Saturday, May 23, 2026

𝟭𝟬 𝗳𝗿𝗲𝗲 𝗔𝗜 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀

𝗛𝗲𝗿𝗲 𝗮𝗿𝗲 𝘁𝗵𝗲 𝘁𝗼𝗽 𝟭𝟬 𝗳𝗿𝗲𝗲 𝗔𝗜 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀 𝗳𝗿𝗼𝗺 𝗶𝗻𝗱𝘂𝘀𝘁𝗿𝘆 𝗹𝗲𝗮𝗱𝗲𝗿𝘀: 

𝟭 - 𝗔𝗻𝘁𝗵𝗿𝗼𝗽𝗶𝗰:
https://lnkd.in/e5fK7QUA

𝟮 - 𝗚𝗼𝗼𝗴𝗹𝗲:
http://grow.google/ai

𝟯 - 𝗠𝗲𝘁𝗮:
https://lnkd.in/et6wz-ta

𝟰 - 𝗡𝗩𝗜𝗗𝗜𝗔:
https://lnkd.in/e8aHmFxc (GOATed)

𝟱 - 𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁:
https://lnkd.in/ej85NeZc

𝟲 - 𝗢𝗽𝗲𝗻𝗔𝗜:
http://academy.openai.com

𝟳 - 𝗜𝗕𝗠:
http://skillsbuild.org

𝟴 - 𝗔𝗪𝗦:
http://skillbuilder.aws

𝟵 - 𝗗𝗲𝗲𝗽𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴𝗔𝗜:
http://deeplearning.ai

𝟭𝟬 - 𝗛𝘂𝗴𝗴𝗶𝗻𝗴 𝗙𝗮𝗰𝗲:
https://lnkd.in/eajVCrvv



Node.js internals book

Node.js Internals, Runtime & Networking Guide | NodeBook

NodeBook explains what happens inside the runtime when your application handles I/O, schedules work, allocates memory, and serves traffic.



SpaceX(AI) IPO, EWS, Starship V3

Tracking the Singularity: Week of May 21st @metatrends

SpaceX filed for a $75 billion raise at a valuation north of $1.75 trillion, 2.5 times larger than Saudi Aramco’s record. Elon maintains super voting rights with insiders controlling 86% of voting power.

The prospectus claims a total addressable market of $28.5 trillion. That breaks down to $870 billion for Starlink broadband, $740 billion for Starlink mobile, $600 billion for digital advertising through X, $2.4 trillion for AI infrastructure, and a staggering $22.7 trillion from Macrohard, their partnership with Tesla to replicate all digital work with AI.

spacex ipo prospectus summary - Google Search

SpaceX filed its S-1 prospectus with the SEC, targeting a blockbuster IPO valuation between $1.75 trillion and raising roughly \(\$75\) billion. The offering will feature a dual-class share structure, keeping absolute voting control with Elon Musk while exposing the newly merged tech-and-space conglomerate's financials and sci-fi ambitions to public markets. [1, 2, 3]

Key Prospectus Takeaways
1. Financial Health & xAI Drag
  • Revenue: SpaceX generated $18.7 billion in 2025.
  • Net Losses: The company recorded a net loss of $4.9 billion.
  • The AI Factor: Losses are heavily driven by SpaceX's recent merger with xAI, which lost $6.4 billion in 2025. Without the AI division, SpaceX’s core connectivity and aerospace divisions are comfortably profitable. [1]
2. Major Commercial Contracts
  • The prospectus outlines a massive Cloud Services Agreement with Anthropic. Anthropic will pay SpaceX \(\$1.25\) billion per month through May 2029 to use compute capacity from SpaceX's "Colossus" and "Colossus II" data center clusters. [1, 2]
3. Ambitious Addressable Market
  • SpaceX defines an ultimate total addressable market (TAM) of \(\$28.5\) trillion, spanning space broadband, AI infrastructure, and broader enterprise applications. [1]
Corporate Governance & Executive Pay
  • Super-Voting Stock: The IPO introduces a dual-class structure. Public investors will purchase Class A shares (1 vote per share), while insiders led by Musk hold Class B shares (10 votes per share).
  • Mars Compensation Milestone: Musk was granted a restrictive stock award of 1 billion shares, which fully vests only if SpaceX achieves a \(\$7.5\) trillion market capitalization and successfully establishes a permanent human colony on Mars with at least 1 million inhabitants. [1, 2, 3]
Core Business Units
  • Connectivity (Starlink): The financial backbone of the empire. As of early 2026, Starlink operates over 9,600 satellites, servicing 10.3 million subscribers. In 2025, this unit generated \(\$11.39\) billion in revenue with \(\$4.42\) billion in operating income.
  • Aerospace & Space AI Compute: The company plans to lay the groundwork for a lunar economy, asteroid mining, and orbital AI compute satellites by 2028. [1, 2, 3, 4]

The video provides an overview of the SpaceX S-1 filing, which details the company's objective to raise $75 billion at a $1.75 trillion valuation (45:19 - 45:38). Key details include:

