You're reading a free issue of The Heat Sheet

Get it in your inbox every week — AI-powered signals, volume leaders, unusual options flow.

This Week's Setup

The Bottleneck Trade

14 tickers our scanner flagged repeatedly over the past 30 days. The common thread isn't a sector — it's capacity that's already sold out. Memory, photons, foundry slots, AI cloud GPUs. The names with locked contracts are running. The ones still pitching are not.

— Theme 01 —

The Memory Crunch

The bottleneck nobody saw coming three quarters ago is now the loudest story in semis. HBM is allocated through 2026. NAND pricing is set to climb up to 234% this year per Gartner. Both names below are running on signed multi-year contracts — not hope.

Closed at a fresh 52-week high of $576.45 on May 4, then pushed past $666 on May 6, putting market cap around $752B. Q2 FY26 earnings on March 18 were among the largest beats in company history: $12.20 adjusted EPS vs $9.31 expected on $23.86B revenue. Management said HBM is sold out through calendar 2026 under multi-year contracts; HBM4 entered volume production for NVIDIA's Vera Rubin one full quarter ahead of guidance. Forex News reports analyst price targets now hitting four digits for the first time.

AI Memory  ·  HBM4  ·  NVIDIA Supply Chain  ·  Pricing Power

The most extreme single-stock chart on the Nasdaq-100 this year. SNDK was trading at ~$1,468 on May 8 per a TECHi market snapshot — up roughly 500%+ YTD and now Nasdaq-100's top performer. Q3 revenue: $5.95B, up 251% YoY; data-center revenue +645% to $1.47B. The company has signed five long-term customer agreements worth over $42B in future revenue (Susquehanna call), announced a $6B share buyback, extended the Kioxia JV through 2034, and invested $1B in Nanya. Susquehanna's $2,000 target is currently the highest on the Street; Evercore's bull case sits at $2,600.

NAND  ·  Enterprise SSD  ·  Multi-Year Contracts  ·  Buyback

⚡ Verdict — Memory Crunch

Both names are operating in contracted, capacity-constrained regimes — that's a different animal than the 2022 cycle. The real risk isn't demand. It's a Hynix or Samsung capex surge that breaks the discipline. Watch the SK Hynix-ASML order flow and any HBM price cracks before assuming this runs forever.

— Theme 02 —

Photon Rush 2.0

Inside AI data centers, copper is tapping out. Light is the only way to move 800G and 1.6T traffic between racks. That means optical transceivers, retimers, and the exotic substrate materials underneath them. Two micro-caps just had their re-rating moment.

The chart did three years of healing in one morning. Q1 reported April 23: revenue $137M, up 43%, swung to +$0.22 adj EPS from -$0.05 a year ago, and Q2 guide of $160-170M well above the Street. The stock ripped 76% in one session (April 24) and ran from $21 on April 13 to $94.54 by May 8. Catalyst: the Washington 200G four-lane TIA targeting 1.6T optical interconnects (sampling now, volume 2H 2026) plus Panther V — a data-movement accelerator for AI inference bottlenecks, addressable market ~$5B per management. Loop Capital lifted its target from $17 to $75. Five firms upgraded in a week.

Optical Silicon  ·  1.6T Interconnects  ·  Inference Acceleration  ·  Re-Rating

The purest listed bet on indium phosphide (InP) substrates — the exotic wafer underneath the high-speed lasers and photodetectors that power AI data center optics. From a $1.32 52-week low to an all-time high of $129.43 on May 8, the stock closed Friday near $116. The April 26 $632.5M equity raise priced 8.56M shares at $64.25 — used to fund Beijing Tongmei Xtal expansion of InP capacity for global export. Management is calling for sequential revenue growth, record InP demand, and a backlog north of $100M, mostly AI / data-center upgrades. Wedbush raised targets after the raise. Beta around 2.8.

