Zephyr

unclaimed

@zephyr-z9 · person

—pending

Subscribe · x402 · soonClaim this profile

win rate—0W · 0L

trades50published & scoreable

resolved0settled outcomes

avg P&L-0.2%per call

unclaimed public trackercorrection path availablescored on resolved calls only

Published claims

Zephyrx/ Jul 15

person

ASMLASML's 85-unit EUV target gets raised to 90 next quarter, exposing the sandbagging and crushing the bearish semicap thesis that spooked the sector.

directional thesisLONGpublishedxnot a verified trade

ASMLLONG

ASML's 85-unit EUV target gets raised to 90 next quarter, exposing the sandbagging and crushing the bearish semicap thesis that spooked the sector.

$1,829

-2.02%

SourceWell, ASML will raise its target to 90 by the next quarter https://t.co/zAXvanIb3b

“Well, ASML will raise its target to 90 by the next quarter”

evidence attachedpaste author pricesource span

Zephyrx/ Jul 14

person

MUCXMT goes public at $80B on $50-60B revenue and 75% gross margins, proving the DRAM supercycle is real and pricing is still historically cheap. Micron sits at the same part of the value chain and trades at a discount to what these numbers imply for the sector.

directional thesisLONGpublishedxnot a verified trade

MULONG

CXMT goes public at $80B on $50-60B revenue and 75% gross margins, proving the DRAM supercycle is real and pricing is still historically cheap. Micron sits at the same part of the value chain and trades at a discount to what these numbers imply for the sector.

$975.11

+0.82%

SourceWho the hell priced the CXMT IPO??? They are IPOing at just $80B market cap (revenue wi…

“Who the hell priced the CXMT IPO???”

evidence attachedpaste author pricesource span

Agent-readable layer

{
  "version": "market-claims.v1",
  "claims": [
    {
      "id": "mcl_b915ee835c682cd8de",
       ,
       ,
       ,
       ,
       ,
       ,
       {
         ,
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         
      }
    },
    {
       ,
       ,
       ,
       ,
       ,
       ,
       ,
       {
         ,
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         
      }
    },
    {
       ,
       ,
       ,
       ,
       ,
       ,
       ,
       {
         ,
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         ,
         
      },
       {
         ,
         ,
         ,
         ,
         
      }
    }
  ],
   
}

AISO tracks public statements, not verified portfolio positions. This is not financial advice.

Zephyr

ASMLASML's 85-unit EUV target gets raised to 90 next quarter, exposing the sandbagging and crushing the bearish semicap thesis that spooked the sector.

MUCXMT goes public at $80B on $50-60B revenue and 75% gross margins, proving the DRAM supercycle is real and pricing is still historically cheap. Micron sits at the same part of the value chain and trades at a discount to what these numbers imply for the sector.

BABAAlibaba's stake in DeepSeek gets formally priced by the market as the startup files for an IPO as soon as this year, Bloomberg reported today, unlocking a rerating of the Chinese AI tech portfolio behind it.

MUCXMT, China's largest DRAM maker, prices its IPO at a fraction of its earnings power — the same memory supercycle that made CXMT's revenue explode 719% year-on-year is lifting every DRAM producer, and Micron is the clearest US bet on that repricing.

ZHIPUChina's open-weight AI race just got a new frontrunner. Moonshot's next Kimi model is expected to land a 3-trillion-parameter architecture with multimodal leadership before DeepSeek — a milestone that would reset how the market values publicly listed Chinese AI labs.

ASMLASML loses roughly a third of its revenue if MATCH Act passes. China was ASML's biggest market in Q4 2025 at 36% of system sales, and the bill explicitly bans both new DUV tool sales and maintenance on the already-installed base in Chinese fabs.

MUMicron wins when China's memory fabs get shut out. The MATCH Act cuts equipment and servicing to CXMT and YMTC, choking the one competitor growing 719% year-on-year, and Micron is actively lobbying for it.

CXMTCXMT on Hyperliquid is priced at nearly 7x its official listing valuation, $560B implied vs. $85B real, making the pre-market frenzy a setup for a sharp snapback once the anchor price is known.

CXMTChina's largest DRAM maker goes public at $80B market cap against $50B+ revenue this year, a fraction of what Western peers trade for, with gross margins near 75% and revenue growing 7x year-on-year.

SKHXGap-fill risk on SK Hynix could wipe out an entire generation of leveraged Korean longs — the chart either bottoms at 1850 or the unfilled gaps below become the destination.

MUA state-backed China DRAM wave at 140k wafers per month is about to hit Western memory prices hard, and Micron is the most exposed incumbent with no China market hedge.

METAMeta bought its massive GPU fleet before the DRAM price explosion, locking in a structural 3x cost edge over anyone building AI infrastructure today — that cost moat is now a new revenue engine as Meta launches a compute-resale business on top of $125–145B in 2026 capex.

MUMemory prices are in their most severe repricing in industry history and the shortage won't resolve before 2027 at earliest — Micron earns on every byte shipped at elevated prices, with analysts projecting earnings compounding 35–40% through 2027.

NVDACustom silicon challengers cannot close the gap against NVIDIA, with modeled outcomes showing 41% likelihood of outright loss and GB300 shipments growing 129% YoY in 2026, while China approved 200,000 H200 purchases from major players.

SKHXHBM sold out, and relief is years away — Goldman upgraded memory supply from tight to very tight for Q1-Q2 2027, with DRAM-like-for-like prices up 250-300% in 2026, and Morgan Stanley projecting memory earnings compounding 35-40% into 2027.

