Ed Zitron
@edzitron
Followers
94K
Following
237K
Media
7K
Statuses
39K
Newsletter https://t.co/D5qDgUKaNR - Better Offline Podcast - https://t.co/pUoGsuaQTw - Column Business Insider - CEO at https://t.co/5idt8AyPqr - Award-Winning Tech PR
Beautful Las Vegas Nevada
Joined December 2008
A 14.5k Premium Newsletter - AI Bubble 2027, my extrapolations of what may happen in the next 18 months, conditions that might accelerate a collapse, and how OpenAI and Anthropic overstate user numbers and revenues while hiding their awful burnrates.
wheresyoured.at
Soundtrack: The Dillinger Escape Plan - One Of Us Is The Killer An MIT study found that 95% of organizations are getting "zero return" from generative AI, seemingly every major outlet is now writing...
18
30
196
RT @edzitron: A 14.5k Premium Newsletter - AI Bubble 2027, my extrapolations of what may happen in the next 18 months, conditions that migh….
wheresyoured.at
Soundtrack: The Dillinger Escape Plan - One Of Us Is The Killer An MIT study found that 95% of organizations are getting "zero return" from generative AI, seemingly every major outlet is now writing...
0
30
0
Gas leak?.
THE NEW MAG 7 OF AGENTIC AI. Stage one of AI was all about hardware -- servers, GPUs, and the scaffolding. Stage two is applications: embedding intelligence directly into the workflows, infrastructures, and physical systems that run trillion-dollar economies. This is where true
9
8
412
NVIDIA missed on data center revenue estimates for the second quarter running, people ignored it last Q so who knows
7
32
321
RT @edzitron: Newsletter: My 16,000 word opus - How To Argue With An AI Booster, a comprehensive guide to arguing with AI boosters, address….
wheresyoured.at
Editor's Note: For those of you reading via email, I recommend opening this in a browser so you can use the Table of Contents. This is my longest newsletter - a 16,000-word-long opus - and if you...
0
74
0
This piece also includes a beautiful table of contents so you can use it as a reference! .
wheresyoured.at
Editor's Note: For those of you reading via email, I recommend opening this in a browser so you can use the Table of Contents. This is my longest newsletter - a 16,000-word-long opus - and if you...
0
1
19
Newsletter: My 16,000 word opus - How To Argue With An AI Booster, a comprehensive guide to arguing with AI boosters, addressing both their bad faith debate style and their specific (and flimsy) arguments as to why generative AI is the future.
wheresyoured.at
Editor's Note: For those of you reading via email, I recommend opening this in a browser so you can use the Table of Contents. This is my longest newsletter - a 16,000-word-long opus - and if you...
15
74
419
RT @edzitron: bonus ep of Better Offline. I report on how OpenAI’s new "router-based" ChatGPT-5 makes it impossible for OpenAI to cache the….
podcasts.apple.com
Podcast Episode · Better Offline · 08/22/2025 · 29m
0
26
0
bonus ep of Better Offline. I report on how OpenAI’s new "router-based" ChatGPT-5 makes it impossible for OpenAI to cache the static prompt for any model or tool it uses every single prompt, doubling token burn for mediocre gains.
podcasts.apple.com
Podcast Episode · Better Offline · 08/22/2025 · 29m
2
26
160
Steve Burke is one of the best tech journalists of all time, incredible piece.
THE NVIDIA AI GPU BLACK MARKET:
1
9
177
RT @edzitron: Premium newsletter: I have the details on how GPT-5 works. OpenAI's new "router-based" ChatGPT-5 makes it impossible for the….
wheresyoured.at
Welcome to another premium edition of Where's Your Ed At! Please subscribe to it so I can continue to drink 80 Diet Cokes a day. Email me at [email protected] with the subject "premium" if you...
0
43
0
With GPT-5, OpenAI created massive amounts of overhead as a means of juicing more performance out of a model they *had* to release to the market. I'm not sure what they do next.
1
5
56
From what I've been told, ChatGPT-5 splits user prompts into two "legs" where one LLM orchestrates the output, which can mean multiple different model/tool calls, each with their own instructions, every single prompt. OpenAI gave ChatGPT a middle manager.
1
2
54
Premium newsletter: I have the details on how GPT-5 works. OpenAI's new "router-based" ChatGPT-5 makes it impossible for the company to cache the static prompt for any model or tool it uses every single prompt, doubling token burn for mediocre gains.
wheresyoured.at
Welcome to another premium edition of Where's Your Ed At! Please subscribe to it so I can continue to drink 80 Diet Cokes a day. Email me at [email protected] with the subject "premium" if you...
11
43
319
Sure sounds like they're thinking about rate limiting on API!
Looks like OpenAI has opened up "priority processing" to all API customers with the launch of GPT-5 after making it enterprise when they launched in mid-June. Have to wonder if they're gonna start rate limiting/throttling! .
15
22
469
RT @edzitron: Premium Newsletter: The enshittification of AI has begun. With GPT-5, OpenAI is cutting all other models, reducing the value….
wheresyoured.at
Thanks for subscribing to Where’s Your Ed At Premium, please shoot me an email at [email protected] if you ever have any questions. Yesterday, OpenAI launched GPT-5, a new “flagship” model of some...
0
64
0