InP Substrates  ·  Silicon Photonics  ·  High-Beta Inflection  ·  Equity Raise

⚡ Verdict — Photon Rush 2.0

Both have re-rated hard and fast. The fundamental story is real — copper genuinely can't carry AI interconnect at scale — but valuations are now pricing in a clean 2H 2026 production ramp. Any optics qualification slip or hyperscaler push-out hits both names in the same week.

— Theme 03 —

The AI Cloud Spine

Four names that are not building AI models — they're building the floor underneath them. A reborn foundry, two specialist clouds, and a former Bitcoin miner now signing nine-figure NVIDIA contracts. All four are running on multi-year locked revenue.

The biggest turnaround tape in semis. Closed at $113.01 on May 6 — up 19% on the week, 122% on the month, and per 24/7 Wall St roughly 466% over the trailing twelve months. Q1 FY26 (April 23): non-GAAP EPS $0.29 vs $0.0127 consensus; revenue $13.58B; Data Center & AI +22% YoY to $5.05B; Foundry +16% YoY to $5.42B. The May 5 catalyst: a Bloomberg report that Apple is in talks with Intel and Samsung for U.S. foundry production. Lip-Bu Tan's line — "the CPU is the orchestration layer and critical control plane for the entire AI stack" — is reframing INTC as an agentic-AI play, not just a turnaround. Forward P/E ~119; consensus target $79.05.

Foundry  ·  18A Node  ·  Apple Talks  ·  NVIDIA Partnership

The neocloud that 18 months ago had $120M in annualized revenue — and is reporting Q1 2026 on Wednesday, May 13. Closed near $193 on May 11, up 133% YTD and 610% over the past year. Anchored by multi-billion-dollar Microsoft and Meta capacity deals, with NVIDIA committing $2B in March. The April $643M acquisition of Eigen AI is the strategic move — it pushes NBIS from raw GPU rental (IaaS) into post-training optimization (PaaS), a higher-margin layer. Management has guided $16–20B of 2026 AI infrastructure capex and is targeting $7–9B in full-year ARR. Q4 revenue was $227.7M; the Street is watching for the sequential jump.

Neocloud  ·  Microsoft Anchor  ·  PaaS Transition  ·  May 13 Earnings

Closed at $35.63 on May 4 with ~572% YTD outperformance vs the S&P's ~5%. The story compounds: a new $7.5B, 15-year lease with a second U.S. investment-grade hyperscaler at the Delta Forge 1 campus, lifting total contracted lease revenue beyond $23B. Q3 revenue surged 139% YoY to $126.6M, beating estimates by 61%. On May 5, APLD completed the spin-off of its cloud business into ChronoScale (NASDAQ: CHRN) — retaining 97% — repositioning the parent as a pure-play HPC / AI data center operator. Goldman is leading a $300M senior secured bridge facility for Polaris Forge 1 development in Ellendale, ND.

Hyperscaler Leases  ·  AI Factories  ·  ChronoScale Spin  ·  $23B Backlog

The Bitcoin miner that became a hyperscaler partner. May 7 brought a triple announcement: a $3.4B, 5-year AI cloud contract with NVIDIA, a 5GW strategic partnership pipeline, and NVIDIA receiving a 5-year right to invest up to $2.1B via warrants exercisable at $70. Layer that on top of the existing $9.7B Microsoft contract and the $625M Mirantis acquisition (closes the software / orchestration gap for the NVIDIA cloud delivery). Q3 revenue came in light at $144.8M vs $219.87M expected — the market took the stock down to $59.50 on the print, but it had run from $38 to $72 in the prior two weeks. ARR under contract: $3.1B today, target $3.7B by year-end.

NVIDIA Anchor  ·  5GW Pipeline  ·  Crypto-to-AI Pivot  ·  Mirantis

⚡ Verdict — AI Cloud Spine

Four very different stories with one shared truth: customer concentration is the real risk. Microsoft, NVIDIA, Apple — these are the names whose capex calendar moves all four stocks together. NBIS prints Wednesday. APLD just spun. IREN just missed but signed mega-deals. INTC needs a confirmed foundry contract to keep the multiple from compressing.

