Don’t let your laptop tell you you’re doing too much. With #IntelCoreUltra# Series 3 you can create and edit and record and upload and stream and game, all in one thin, light laptop. #AICompute# #Processor# #CPU#

AI infrastructure is scaling globally. Hyperscalers (Microsoft, Amazon, Google, Meta and others) are on track for roughly $765 billion in annual AI-related CapEx in 2026 alone, with cumulative AI data center capital expenditures projected to reach $5.2 trillion by 2030 in the base case (and up to $7.9 trillion in accelerated scenarios), according to McKinsey (March 2026). The global AI data center market itself is expected to grow from $147 billion in 2025 to $811 billion by 2033 at a CAGR of 23.9%, per Grand View Research. Meanwhile, global data center electricity consumption hit ~485 TWh in 2025 (up 17% YoY) and is projected to roughly double to ~950 TWh by 2030, with AI-focused facilities growing even faster (IEA, April 2026 report). The financial layer around it is still early. Despite these trillions in required capital, the entire tokenized Real World Assets (RWA) market (excluding stablecoins) stands at only ~$30–37.5 billion as of May 2026 — still tiny relative to the physical buildout and overwhelmingly dominated by traditional assets like Treasuries and private credit rather than AI compute, energy, or data centers ( and market reports, May 2026). As more compute, energy, and data infrastructure come online, new models for access, coordination, and capital formation may emerge on-chain. Rax Finance is exploring this direction by building a full-stack on-chain registry, metering, and settlement layer that tokenizes GPU capacity, data center power, and energy resources into verifiable, insured, yield-bearing RWAs — turning physical AI infrastructure into globally accessible, programmable assets. What are your thoughts on the future of AI infrastructure RWAs, Rax Fam? Would love to hear your ideas below 👇

The AI compute scaling showcased at #GTC# is incredible. But in Web3 trading, an AI is only as powerful as the data pipeline feeding it. We built Derivio’s proprietary Data Stream Engine to solve this exact bottleneck—delivering real-time, high-fidelity on-chain data for the Agentic era. ⚡️🛠️

Ex-Google CEO Eric Schmidt says AI may hit a money wall before it hits a power wall. Everyone is talking about energy constraints. Schmidt thinks they're focused on the wrong bottleneck. Here's the math he laid out: Roughly $50 billion per gigawatt of AI compute. 10 gigawatts of capacity costs half a trillion dollars. To seriously scale the infrastructure this industry needs, you are talking about deploying a trillion dollars or more into a single sector. "How many companies, countries, and so forth can hand an industry a trillion dollars of capital? Very, very few." China could do it. He's not sure if they are but they have the state capacity to try. Europe can't. Their capital markets aren't built for it. America can. And that's the edge. The brilliance of US capital markets is the ability to borrow and deploy that kind of money faster than any other system in the world. That structural advantage is what allows the AI buildout to happen here at a scale that other countries simply cannot match. "The Europeans can't do this, which they're sort of sore about. But this is good for America."