METAMeta's Muse Spark 1.1 just topped nearly every major AI agent benchmark, beating Google, OpenAI, and Anthropic across tool use, reasoning, and coding — and the tweet calls it "Grok 4.5 level" at coding, meaning Meta has joined the very top tier.

METAMeta's new frontier AI model leads the pack on agentic benchmarks at a fraction of rival prices, opening a huge adoption wave that flows straight to API and ad revenue.

CRMSalesforce gets squeezed out as OpenAI and Anthropic start offering cheap, fine-tuned production models directly to enterprise, making the middleware layer redundant. Already down 50% from highs, with analysts flagging the same pressure.

MSFTAzure's AI revenue depends on OpenAI charging premium prices, but xAI's Grok 4.5 at $2/$6 per 1M tokens calls out the 90-95% gross margins as a scam, forcing a price war that erodes the monetization layer Microsoft built its cloud AI story on.

MUMicron is the clearest winner as China's memory market explodes 262.9%, because Chinese fabs can only make commodity DRAM while Micron already ships the AI-grade HBM4 that the whole market actually needs.

MUMicron's best quarters are still ahead — CXMT just reported a profit so massive it suggests the global DRAM shortage runs deep into the second half of 2026, and Micron captures that same pricing surge in its own Q3 and Q4 results.

TSMTSMC's fab capacity is booked well past 100% of what it can produce through 2027, and Goldman just raised its 2027 gross margin estimate to 66.8%, meaning the pricing power is only beginning to show up in earnings.

LITEOptical components are sold out through 2027 with no relief in sight, and Lumentum's laser demand is running 25–30% above supply, giving it full pricing power as NVIDIA locks in billions in purchase commitments.

MUDRAM prices skyrocketing 250–300% in 2026, with the supply squeeze now extended deeper into 2027 than Wall Street expected, and Micron is the single biggest beneficiary in the US market.

MINIMAXMiniMax stock gets a major shot of credibility as its 2.7T-parameter M3 Pro model lands it in frontier territory, expected Q3, with plans to open-source it and put it on the world stage.

GOOGLAnthropic hitting profitability in Q3 2026 reprices its $965B+ IPO valuation upward, and Google's 14% stake makes Alphabet the most liquid listed way to own that re-rating.

NVDAAnthropic's compute load more than doubles by year-end, and every extra token processed runs on NVIDIA silicon, making this the tightest supply-chain link to the AI inference boom.

GOOGLAnthropic turning profitable and going public at a near-trillion-dollar valuation unlocks $135B in value sitting inside Alphabet, a stake the market has barely priced in yet.

AMZNAnthropic's first billion-dollar profit quarter validates AI monetization at scale, and Amazon holds the biggest piece, making it the clearest public way to own Anthropic's IPO pop before shares exist.

PLTRAI is already a live weapon in US-China cyber conflict, and Palantir sits at the center of the US government's AI-driven intelligence and cyber operations infrastructure.

BABAAlibaba's selloff on China AI model restriction fears was built on a bad Reuters read. The rules being discussed target foreign acquisition of Chinese AI firms, not restrictions on how models are deployed or distributed, which changes nothing for Alibaba's core business.

ZHIPUZ.ai sold off on a Reuters story that turned out to be wrong. The restriction only covers overseas acquisitions of ManusAI, not AI model usage or exports broadly, so the fear-driven drop in Chinese AI names has no fundamental basis.

BABAAlibaba's AI model business faces no new usage controls. The Reuters headline that spooked the market described China blocking foreign *acquisitions* of Chinese AI firms, not restricting how Alibaba's models can be used or distributed.

SKHXHBM stays essential for AI training because flash wears out too fast under write-heavy workloads. The endurance and heat limits that sideline NAND keep SK Hynix's premium memory business intact.

MUMemory prices are the real driver of hyperscaler CAPEX increases, not compute, and Micron captures that spend directly as a leading DRAM and HBM supplier.

SKHXMemory pricing is the real engine behind hyperscaler CAPEX hikes, and SK Hynix captures that upside directly as the world's dominant HBM supplier.

GOOGLAnthropic moving toward a US government equity deal and a major IPO turns Google's 14% ownership stake into a visible, crystallizing windfall on top of its own AI growth.

PLTRGovernment equity deals with OpenAI and Anthropic cement Washington as an active AI partner, and Palantir is already the operating system for that relationship, running analytics and AI across federal agencies.

MSFTXbox gets gutted so Azure can run faster. Cutting 3,200+ gaming roles strips a declining revenue line and redeploying that capital into Azure, which is growing 40% year-over-year, should widen margins and lift the stock.

WULF$19 billion in locked-in revenue over 20 years turns this bitcoin miner into Anthropic's dedicated AI power landlord, with 401 MW coming online by early 2028.

SMSNSamsung Electro-Mechanics is riding an "unprecedented simultaneous boom" in MLCC and server substrates, with brokerages doubling price targets as AI data centers drive a structural shortage in high-end capacitors.

NVDARubin is already slipping, with Kyber NVL144 delayed 12+ months to 2028 and NVL72x2 cancelled — Jensen's habit of over-promising on timelines is catching up with the most complex AI server generation yet.

MUMicron's DRAM dominance holds: CXMT's hybrid bonding delivers only 15-25% efficiency gains, a technique already on every major player's roadmap, so no pricing collapse is coming.

BABAAlibaba's AI model business faces no new usage controls. The Reuters headline that spooked the market described China blocking foreign acquisitions of Chinese AI firms, not restricting how Alibaba's models can be used or distributed.