Themes & Narratives

1. 

   If it holds: challenges AI compute · challenges AI platforms

   AI spending is outrunning its returns

   MSFT ↓, would come under pressure if this holds GOOGL ↓, would come under pressure if this holds AMZN ↓, would come under pressure if this holds META ↓, would come under pressure if this holds NVDA ↓, would come under pressure if this holds

   The claim faces diverging evidence across hyperscalers. Microsoft's disclosure of $10 billion in quarterly revenue tied to OpenAI demonstrates that direct monetization is scaling alongside infrastructure outlays, weakening the thesis of unbacked spending. However, investor unease over Alphabet's ability to protect search margins while doubling cloud revenue shows that returns remain uneven across core business models. The central question is whether third-party model partnerships can generate durable operating margins rather than gross revenue.

   No material update

   Last development
   29 Oct 2026

   Evidence & history →
2. 

   If it holds: challenges AI platforms · supports Power & minerals

   Power constraints are slowing data-centre expansion

   AMZN ↓, would come under pressure if this holds MSFT ↓, would come under pressure if this holds GOOGL ↓, would come under pressure if this holds VRT ↑, would benefit if this holds ETN ↑, would benefit if this holds GEV ↑, would benefit if this holds BE ↑, would benefit if this holds PWR ↑, would benefit if this holds CEG ↑, would benefit if this holds VST ↑, would benefit if this holds TLN ↑, would benefit if this holds

   Evidence increasingly supports the claim as power constraints produce direct financial consequences. NextDC's lowered guidance shows that connection delays and local opposition can directly impair facility timelines and operator earnings, confirming grid friction extends beyond US regional queues. While hyperscalers are funding on-site generation like fuel cells to circumvent backlogs, localized grid and community barriers remain binding. The open question is whether off-grid power can be deployed fast enough to prevent project deferrals.

   No material update

   Last development
   21 Oct 2026

   Evidence & history →
3. 

   In no theme: tracked for the companies it names

   Custom silicon is reducing dependence on merchant GPUs

   NVDA ↓, would come under pressure if this holds AMD ↓, would come under pressure if this holds AVGO ↑, would benefit if this holds GOOGL ↑, would benefit if this holds AMZN ↑, would benefit if this holds MSFT ↑, would benefit if this holds

   Evidence supporting the claim is broadening from single-firm internal use to commercial distribution. Google supplying TPUs to Oracle Cloud, alongside Oracle adopting Cerebras and OpenAI custom inference chips, directly challenges the assumption that proprietary silicon remains confined to internal hyperscaler workloads. This expands options for buyers seeking alternatives to merchant GPUs. However, merchant GPUs retain deep software ecosystems and dominance in frontier training, leaving open whether multi-cloud custom silicon can win large external developer share.

   No material update

   Last development
   28 Oct 2026

   Evidence & history →
4. 

   If it holds: challenges AI software

   AI agents are challenging per-seat software pricing

   CRM ↓, would come under pressure if this holds NOW ↓, would come under pressure if this holds MSFT →, cuts both ways if this holds ADBE ↓, would come under pressure if this holds

   AI agents may do work that people did, shrinking the paid seats enterprise software is sold by, or move vendors towards pricing by usage, credits or outcomes.

   Not yet assessed

   Awaiting evidence Evidence & history →
5. 

   If it holds: supports AI software

   Falling inference prices are commoditising model access

   MSFT →, cuts both ways if this holds GOOGL →, cuts both ways if this holds AMZN →, cuts both ways if this holds META →, cuts both ways if this holds CRM ↑, would benefit if this holds NOW ↑, would benefit if this holds PLTR ↑, would benefit if this holds

   Evidence continues to support the claim as frontier labs steadily lower access costs for agent-capable models. OpenAI's 50% price cut for its mid-tier model extends an established pattern of price reductions, following cuts from Anthropic and OpenAI's earlier releases. While lower prices expand agent adoption and token volumes, continuous deflation in API pricing compresses unit margins across model providers. The open question is whether labs can offset lower prices with sheer volume or if differentiation will erode.