  • Business Units: The company is divided into three main pillars: Starlink, which is described as a "money printer" with $11.4 billion in revenue and 50% growth; a space business generating $4 billion; and an AI division (45:47 - 46:27).
  • Elon Web Services (EWS): A significant development mentioned is a $45 billion deal with Anthropic to rent compute capacity on Colossus clusters, amounting to $15 billion annually (46:47 - 47:08).
  • Infrastructure Efficiency: The filing highlights SpaceX's ability to build data centers significantly faster than competitors, with build times dropping from 122 days to 66 days for recent projects (48:17 - 48:33).
  • IPO Timing: The listing is expected to occur in mid-June, with the ticker set as SPCX (45:38 - 45:41).
“$28.5 TRILLION” - Musk’s SpaceX IPO Could Break EVERY Record In History - YouTube Valuetainment









AI HW: $95B IPO Cerebras vs Groq

Cerebras and Groq are both cutting-edge semiconductor startups challenging Nvidia by focusing on ultra-fast AI inference rather than generalized processing. Cerebras uses massive, wafer-scale chips to deliver record-breaking token throughput for massive models, while Groq uses smaller Language Processing Units (LPUs) optimized for extremely low latency. [1, 2, 3]
For an in-depth breakdown of how the hardware architectures and specific advantages of each chip stack up, watch this video:


Key Differences
Feature [1, 2, 3, 4, 5]CerebrasGroq
ArchitectureWafer-Scale Engine (WSE): One massive chip that keeps models completely on-chip with zero off-chip memory access.Language Processing Unit (LPU): Custom deterministic architecture with 230MB of on-chip SRAM, scaling via proprietary fabric.
Model Size HandlingExcellent. A single Cerebras device can hold multi-billion parameter models in fast SRAM, reducing complexity and points of failure.Smaller capacity per LPU. Large models require hundreds of networked chips, introducing complex clusters.
Inference SpeedGroundbreaking. Frequently sets industry records for highest token throughput (e.g., hundreds to thousands of tokens per second).Industry-leading latency. Highly deterministic with steady output, ideal for real-time voice, robotics, and edge applications.
Primary FocusHeavy enterprise inference, high-performance computing (HPC), and large-scale model deployment.Latency-sensitive inference, cloud compute (GroqCloud), and immediate edge/real-time processing.
AccessibilityAvailable on-premises or via inference APIs (Meta, Hugging Face, OpenRouter, Vercel).Available via GroqCloud, on-prem (GroqRack), and through API ecosystems like Hugging Face.

For more specific details and pricing considerations, review the official Cerebras CS-3 vs Groq LPU comparison and test out models directly on GroqCloud.


Cerebras Systems - Wikipedia


AI chipmaker Cerebras Systems completed one of the largest U.S. tech IPOs in history. Trading on the Nasdaq under the ticker CBRS, the stock priced at \(\$185\) per share and opened its first day of trading at $350. [1, 2, 3]

Key Details of the Market Debut:
  • Date: May 14, 2026
  • Ticker: CBRS (Nasdaq)
  • Initial Pricing: $185 per share (above the initially marketed $115-$125 range)
  • First-Day Action: Surged up to \(\$385\) intraday and closed up 68% at \(\$311.07\), raising \(\$5.55\) billion.
  • Market Capitalization: Reached approximately $67 billion at the close of its first day. [1, 2, 3, 4, 5, 6]
What You Should Know:
  • Business Model: Cerebras specializes in massive, wafer-scale chips designed to speed up AI model training and inference.
  • Key Customers: The company has recently diversified its revenue through massive infrastructure and deployment deals with major players like Amazon and OpenAI.
  • Valuation: The stock has drawn intense institutional demand, heavily oversubscribed ahead of its launch, though some analysts caution that it is trading at a significant premium based on current revenue. [1, 2, 3, 4, 5]

On May 14, Cerebras opened on the Nasdaq at $350 a share, peaked at $386, and closed at $311, valuing the AI chipmaker at ~$95 billion on day one. The company priced its IPO at $185 the night before, which was already a twice-upsized range, and raised $5.55 billion. It is the largest US tech IPO since Snowflake’s $3.8 billion debut in 2020. The stock ended the week at around $278.

The video features a deep dive into Cerebras Systems and their industry-defining work in AI hardware, led by co-founder and CEO Andrew Feldman. Here are the key points regarding the company's trajectory and technology:

  • Record-Breaking IPO: The company recently achieved a major milestone with a 95 billion** 0:43 - 0:46, 1:33:00 - 1:33:45.
  • Wafer-Scale Computing: Cerebras distinguishes itself from standard GPU manufacturers by pioneering wafer-scale computing. Instead of using individual chips, they utilize an entire silicon wafer to accelerate AI training and inference, aiming to solve performance bottlenecks in large-scale model development 1:33:25 - 1:34:34, 1:54:24 - 1:55:30.
  • Focus on Inference: While the company is well-known for training capabilities, they have aggressively pivoted to meet the massive demand for fast inference. Andrew Feldman emphasizes that for AI to be truly useful, it must be capable of high-speed, cost-effective inference 1:40:20 - 1:41:23.
  • Hardware Advantages: The company’s architecture involves significant innovations in lithography, cooling, power delivery, and compiler design. Because they use a massive, singular engine, they had to solve unique problems regarding fault tolerance—specifically the ability to shut down a core and route around flaws—which makes their technology exceptionally resilient 1:54:53 - 1:55:30, 2:04:19 - 2:05:03.
  • Performance Benchmarks: Andrew Feldman noted that their systems can run trillion-parameter models with significantly higher token-per-second output compared to standard high-end GPU clusters 1:43:55 - 1:44:23.