#AICapExPushContinues

#SandiskJoinsSP100 Sandisk is about to gain a new kind of buyer 👀 After jumping 10.99% on Sep 18, Sandisk joins the S&P 100 before the Sep 21 open, potentially triggering passive fund demand. What caught my attention is the timing. AI data-center growth already strengthened the storage story, and index inclusion now adds a flow catalyst. The real test starts after inclusion. Once passive buying settles, earnings and AI storage demand need to justify the $1,791.82 price.

🚨 $NVDA: NVIDIA IS BACK IN THE AI SPOTLIGHT 👀 $NVDA is moving with the broader AI/semiconductor rally today, while investors continue watching demand for its next-generation AI infrastructure. NVIDIA’s latest updates highlight its Vera Rubin platform and strong inference performance, keeping AI infrastructure at the center of the market narrative. 🔥 THE BIG STORY AI-chip demand remains a major driver for semiconductor stocks, with AMD also surging today and briefly reaching a $1T valuation. ⚠️ THE RISK NVDA remains sensitive to AI spending expectations, U.S.–China chip restrictions, and the ability of customers to secure enough power and infrastructure for new AI systems. 🔥 WHAT I’M WATCHING 1️⃣ NVDA price action during the U.S. session 2️⃣ Semiconductor sector momentum 3️⃣ AI-chip demand headlines 4️⃣ Any new U.S.–China developments **👀 NVIDIA IS STILL ONE OF THE BIGGEST NAMES IN THE AI TRADE — WATCH THE NEXT MOVE.** #CryptoCapReclaims2.8T #ZEC38KShortClosed #TrumpGulfIranTalks

AI-exposed jobs are seeing the largest pay growth in the US: Advertised pay in the most AI-exposed occupations has surged +46% since the start of 2021. At the same time, moderately AI-exposed jobs have seen +41% growth. This compares to a +39% increase in all US posted wages and just +25% for the least AI-exposed jobs. The gap began widening materially in mid-2025 and has continued to do so throughout this year. The divergence suggests employers are increasingly competing for workers in AI

TRUMP: “WHOEVER WINS $AI , WINS” President Trump says the U.S. must avoid stifling AI growth, arguing the technology could become bigger than the Industrial Revolution or the internet. Trump said the U.S. is currently leading China and other countries in $AI and pledged to preserve that position. He added that law enforcement could intervene if necessary, but said his administration would otherwise encourage $AI and “super intelligence.”

Server DRAM demand is expected to grow nearly 6x by 2030 and account for ~60% of the entire market. That matters most for $MU and $SKHY since servers are expected to drive ~80% of all incremental DRAM demand as AI infrastructure becomes biggest source of growth#CryptoCapReclaims2.8T #ZEC38KShortClosed #TrumpGulfIranTalks

$SNDK has made multiple attempts to break through 1800, but each time it failed to hold and pulled back. Now comes the classic question: third time’s the charm? 👀 Both $MU and $SNDK are benefiting from renewed interest in the storage and semiconductor narrative. AI data-center demand, HBM, enterprise SSDs, and renewed capital flows into semiconductors are all keeping the sector in focus.#DailyOrbit

$VVV — Venice Token Retests $22 Support as AI Narrative Strengthens Venice Token has remained one of the stronger AI-related assets in September, gaining about 33% over the week before pulling back from its September high near $29.29. Market Structure: VVV is currently retesting the $22 area, which aligns closely with its previous record-high zone. Holding this region could keep the broader recovery structure active. Key Levels: Support: $21.50–$22.00 Major Support: $19.50–$20.50 Resistance: $24.00–$25.00 Higher Zone: $28.00–$30.00 News Catalyst: VVV recently gained additional market access through an OKX VVV/USDT spot listing. Venice is also gaining attention as a privacy-focused AI platform, with VVV used for staking and access to AI inference capacity. TenX Protocols also added VVV to its digital-asset treasury, purchasing more than 7,700 VVV at an average price of about $17.17. Bottom Line: VVV is holding an important $21.50–$22 support zone after a strong rally. A recovery above $24 could bring the $28–$30 area back into focus.#CryptoCapReclaims2.8T #ZEC38KShortClosed #TrumpGulfIranTalks

