$1.7 Trillion IPO
April 03, 2026
A rocket company is heading to the stock market.
A database giant is cutting jobs.
A social media company is buying energy.
All because of AI.
-----------------------
SpaceX moves toward a record-breaking IPO
SpaceX has taken a major step toward going public, and it could be one of the biggest listings in history.
The company has confidentially filed for a U.S. IPO, with reports suggesting a valuation of up to $1.75 trillion and a potential raise of $50-75 billion, which would make it the largest IPO ever recorded.
The scale is backed by real numbers. SpaceX’s Starlink business alone has reportedly crossed 9 million subscribers, generating recurring revenue that underpins its valuation, while total company revenue is estimated to be nearing $20 billion annually.
But the bigger play is strategic. Following its $250 billion merger with xAI, SpaceX is positioning itself at the intersection of satellites, AI compute, and space-based infrastructure, including long-term plans for orbital data centers.
-----------------------
Oracle cuts jobs as it doubles down on AI
Oracle is laying off employees as it accelerates its shift toward AI and cloud, a move increasingly common across Big Tech.
The company hasn’t disclosed exact numbers, but the cuts come as AI spending surges. Training advanced models now routinely costs $100 million+, with estimates showing models like GPT-4 at $78 million and Gemini Ultra at $191 million in compute alone, according to Stanford University.
Oracle is leaning into this shift, expanding its cloud infrastructure and positioning itself against AWS, Microsoft Azure, and Google Cloud for AI training and deployment.
-----------------------
Meta’s AI expansion is turning into an energy race
Meta is quietly making one of its most strategic moves yet: locking in energy supply to power its AI ambitions.
The company is reportedly securing large-scale natural gas resources to support new data center projects, including developments that could significantly increase energy demand in states like South Dakota.
A single large AI data center can consume as much electricity as tens of thousands of homes, and global data center power usage is projected to double by the end of the decade as AI adoption accelerates.
Meta’s approach signals a long-term reality: access to energy is becoming just as critical as access to GPUs.
-----------------------
Glia wins as banks shift toward “safe AI”
Glia has won the 2026 Artificial Intelligence Excellence Award in the banking category, but the bigger story is why it stood out.
The company’s platform helps automate up to 80% of customer interactions while keeping humans in the loop, critical in a highly regulated industry. It also claims to offer the first contractual protection against AI hallucinations and prompt injection attacks in banking.
Glia now works with 700+ banks and credit unions, signaling strong demand for AI that can operate within compliance-heavy environments.
-----------------------
Worth Reading
AI models reason out loud so we can watch them think. But what happens when they start thinking in a language we can't read? Nicholas Andresen's piece on Chain-of-Thought, "Thinkish," and why our window into machine reasoning may not stay open much longer.
-----------------------
Also Read
Z.ai drops a multimodal coding model
GLM-5V Turbo combines vision and coding capabilities, built for agent-driven workflows and high-capacity task execution. View the official post here
Google AI Studio makes it faster to ship apps
Logan Kilpatrick shared updates that let users turn chats into apps in two clicks, alongside improved mobile usability.
Alexa+ expands into food ordering
Amazon is integrating Uber Eats and Grubhub, pushing Alexa deeper into real-world, transactional use cases.
AI in healthcare faces growing scrutiny
Researchers from Northeastern University highlight risks around bias, accountability, and ethical deployment in medical AI systems.
-----------------------
Top Tweet of the Week
What if ChatGPT is designed to keep you talking?
A post questions whether AI systems are subtly tuned to keep users hooked, not just informed, raising concerns about the incentives behind model design.
-----------------------
Worth a Try AI Tools
Capy
Capy is an AI-native IDE designed for parallel development, allowing multiple coding agents to run simultaneously while managing workflows from task to pull request. Free for Open Source Projects.
Why try it?
Useful for developers aiming to speed up execution across multiple workstreams.
Before You Go
The question the week leaves open: if energy and compute are the new moats, what happens to the companies with great models but no path to owning the infrastructure underneath them?
See you next week! 👋
🚧☠☠☠☠☠🚧