Chainfeeds Summary:
The AI supercycle will last 15 years, and we are only in the third year now.
Article source:
https://x.com/cryptorand/status/2052795669426172219
Article Author:
Rand Group
Opinion:
Rand Group: Phase One is complete (2023–2025), with the infrastructure layer finished. AMD rose 78% in 2025, NVDA rose 39%, and INTC just delivered an extremely strong Q1 earnings report, pushing the Philadelphia Semiconductor Index above 10,000 points for the first time. Chips will continue to support every phase of the AI cycle, but the "generational" low-entry opportunities have disappeared, and the risk-reward ratio has been compressed. Phase Two is the peak of infrastructure construction (2025–2027), a phase that most investors are just beginning to realize. CEG acquired Calpine, becoming the largest privately held power producer in the US with a total installed capacity of 55GW. GEV rose more than 200% in one year. VRT is co-developing a cooling system for NVIDIA's Rubin architecture. GLW has risen 74% this year due to increased demand for fiber optics. The real breakout is small modular nuclear reactors (SMRs), including OKLO, SMR, and BWXT. These companies are deploying to directly power data centers. There is still room for growth in this phase, but the most obvious targets have already been identified by the market. The third phase is the deployment window (2026–2028), which is the stage where AI moves from the data center to the physical world. Most people will be late. TSLA is converting its Fremont factory to produce Optimus robots, with capital expenditures reaching $25 billion, aiming to start mass production in the second half of 2026. RKLB just set a record of $602 million in revenue, with an order backlog of $1.85 billion. LUNR is up 47% this year, with contract value reaching $943 million. KTOS' Valkyrie drone has been selected by the US Marine Corps. And now is the deployment window. The fourth phase is the final frontier (2028+), which is the endgame. Microsoft's capital expenditures reach $190 billion. Google $190 billion. Amazon $200 billion. Meta $145 billion. Google Cloud's backlog has exceeded $460 billion. They are building the infrastructure for AI software dominance and AGI. Quantum computing is still in its early stages, but IONQ and D-Wave are already laying the groundwork. Ultimately, the platform controlling the software layer will win the entire supercycle.
Content source 