I think $Surplus is worth starting to pay attention to. 0xc52aedec3374422d7510e294cfaa90799595cba3 I think this rally is not simply AI Meme speculation, but rather the market starting to reprice the AI Inference Marketplace product narrative. SURPLUS’s core product is an AI Inference Marketplace, which can be understood as OpenRouter + an order book + a secondary market for AI inference. The problem it solves is straightforward: inference costs vary significantly across different models, different providers, and different time periods, while Surplus is trying to turn AI inference from a fixed API price into a market-based trading market. Simply put:AI Model / Provider → Inference Capacity → Surplus Marketplace → Developer / AI Agent Whoever can provide cheaper inference gets more orders. So it is not trying to train a new AI model itself. Instead, it is trying to become the trading and routing layer for AI inference. This is also why I think its Narrative Ceiling is higher than that of ordinary AI Agent Tokens. The AI industry is gradually moving from “who trained the strongest model” toward “who can access these models at the lowest cost.” Especially as AI Agents, Coding Agents, and Autonomous Agents continue to increase, inference will gradually become one of the core costs of the AI Economy. And Surplus is positioned exactly here:Model Provider → Inference Capacity → Marketplace → Price Discovery → AI Agent / Developer One of the most important things to watch about this recent rally is that the product data is starting to become very real. According to the official Analytics, over the past 28 days it has generated approximately 43.8M Requests, 2.45T Input Tokens, 38.57B Output Tokens, and 2.02T Cache Tokens. A large number of requests have also been able to receive very high inference discounts during this period. This means that what the market is starting to trade is no longer “an AI concept project,” but:Real API Requests → Real Token Consumption → Real AI Inference Demand In other words, Surplus’s Product Thesis is beginning to receive actual data validation. Another important Catalyst is x402 + AI Agents. Surplus already supports x402, allowing AI Agents to pay directly in USDC on a per-request basis without traditional registration or API keys. If AI Agents eventually begin autonomously calling models at scale, then:Agent Count ↑ → Inference Requests ↑ → AI Spending ↑ → Marketplace Volume ↑ Surplus could have the opportunity to evolve from an ordinary AI API Router into the Inference Settlement Layer of a Machine-to-Machine AI Economy. This would significantly increase its Narrative Ceiling. At the same time, Surplus has already integrated with Claude Code, Cursor, OpenCode, Aider, Continue, Cline, Codex CLI, and other development tools, while also providing an OpenAI-compatible API. This means it solves a very important Adoption Problem: users do not need to learn an entirely new AI product. They only need to change the API Endpoint and can continue using the tools they already use. If this model continues to expand, the theoretical flywheel would be:More Tools Integrated → More Developers → More API Requests → More Inference Volume → More Sellers Providing Capacity → Deeper Order Book → Lower Prices → More Buyers This is what makes Surplus truly worth watching. What it is really trying to build is not an ordinary AI aggregator, but an AI Inference Marketplace Flywheel. So what I think the market is actually betting on right now is not:“Surplus is an AI Token.” It is:AI Inference will eventually become a resource that can be marketized, routed, and traded, and Surplus has the opportunity to become the trading layer for this market. In other words:AI Models → Inference Capacity → Marketplace → Price Discovery → Agents / Developers → AI Economy If this Thesis works, then Surplus’s valuation logic would no longer be that of an ordinary small-cap AI project, but would be closer to:AI Inference Exchange / Marketplace Infrastructure But there is one very important question here: Product success ≠ Surplus necessarily capturing product value. This is currently my biggest question about it. The market valuation it receives is still relatively low, while the Surplus product itself has already generated significant inference usage. The problem is that, according to the currently public product documentation, Marketplace settlement is still mainly:Buyer → USDC → Seller along with the relevant Fee Recipient. The currently public mechanism does not sufficiently demonstrate:Marketplace Revenue → Surplus Buyback / Burn / Staking / Holder Revenue So the market is currently trading ahead of:Product Growth → Future Token Value Capture rather than an already-existing:Product Revenue → Token Buyback This is also what makes SURPLUS particularly interesting right now. The market is beginning to believe that the product may succeed, but the value-capture mechanism between the Token and the product is still in an early stage. So the current rally logic can be understood as:AI Narrative + Real Product Data + AI Agents + x402 + Marketplace Model + Low Market Cap → Attention ↑ → Volume ↑ → Price ↑ → AI Narrative Strengthens → More Attention This is a very typical Crypto Reflexivity cycle. For future upside Catalysts, I would focus on several directions. First is continued Product Usage growth. If current Requests continue growing from tens of millions to:100M → 500M → 1B+ then it will become increasingly difficult for the market to view Surplus as merely a