Zephyr here,
We have some absolutely massive AI news today.
NVIDIA just announced they're investing $100 billion in OpenAI to build the biggest AI infrastructure project in history.
Huawei's building supercomputers with over a million GPUs to compete with US tech.
And new reports show AI's electricity demands might not be as crazy as everyone thought.
Let's dive in.
Become an email marketing GURU.
Join us for the world’s largest FREE & VIRTUAL email marketing conference.
Two full days of email marketing tips & trends, famous keynote speakers (Nicole Kidman!), DJ’s, dance contests & networking opportunities.
Here are the details:
100% Free
25,000+ Marketers
November 6th & 7th
Don’t miss out! Spots are LIMITED!
1. NVIDIA Commits $100 Billion Investment in OpenAI's 10-Gigawatt Infrastructure
NVIDIA and OpenAI announced a landmark partnership to deploy at least 10 gigawatts of AI data centers powered by millions of NVIDIA GPUs, with NVIDIA investing up to $100 billion progressively as each gigawatt is deployed. The first gigawatt using NVIDIA's Vera Rubin platform will come online in late 2026, representing 4-5 million GPUs that will create infrastructure equivalent to a billion times more computational power than OpenAI's first server. OpenAI CEO Sam Altman called compute infrastructure "the basis for the economy of the future" as the company serves over 700 million weekly active users.
2. Huawei Unveils Million-GPU Supercomputers to Challenge US Platforms
Huawei announced its Atlas 950 and 960 SuperClusters designed to "circumvent the limitations in China's chip manufacturing process" with the Atlas 960 featuring over 1 million Ascend NPUs by 2027. The Atlas 950 SuperPoD with 8,192 Ascend NPU cards launches in Q4 2026 while the larger Atlas 960 SuperPoD with 15,488 cards arrives by end of 2027. Deputy chairman Eric Xu claims these will be the "world's most powerful" computing clusters, providing China with abundant computing power to compete directly with NVIDIA and xAI platforms despite US export controls.
3. AI Electricity Demand Projections May Be Overblown, New Analysis Shows
Despite alarming headlines, data centers currently consume only 1-2% of global electricity, expected to reach just 3% by 2030 according to the IEA's latest analysis. The energy sector has historically overestimated tech power demands, like when EPA predicted data center energy would double from 2005-2010 but actual increase was only 36%. While AI does use more power (ChatGPT query needs 2.9 watt-hours versus Google search's 0.3), efficiency improvements and the fact that NVIDIA can't produce servers fast enough naturally limit growth to manageable levels.
4. Google's Gemini AI Arrives on 300 Million TV Devices
Google is bringing Gemini AI to Google TV, enabling natural language conversations on over 300 million active devices including TCL, Hisense, and Walmart TV models. Users can ask Gemini to help find shows based on shared interests, catch up on past seasons, get homework help, plan vacations, or learn new skills directly through their TV. The rollout begins today on TCL QM9K series TVs with broader availability coming later this year to Google TV Streamer and 2025 models from multiple manufacturers.
5. Facebook Dating Gets AI Assistant to Combat Swipe Fatigue
Facebook launched two AI features for its dating app: a "dating assistant" chatbot that finds matches based on interests like "Find me a Brooklyn girl in tech" and helps improve profiles. The second feature, "Meet Cute," uses a personalized algorithm to deliver weekly surprise match suggestions based on compatibility assessments, though Meta hasn't explained how the algorithm evaluates potential dates. Both features are rolling out to users in the US and Canada as Facebook tries to differentiate its dating platform through AI-powered matching rather than endless swiping.
6. OpenAI Expands Budget ChatGPT Go Plan to Indonesia at $4.50/Month
OpenAI launched its ChatGPT Go subscription in Indonesia for $4.50 monthly, following its successful India launch that doubled paid subscribers within a month. The mid-tier plan offers 10 times higher usage limits than free ChatGPT for prompts, image generation, and file uploads, plus better conversation memory for personalized responses. This directly competes with Google's similarly priced AI Plus plan in Indonesia which includes Gemini 2.5 Pro, creative tools, and integration with Gmail and Docs plus 200GB cloud storage.
Want to Get More Done Faster?
You read about AI every day here.
But you're still moving at the same pace as everyone else.
I just launched Zephyr Elite - where I teach you to get more done faster with AI and automation.
Master AI prompts that deliver exactly what you need. Automate any repetitive task with Make and n8n. Build workflows that work while you sleep.
Get 20+ hours back every week while your competitors stay stuck in slow mode.
How would you rate today's newsletter?
Stay tuned for more updates, and have a fantastic day!
Cheers,
Zephyr