   No material update

   Last development
   27 Oct 2026

   Evidence & history →
6. 

   If it holds: supports AI compute

   Advanced memory is becoming an AI supply bottleneck

   MU ↑, would benefit if this holds TSM ↑, would benefit if this holds NVDA ↓, would come under pressure if this holds AMD ↓, would come under pressure if this holds

   The timeline for relief from advanced memory constraints is extending further into the future. Industry executives now project that RAM shortages will persist through 2028, pushing memory prices for 2027 significantly higher than current levels. This prolonged supply-demand imbalance reinforces the status of high-bandwidth memory as a durable bottleneck in the AI hardware supply chain.

   No material update

   Last development
   1 Oct 2026

   Evidence & history →
7. 

   If it holds: supports AI platforms · supports AI software

   Enterprise AI is moving from pilots into production

   MSFT ↑, would benefit if this holds CRM ↑, would benefit if this holds NOW ↑, would benefit if this holds PLTR ↑, would benefit if this holds GOOGL ↑, would benefit if this holds

   Palantir's commercial revenue overtaking its government business offers tangible proof that corporate AI adoption is moving into production at scale. Triple-digit commercial revenue expansion indicates sustained willingness among corporate clients to fund AI platforms. However, analyst caution regarding decelerating contract growth highlights that future pipeline conversion could moderate. The key open question is whether enterprise AI spending can maintain this pace beyond early high-value customer implementations.

   No material update

   Last development
   3 Nov 2026

   Evidence & history →
8. 

   If it holds: challenges AI compute

   Export controls are fragmenting the technology supply chain

   NVDA ↓, would come under pressure if this holds AMD ↓, would come under pressure if this holds ASML ↓, would come under pressure if this holds AMAT ↓, would come under pressure if this holds LRCX ↓, would come under pressure if this holds

   The enforcement of US export controls remains a complex challenge, as evidenced by the ongoing illicit flow of restricted hardware. The arrest of a tech CEO accused of smuggling $300 million worth of Nvidia chips into China highlights the massive financial incentives driving black-market supply chains. This underscores the difficulty of fully isolating the Chinese AI ecosystem through trade restrictions alone.

   No material update

   Last development
   3 Oct 2026

   Evidence & history →
9. 

   If it holds: supports Power & minerals

   Nuclear power is being revived to run data centres

   CEG ↑, would benefit if this holds VST ↑, would benefit if this holds TLN ↑, would benefit if this holds OKLO ↑, would benefit if this holds BWXT ↑, would benefit if this holds LEU ↑, would benefit if this holds CCJ ↑, would benefit if this holds DNN ↑, would benefit if this holds UUUU ↑, would benefit if this holds

   Evidence continues to strengthen the claim through coordinated public financing and supply-chain commitments. The DOE's $5 billion loan offer to Constellation builds directly on earlier federal backing for Vistra and hyperscaler power purchase agreements, showing policy momentum to expand output at existing reactors where grid interconnections already exist. Commercial microreactors also took a step forward with fuel contracting. However, loan commitments remain conditional, and uprates require multi-year outages and regulatory approvals.

   No material update

   Last development
   28 Oct 2026

   Evidence & history →
10. 

    If it holds: supports Power & minerals

    Western critical-mineral supply chains are being rebuilt

    MP ↑, would benefit if this holds USAR ↑, would benefit if this holds UUUU ↑, would benefit if this holds LEU ↑, would benefit if this holds FCX ↑, would benefit if this holds

    Governments and manufacturers may be paying to build mining, refining and magnet production outside the dominant supplier countries.

    Not yet assessed

    Awaiting evidence Evidence & history →
11. 

