- The AI Pulse
- Posts
- 🤖 How Fast Do AI Chips Lose Value?
🤖 How Fast Do AI Chips Lose Value?
PLUS: 500 Global CEOs Share Their Biggest AI Confession
James Roesser
November 25, 2025
Welcome back AI enthusiasts!
In today’s Daily Report:
📉How Fast Do AI Chips Lose Value?
🤐500 Global CEOs Share Their Biggest AI Confession
🛠Trending Tools
🥪Brief Bites
💰Funding Frontlines
💼Who’s Hiring?
Read Time: 3 minutes
🗞RECENT NEWS
AI INFRASTRUCTURE
📉How Fast Do AI Chips Lose Value?
Image Source: Reve Image/Free Online Next-Gen AI Image Generator
As Big Tech companies prepare to spend $4 trillion on AI infrastructure over the next five years, a crucial accounting dilemma looms: “What’s the life expectancy of a GPU?”
Key Details:
Best known for predicting and profiting from the 2008 financial crisis, “The Big Short” hedge fund manager Michael Burry recently went viral for allocating 82.75% of the $1.1 billion he manages to bet against NVIDIA. Shortly after placing these bets, he announced he was shutting down Scion Asset Management. Although he no longer holds a financial position against NVIDIA, his stance remains unchanged.
He believes Big Tech companies are extending the useful life of GPUs to artificially inflate reported earnings. In other words, these AI chips have a much shorter lifespan than any of the Magnificent 7 claim.
Simply put, the big worry centers around GPUs, which are essential for training and running advanced AI models. As newer GPUs come out, older GPUs gradually fall behind in performance. As a result, Big Tech companies must reduce the value of older GPUs over a realistic time frame. The faster older GPUs depreciate, the larger the impact on reported earnings.
Why It’s Important:
Burry estimates the true useful life of these AI chips is likely two years, not the six years many Big Tech companies claim. That said, his estimates overlook the critical distinction between different compute lifecycles.
Big Tech companies operate on a “value cascade”: GPUs have an extended useful life because they transition through multiple stages of compute. For example, Training is the process of teaching an AI Model, which requires newer GPUs. AI Inference is the process of using an AI Model, which only requires older GPUs.
AI RESEARCH
🤐500 Global CEOs Share Their Biggest AI Confession
Image Source: Canva’s AI Image Generators/Magic Media
According to a recent poll of 500 global CEOs, 94% believe AI could offer better counsel than at least one of their human board members. So, are they right?!
Key Details:
Wharton’s Mack Institute and INSEAD’s Center for Corporate Governance ran an experiment comparing the performance of an “AI Board” with “Human Boards” on the same corporate scenario.
Wharton’s Mack Institute created the “AI Board,” which was made up of multiple AI Agents: proactive personal assistants that autonomously analyze, execute, and fulfill on your behalf. The AI Agents were equipped with memory to store evolving goals and track ongoing deliberations.
INSEAD’s Center for Corporate Governance divided human participants into six groups to form multiple “Human Boards” for a fictional company called Fotin. Each group received a board book with the meeting agenda and detailed descriptions of their individual executive roles.
Why It’s Important:
Transcripts from the “AI Board” and the “Human Boards” were blindly evaluated against eight governance criteria by three LLMs and three independent experts using a three-point scale: poor, satisfactory, and good.
The “Human Boards” scored significantly lower across all eight governance criteria. For example, when implementing decisions, human participants rarely translated spoken strategy into concrete performance targets because discussions often revolved around broad statements.
THE STOCK MARKET
📍AI Tier Tracker
TIER 0: ENERGY Fluence Energy, Inc. |
|---|
TIER 1: SILICON Micron Technology, Inc. |
TIER 2: DATA CENTERS Digital Realty Trust, Inc. |
TIER 3: AI MODELS NVIDIA Corp. |
TIER 4: SOFTWARE STACK Exela Technologies, Inc. |
TIER 5: AI AGENTS NICE Systems Ltd. |
🔔CLOSING BELL: As of 11/24/2025 market close.
💡STOCK SPOTLIGHT: Each tier showcases a new stock every day.
🛠TRENDING TOOLS
🐋BestStock AI delivers data-driven investing strategies.
✨SafeNew converts AI-generated text to human-like text.
🎨Cartoonize AI transforms pets and people into cartoons.
📊Excelmatic makes sense of complex Excel spreadsheets.
🖋️RightBlogger helps you create and promote your content.
🧰 Browse our Always Up-To-Date AI Toolkit.
🔩BOLT BOOST: 3 new AI tools are added to the AI toolkit every day.
🥪BRIEF BITES
The Trump Administration launched “The Genesis Mission,” a coordinated federal effort to unleash a new age of AI-accelerated innovation that strengthens national security.
MIT LIDS developed “VGGT-SLAM,” enabling emergency-response robots to quickly, efficiently, and accurately generate terrain layouts for navigating hazardous environments.
Anthropic released “Claude Opus 4.5,” the world’s most intelligent frontier AI model for coding, which achieved 80.9% accuracy when solving real-world software issues sourced from GitHub.
OpenAI CEO Sam Altman warned of “Economic Headwinds,” according to a leaked internal memo where he mentioned that Google’s pre-training breakthroughs put significant short-term competitive pressure on OpenAI.
💰FUNDING FRONTLINES
Momentic raised a $15M Series A to automate software testing.
Wispr closed a $25M Series A Extension to don’t type, just speak.
Function Health landed a $298M Series B to help you own your health.
💼WHO’S HIRING?
📒FINAL NOTE
FEEDBACK
How would you rate today’s email?It helps us improve the content for you! |
❤️TAIP Review of The Day
“Good pros and cons.”
REFER & EARN
🎉Your Friends Learn, You Earn!
You currently have 0 referrals, only 1 away from receiving 🎓3 Simple Steps to Turn ChatGPT Into an Instant Expert.
Share your unique referral link: https://theaipulse.beehiiv.com/subscribe?ref=PLACEHOLDER