Narrative Project. Second is the actual formation of Marketplace Liquidity. The most important question right now is not how many models there are, but whether there are enough Sellers. If:Seller ↑ → Offer ↑ → Model Coverage ↑ → Price Competition ↑ → Buyer ↑ then Surplus will truly begin to form a network effect. Third is AI Agent Adoption. If Claude Code, Codex, and more Autonomous Agents begin using Surplus for inference at scale, then its positioning could move from:AI API Router to:Inference Layer of the AI Agent Economy That would be a major Narrative Upgrade. Fourth is an explosion in x402 usage. If we eventually see large amounts of Agent Payments, x402 Requests, and Machine-to-Machine Inference, that would further prove that Surplus is entering a new AI Economy rather than simply being a traditional API Aggregator. Fifth, and the one I most want to see: Token Value Capture. If we eventually see:Marketplace Fees → Surplus Buyback → Burn or:SURPLUS → Staking → Inference Credits or even:Holding SURPLUS → API Discount / Inference Allocation then:Product → Revenue → Token would finally form a complete loop. Once this mechanism appears, the market’s valuation logic for Surplus could change significantly. On the other hand, its risks are also very clear. The biggest risk is Token and Product decoupling. If: Requests ↑ Users ↑ Revenue ↑ but:SURPLUS has no value capture then we could ultimately see:The product is a good product, but the Token is merely a Narrative Proxy for Surplus Intelligence. The second risk is competition from OpenRouter and other existing infrastructure. Surplus must prove that:Marketplace / Order Book / Seller Capacity has stronger network effects than a normal AI Aggregator. Otherwise, it could ultimately become:Another OpenRouter and its valuation ceiling would be limited. The third risk is Seller Liquidity. The biggest problem for a marketplace is:There are Buyers, but not enough Sellers. Without enough Sellers: → No price competition → No cheap inference → Product advantage declines → Buyers leave So going forward, what should really be monitored is Active Sellers, Order Book Depth, and actual trading volume, rather than simply the number of Models. The fourth risk is Provider / API Terms. Part of Surplus’s model involves selling unused API Capacity. This means that different AI Providers’ policies regarding API Credits, Subscription Capacity, Account Sharing, or Resale could all affect Seller Supply. The fifth risk is continued decline in Inference Prices. This is actually a double-edged sword. For users:The cheaper the inference, the better. But for a Marketplace:The economic value of every 1M Tokens is also declining. Therefore, Surplus needs:Huge Usage Volume to compensate for:Declining Inference Unit Prices. If prices decline faster than usage grows, the economic value of the Marketplace may not necessarily grow at the same pace. So my core view on SURPLUS right now is:It is no longer simply an AI Narrative Token. It is an AI Infrastructure Bet that is beginning to show real product usage, while Token Value Capture is still in the early validation stage. What the market is really betting on right now is:The core infrastructure of AI in the future will not only be the Model itself, but that Model Inference could become a massive market, and Surplus has the opportunity to become the Exchange / Router / Liquidity Layer for that market. The most important thing to watch next is not simply the price, but whether this entire chain can run through:Requests ↑ → Sellers ↑ → Order Book Depth ↑ → AI Agent / x402 Usage ↑ → Marketplace Revenue ↑ → SURPLUS Value Capture ↑ If the first four continue to grow, it means the Product Thesis is being validated. But if the final step continues to fail to appear, then Surplus is more likely to remain a Narrative Proxy for Surplus Intelligence. If it can ultimately establish:Product → Revenue → Token then the entire valuation logic will finally form a complete loop. And that is where I believe SURPLUS’s biggest upside lies going forward: not simply in running another AI Narrative pump, but in the market confirming that it truly has a growing AI Inference Economy—and that SURPLUS can capture value from it. Chart:

SoftBank is raising high-yield debt to fund its latest $10 billion investment in OpenAI, with an additional €1 billion also planned. Its BB+ credit rating places the notes in junk-bond territory. The payment is the third of three $10 billion installments agreed in February. Once completed, SoftBank’s total cash investment in OpenAI will reach $64.6 billion for an estimated 13% stake. OpenAI’s financing documents project revenue rising from $36 billion in 2026 to $350 billion in 2030, but also forecast $278 billion in cumulative negative free cash flow over that period. Its preferred shares cannot be freely traded before an IPO. Why it matters: SoftBank is financing a large, illiquid AI bet with borrowed money while OpenAI expects to consume vast amounts of capital. Read the full article: https://t.co/6ovJe6kFzW

Let's talk about something outside the crypto circle but bound to impact the crypto world sooner or later. UBS said today that global capital expenditure in the AI sector will reach $1.4 trillion by 2027, with 90% of the increase coming from rising memory prices; on the same day, Changxin's G5 went into mass production, preparing to enter the flash memory market to compete head-on with Samsung, SK Hynix, and Micron. #CryptoCapReclaims2.8T #ZEC38KShortClosed #TrumpGulfIranTalks