    If it holds: supports Robotics

    AI is moving into robots and machines

    TSLA ↑, would benefit if this holds ISRG ↑, would benefit if this holds SYM ↑, would benefit if this holds TER ↑, would benefit if this holds ROK ↑, would benefit if this holds CGNX ↑, would benefit if this holds AEVA ↑, would benefit if this holds OUST ↑, would benefit if this holds MBLY ↑, would benefit if this holds

    Tesla's start of Optimus production and planned Cybercab sales show hardware advancing toward commercialization, while Waymo continues geographic expansion into airport transit. However, Tesla's commercial timelines remain company projections facing active federal safety reviews for vehicles without manual controls. The evidence shows fleet operators making steady operational headway in defined service zones, but widespread commercial readiness and regulatory approval for consumer autonomous vehicles remain significant hurdles.

    No material update

    Last development
    22 Oct 2026

    Evidence & history →
12. 

    If it holds: challenges Power & minerals

    Data-centre power demand is falling short of forecasts

    CEG ↓, would come under pressure if this holds VST ↓, would come under pressure if this holds TLN ↓, would come under pressure if this holds GEV ↓, would come under pressure if this holds VRT ↓, would come under pressure if this holds ETN ↓, would come under pressure if this holds PWR ↓, would come under pressure if this holds BE ↓, would come under pressure if this holds OKLO ↓, would come under pressure if this holds

    The sharp reduction in PJM's planned backstop procurement—potentially shrinking by over 90% to 630 MW as regional utilities trim large-load forecasts—provides initial evidence that speculative data-centre interconnection requests are being weeded out. This tempers recent bullish disclosures from utilities such as AEP and FirstEnergy, which reported surging long-term pipelines. The central open question is whether these downward revisions reflect genuine cancellations and efficiency gains, or simply adjustments while compute demand grows.

    New evidence

    On November 3, 2026, Jefferies analysts reported that planned backstop generation procurement in PJM Interconnection could drop from 6,800 MW to as little as 630 MW after regional utilities lowered their large-load forecasts.

    Last development
    4 Nov 2026

    Evidence & history →
13. 

    If it holds: challenges Robotics

    Safety concerns are holding back self-driving and robot deployments

    TSLA ↓, would come under pressure if this holds MBLY ↓, would come under pressure if this holds AEVA ↓, would come under pressure if this holds OUST ↓, would come under pressure if this holds

    Crashes, recalls, regulator investigations and new safety rules may limit where and how fast self-driving vehicles and autonomous robots are allowed to operate.

    Not yet assessed

    Awaiting evidence Evidence & history →
14. 

    If it holds: challenges AI platforms

    The AI build-out is outgrowing the cloud giants' cash flow

    META ↓, would come under pressure if this holds MSFT ↓, would come under pressure if this holds GOOGL ↓, would come under pressure if this holds AMZN ↓, would come under pressure if this holds ORCL ↓, would come under pressure if this holds

    Meta's planned $20 billion to $25 billion debt sale reinforces the claim that hyperscaler infrastructure commitments are outstripping organic cash generation. Coming alongside Microsoft's $25 billion debt plan and Meta's recent shift into negative free cash flow, large bond issues show tech giants turning to debt markets to sustain spending. Private market leaders like Blackstone are also warning of mounting debt concentration risks. The key question is whether rising borrowing costs will eventually force spenders to moderate their outlays.

    New evidence

    On November 3, 2026, reports indicated Meta is aiming to raise $20 billion to $25 billion in a bond offering to fund rising AI infrastructure outlays, following recent negative free cash flow and a negative outlook from S&P Global Ratings.

    Last development
    4 Nov 2026

    Evidence & history →
15. 

    If it holds: challenges AI compute

    Credit is tightening for debt-funded AI cloud builders

    CRWV ↓, would come under pressure if this holds NBIS ↓, would come under pressure if this holds APLD ↓, would come under pressure if this holds ORCL ↓, would come under pressure if this holds NVDA ↓, would come under pressure if this holds MSFT ↑, would benefit if this holds GOOGL →, cuts both ways if this holds AMZN ↑, would benefit if this holds

    The development supports the claim that credit is tightening for specialised cloud builders. Having a dominant chipmaker step in to cosign debt indicates that independent data-centre operators struggle to secure private financing on their own merits. While vendor backing keeps construction on track, it shifts credit and facility utilisation risk directly back onto the hardware supplier. The key question is whether such vendor guarantees will remain isolated or become necessary for broader neocloud expansion.