Q1 2026 Shareholder Update We continued to make meaningful progress on the build out of the infrastructure & AI software that underpins our Robotaxi & future robotics businesses in Q1. That meant commencing the ramp of new factories across AI compute, battery & battery materials, as well as preparing lines for start of production of Megapack 3, Cybercab & Tesla Semi. Demand for our vehicles continued to grow in APAC & South America markets, with a rebound of demand in EMEA markets & North America. As trade and geopolitics become more uncertain, we're further regionalizing and vertically integrating critical supply chains to ensure access to key materials & componentry in each region across vehicle, energy & AI. Automotive – Optimizing our vehicle product portfolio with an emphasis on vehicles designed for a fully autonomous future – More affordable trims of Model 3/Y & rollout of Model Y L in markets outside of China – Began deliveries of Cybertruck in the UAE – Volume production of Cybercab & Tesla Semi this year Energy Generation & Storage – Good progress with new Megafactory outside Houston (will produce Megapack 3 for Megablock). Start of production on track for later this year – We began meaningful customer deployments of Tesla’s first in-house designed solar panel produced at Giga New York Robotics – Preparations for our first large-scale Optimus factory will begin shortly in Q2. First-gen line designed for 1M robots/year will replace Model S/X lines in Fremont Factory, second-gen line is being prepared at Giga Texas (long-term annual capacity of 10M robots/year) AI Training Compute – Cortex 2 is now online & has started running training workloads – Also ramping on-site training infrastructure to ensure sufficient compute resources for AI products & services – Continuing with custom silicon development (Dojo 3) to reduce training cost over time Battery – Ramping new battery & material factories, including LFP cells in Nevada, cathode material & lithium refining in Texas – Battery vendor cell availability continues to be a limiting factor on ramping vehicle production, so we're working on initiatives to de-bottleneck, including using 4680 cells at Giga Berlin Other Supporting Infrastructure – Giga New York is now producing V4 Supercharging cabinets (3x power density & 2x the number of stalls vs V3) – Alongside the ramp of Tesla Semi, we're deploying public Megachargers, including our first one in SoCal – Over 2,200 new Supercharger stalls, growing the network 19% YoY AI Software – FSD 14.3 launched in April – Upgraded Reinforcement Learning (RL) stage to better handle long-tail edge cases, enhanced the neural network vision encoder for sharper perception in low-vis scenarios & rewrote the AI compiler to accelerate model iterations & cut inference latency by 20% (faster reaction time for FSD!) This accelerates our efforts to eventually deploy unsupervised autonomy to both the Robotaxi fleet & customer owned vehicles – Digital Optimus: our next evolution of AI development. We're working on automating digital workloads, building an intelligence layer that will complement real-world AI in vehicles & robots AI Inference Compute – Expanding our scope of manufacturing to include semiconductor fabrication (coinciding with Robotaxi & Optimus ramps) = step towards ensuring sufficient & resilient chip supply – Partnership with SpaceX aims to build the largest chip fab ever, vertically integrating logic, memory & advanced packaging to allow for rapid iteration – Completed final chip design of AI5 (our next-gen inference processor) in April Automotive & Other Software – Rolled out Spring Update which includes a new Self-Driving app with tutorials & stats, "Hey Grok" wake word w/ location-based reminders, accent lights for blind spot alerts, updated Pet Mode & more Robotaxi – Paid Robotaxi miles doubled sequentially in Q1 – Cybercab will begin replacing Model Y fleet once in production & be the largest volume vehicle in the fleet over time – Continuing to lay the groundwork for expanding into new cities (testing, permitting), so we can launch quickly once ready. Safety remains top priority – Expanded unsupervised ops in Austin & launched in Dallas & Houston in April FSD Supervised – Record net new FSD subscriptions in Q1 – Received approval to deploy FSD Supervised in the Netherlands in April, clearing the path for potential approval in other EU countries – Continuing to make progress on approval in China Automotive Services – Safety Score v3.0 enables every mile driven with FSD Supervised engaged to receive a score of 100. Higher Safety Score over time = lower premiums for Tesla Insurance customers

Allbirds $BIRD to sell its brand and footwear assets, rename itself NewBird AI, and use a new $50M convertible financing facility to buy high-performance GPUs to pivot into AI compute infrastructure.

⚡️SPACE DATA CENTERS ARE NO LONGER SCI-FI Polymarket now price a 36% chance that a space-based data center launches by the end of next year. The AI compute race is getting so extreme, even orbit is becoming part of the infrastructure story.

with PRL trading around $1.5b fdv, I’m struggling to see why NOCK at roughly $75m fdv is not a layup over the next 3-12 months? I understand why PRL gets the AI premium. But a ~20x valuation gap feels excessive given that nock has several advantages: - It is further along technically - A large portion of miner emissions has already been absorbed by the market - It is a general-purpose verifiable compute / PoUW protocol rather than a single-workload bet - It has a credible path to expand into AI workloads such as MatMul if that demand materializes - It is aiming at app settlement, private/verifiable computation, and ZK-based infrastructure, not just one AI compute niche - It has a native privacy/verifiability angle through ZK proofs - Its fair-launch supply dynamics may now be more favorable after months of miner distribution im very bullish on proof of useful work becoming a major narrative over the next year. But with fair-launch PoW assets, the earliest phase is often dominated by emissions and miner selling. That can make the asset uninvestable until enough supply has been distributed and absorbed. That is where I think nock may have a structural advantage: it is already further through that emission overhang while still being valued like an obscure technical project, not like a PoUW narrative asset. with prl proving the demand for PoUW projects and nock being the furthest along, I dont see how nock isnt a pretty solid bet here over the next 3-12 months. nock nock