the little engine that could.’


WRAP

Today’s AI-RTZ #1150 — Europe’s DMA Rulings Set Tough Choices for Google and Apple — the EU’s Digital Markets Act rulings setting up tough choices for both companies over the next couple of years.

AI Ramblings Daily on AI-RTZ is here to think through AI and reset. Together.

Tomorrow — the Saturday Weekly Roundup, and Sunday’s The Bigger Picture.

Monday — back with ARD 122 on AI-RTZ 1154.

Have a great weekend.

Thanks for joining us today, AI Curious Folk. Stay tuned.

— MP


Full Source Reading —

For the broader context, see the canonical sources for ARD 121 — in today’s narrative order:

Event 1 — Google Falls Back Again in the AI Coding Race

Event 2 — Apple Twists Differently (Apple Intelligence Approved in China)

Event 3 — DoorDash’s Command-Line Ordering for AI Agents

Gadget AI — ‘RAMageddon’s’ 8GB Squeeze Hits Windows Harder Than Apple


Clips from today

Clip 1 — MacBook Neo: More Than Good Enough

Watch on YouTube Shorts

‘RAMageddon’ is squeezing 8GB laptops — but Apple’s $699 MacBook Neo (up $100) handles it. The Verge found even Microsoft couldn’t make Windows 11 run well on 8GB of RAM.

MP Take: Apple is better positioned at leveraging lower-RAM machines, thanks to its ‘Unified Memory’ architecture that pools memory with the CPU and GPU/neural chips. The Neo is more than adequate for most tasks, even AI — still the best value in laptops — while comparable Windows laptops slow down. It’s ‘the little engine that could.’

Clip 2 — DoorDash’s Command Line Ordering

Watch on YouTube Shorts

DoorDash now lets you order from the command line — and lets AI agents place real food orders. Another brick in rebuilding the internet for AI agents.

MP Take: No more blanket ‘no’s to bots. This follows yesterday’s 1Password accommodation. A bit funny, but it points to the kind of experiments needed for agents to interact with e-commerce sites machine-to-machine (m2m). Every small technical step counts in these early days of the AI Tech Wave — a perfect case in point of the turns ahead.

Clip 3 — US Tech Giants Lobby Against a China Open-Source Ban

Watch on YouTube Shorts

Not just Nvidia, Apple and Tesla — now Microsoft, Google, Meta and Amazon are also lobbying Washington not to block China’s open-source AI for geopolitical reasons.

MP Take: Competition from China is actually expanding the market — it makes the US closed-model companies better, because they’ll have to compete, and develop their models differently, faster. Competition is the essence of American capitalism, and how America has led in technology for 50-plus years. The AI Tech Wave is the biggest of them all; stepping back over political concerns is the wrong move.

Clip 4 — US Big Tech’s China Opportunity

Watch on YouTube Shorts

The perceived US-China AI gap may be months now, not years — and that’s an opportunity for US big tech to do more with companies in China. Moonshot’s Kimi K3, Alibaba’s Qwen and DeepSeek are all coming on strong.

MP Take: With Apple’s China approval and China’s open models catching up, US-China opportunities are getting bigger even as geopolitics tries to shrink them. It’s not out of the question to see Google or others work with China’s AI firms — pushing both governments to thread the needle rather than balkanize AI further. Take the ups and downs with patience.


About AI Ramblings Daily (ARD), and AI-RTZ

Both are daily. Both are free. Both are about AI. But they’re different mediums carrying different messages.

AI-RTZ is the morning text — a deeper written take on one idea, published by at least 5 AM EST. Today: post #1150.

AI Ramblings Daily is the afternoon video + podcast — my ad hoc takes and perspective on the day’s AI issues & news flow, around 20 minutes, with short 1-2 minute clips for quick topic views. Today: episode #121.

Subscribe to either or both on michaelparekh.substack.com. They run as separate Sections you can opt into or out of.


Links used in today’s show (already embedded inline above; listed here for reference)

Take 1 — Google Falls Back Again in the AI Coding Race:

Take 2 — Apple Twists Differently (Apple Intelligence Approved in China):

Take 3 — DoorDash’s Command-Line Ordering for AI Agents:

Gadget AI — ‘RAMageddon’s’ 8GB Squeeze Hits Windows Harder Than Apple:

Companion text:


(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here.)

Subscribe to AI: Reset to Zero for daily AI Ramblings + Sunday Bigger Picture posts.





Want the latest?

Sign up for Michael Parekh's Newsletter below:


Subscribe Here