— Don't Miss Next Week —

Get The Heat Sheet free, every week

The exact tickers our scanner is flagging — clustered into themes, with the data behind each move. Sent every Monday.

— Theme 04 —

Network Layer For AI

The CDN/edge players were left for dead two years ago. Then AI bots hit the internet at scale — Akamai's State of the Internet report flagged a 300% jump in AI bot activity in 2025. Suddenly the network layer is monetizing inference, not just delivering video.

The single biggest contract in company history dropped May 7: a $1.8B, seven-year commitment for Cloud Infrastructure Services from "a leading, U.S. based frontier model provider" (the Street is betting on a major AI lab). Q1 revenue $1.074B (+6%), Cloud Infrastructure Services +40% YoY to $95M, Security +11% to $590M. Stock jumped 21% to $147.71 the next day and pushed toward $145+ intraday on May 8. Akamai was also named sole Customers' Choice for API Protection in Gartner Peer Insights with a 93% recommendation rate. Management raised full-year CIS growth guidance to at least 50% in constant currency.

Frontier Model Contract  ·  API Security  ·  CIS 50% Growth  ·  CDN Rebirth

The classic "good print, bad reaction" tape. Q1 (May 6): record revenue $173M (+20% YoY), security +47%, RPO +63% to $369M, non-GAAP EPS $0.13 (vs $0.09 consensus). Full-year guide raised to $717.5M at midpoint. Stock dropped 27.2% the next day — investors saw infrastructure capex going from 5% to 10–12% of revenue and got cold feet. The product story is the most interesting part: ContentGuard launched in Q1 specifically to protect IP from unauthorized AI bot scraping, addressing the exact problem Akamai's State of the Internet flagged. Named a Leader in Forrester Wave: Edge Development Platforms, Q1 2026.

Edge Compute  ·  AI Bot Protection  ·  Margin Pressure  ·  Forrester Leader

⚡ Verdict — Network Layer

Two different ways the network layer monetizes AI. AKAM landed the elephant — one frontier-model contract that re-rates the entire CIS business. FSLY is taking the bottom-up approach: ContentGuard plus security. Both are now infrastructure stocks, not media-delivery stocks. The CapEx scale is the new gating factor.

— Theme 05 —

Software Eating Itself + Wild Cards

Two software names betting their futures on agentic AI (with very different market reactions), a stablecoin issuer that just dodged a regulatory bullet, and Bill Ackman's brand-new NYSE listing. Different stories — same shared property: high event-driven volume.

Earnings dropped this morning (May 11). Q1 revenue $351M, +24% YoY; AI products contributed 10% of net new ARR; record $49M operating profit. The big strategic shift: monday.com is moving to a consumption-based pricing model on top of seat-based, just acquired OneAI to add native voice agents, and launched a re-architected AI Work Platform. FY guide raised to $1.466–1.474B. Stock was up 25% in premarket. Worth noting context: MNDY had collapsed from $317 in 2024 down to $57.50 — Barclays cut targets to $95 in April, Piper downgraded to Neutral, and a shareholder lawsuit landed on April 27 over long-term AI growth claims. Today's print is the rebuttal.

Consumption Pricing  ·  OneAI Acq  ·  Agentic AI  ·  Earnings Bounce

The other side of the SaaS-AI trade. NICE beat Q1 (May 6) — revenue $768.6M (+9.8%), cloud +14.6% to $603.4M, AI ARR +66% YoY, AI in 100% of new CXone enterprise deals — and the stock fell over 20%. Why? Q2 guide of $761–771M is below the Street's $780M, and Calcalist reported investor concerns that the raised EPS guide was driven mostly by a $600M share buyback rather than organic growth. The Cognigy acquisition (closed 8 months ago) is "ahead of plan." Management is also exploring divestitures of non-CX assets.

AI Contact Center  ·  Cognigy Integration  ·  Buyback-Driven EPS  ·  Guidance Cut

Two giant catalysts in one week. (1) May 4: Senators Tillis and Alsobrooks released a bipartisan CLARITY Act compromise that preserves activity-based stablecoin rewards (only passive yield is banned), removing the biggest overhang. Stock jumped ~20% to ~$115. (2) May 11 earnings: USDC in circulation +28% to $77B, on-chain transaction volume +263% to $21.5T, revenue $694M (+20%). Operating expenses surged 76% — that's the bear case. Mizuho data shows USDC at ~64% of adjusted stablecoin volume YTD — first time leading USDT since 2019. Stock at ~$125 on May 11, still 56% below its 52-week high of $298.99. Arc token presale raised $222M at a $3B FDV.

USDC  ·  CLARITY Act Compromise  ·  Arc Blockchain  ·  Agentic Commerce

Bill Ackman's hedge fund firm just went public. Pershing Square Inc. (NYSE: PS) and the closed-end fund PSUS listed April 29 in a combined $5B IPO — among the largest closed-end fund debuts in U.S. history. Buyers of PSUS shares received one PS share for every five PSUS shares at no extra cost. PS opened at $24, closed at $24.20 on day one; as of May 5 it traded around $34.21 with a 52-week range of $22–$38. Institutional investors took over 85% of orders. The thesis is Ackman finally getting his "Warren Buffett-style" permanent capital vehicle in public form — a different beast than betting on his individual ideas. Heavy initial volume is the scanner trigger.

Brand New IPO  ·  Ackman Vehicle  ·  Permanent Capital  ·  Asset Manager

⚡ Verdict — Wild Cards

The pattern across these four: the trade isn't the company, it's the catalyst. MNDY swung on a Q1 print today. NICE on a soft guide. CRCL on legislation. PS on first-month-of-trading flows. These names live and die on event days — they reward traders, not buy-and-hold investors.

— Big Picture —

When the Whole Stack Is Sold Out

Step back from the 14 names and notice a single sentence describes most of them this quarter: "sold out, contracted, capacity-constrained." Micron and SanDisk on memory. Applied Digital and IREN on data center power and GPU slots. Akamai on cloud infrastructure. Even AXTI is racing to add InP capacity that's already pre-sold.

The macro on AI capex backs this up. Gartner has flagged NAND prices potentially rising up to 234% in 2026. IDC projects a $174.1B NAND market this year. Micron raised FY26 capex past $25B. Akamai is taking FY26 CapEx to 40–42% of revenue. These are not the moves of an industry worried about a 2027 air-pocket — these are the moves of operators racing to install capacity they've already promised customers.

The setup our scanner is surfacing is companies that turned signed contracts into visible revenue ahead of consensus. The two failure modes to watch: (1) any hyperscaler trimming their AI capex calendar — which would unwind APLD, NBIS, IREN, and AKAM together; and (2) memory pricing breaking discipline, which still pulls down MU and SNDK hard. Neither is the base case today. Both are tail risks worth tracking. For now, the tape is rewarding capacity, not promises.

— Make This Weekly —

Join readers who get The Heat Sheet first

Same scanner output, same theme clustering, sent every Monday before the open. Free, no upsell in your inbox.

Share this issue

STOCKARKALYTICS

DATA-DRIVEN MARKET INTELLIGENCE

Home  ·  Blog  ·  Subscribe  ·  Terms  ·  Privacy  ·  Disclosure

StockArkalytics is operated by Aspire Catalyst LLC
Irvine, CA · United States

Disclaimer: The Heat Sheet is published for informational and educational purposes only. Nothing in this publication constitutes investment, financial, legal, tax, or accounting advice, or an offer or solicitation to buy or sell any security. All scanner output, price levels, and analyst targets are sourced from public filings, third-party data providers, and news media as of the publication date and may change without notice. Past performance does not guarantee future results. Trade at your own risk. Do your own research.

© 2026 Aspire Catalyst LLC  ·  All rights reserved