    No material update

    Last development
    31 Oct 2026

    Evidence & history →
16. 

    If it holds: supports AI compute

    Agentic AI is shifting compute demand towards inference

    NVDA ↑, would benefit if this holds AMD ↑, would benefit if this holds INTC ↑, would benefit if this holds ARM ↑, would benefit if this holds MU ↑, would benefit if this holds

    AI agents that run continuously may be moving computing demand from one-off model training to inference, which needs more processors and memory for every accelerator installed.

    Not yet assessed

    Awaiting evidence Evidence & history →
17. 

    If it holds: supports AI compute

    The largest AI buyers are still raising their capital budgets

    NVDA ↑, would benefit if this holds AVGO ↑, would benefit if this holds AMD ↑, would benefit if this holds MU ↑, would benefit if this holds TSM ↑, would benefit if this holds MRVL ↑, would benefit if this holds AMAT ↑, would benefit if this holds LRCX ↑, would benefit if this holds VRT ↑, would benefit if this holds GEV ↑, would benefit if this holds MSFT →, cuts both ways if this holds GOOGL →, cuts both ways if this holds AMZN →, cuts both ways if this holds META →, cuts both ways if this holds ORCL →, cuts both ways if this holds

    Anthropic's disclosure of nearly $900 billion in purchase commitments shows that frontier AI labs are locking in massive multi-year capacity agreements with infrastructure and chip providers. This expansive commitment provides a counterweight to flat 2026 capex guidance recently set by Alphabet and Amazon, signaling that multi-year appetite for compute capacity remains vigorous. The central open question is whether labs can convert these enormous long-term commitments into sustainable cash flow if revenue growth decelerates.

    No material update

    Last development
    3 Nov 2026

    Evidence & history →
18. 

    If it holds: challenges Power & minerals

    Rising power bills are turning regulators against data centres

    AMZN ↓, would come under pressure if this holds BE ↑, would benefit if this holds CEG ↓, would come under pressure if this holds GEV →, cuts both ways if this holds GOOGL ↓, would come under pressure if this holds META ↓, would come under pressure if this holds MSFT ↓, would come under pressure if this holds TLN ↓, would come under pressure if this holds VST ↓, would come under pressure if this holds

    Higher electricity and capacity prices may be leading regulators, grid operators and legislators to cap what generators earn, push more of the grid's cost onto data centres, or slow their connections.

    Not yet assessed

    Awaiting evidence Evidence & history →
19. 

    In no theme: tracked for the companies it names

    AI is replacing work companies paid service firms to do

    ACN ↓, would come under pressure if this holds CNXC ↓, would come under pressure if this holds COUR ↓, would come under pressure if this holds EPAM ↓, would come under pressure if this holds IBM →, cuts both ways if this holds NOW ↑, would benefit if this holds PLTR ↑, would benefit if this holds

    Businesses may be using AI agents for customer support, software maintenance, testing, back-office processing and training they used to buy from outsourcing, consulting and education firms.

    Not yet assessed

    Awaiting evidence Evidence & history →
20. 

    In no theme: tracked for the companies it names

    Memory shortages are squeezing the margins of device and server makers

    AAPL ↓, would come under pressure if this holds DELL ↓, would come under pressure if this holds HPE ↓, would come under pressure if this holds HPQ ↓, would come under pressure if this holds QCOM ↓, would come under pressure if this holds

    Apple's inventory surge and warning on rising memory prices broaden the margin squeeze from PC manufacturers to consumer smartphone makers. Even with record iPhone revenue, Apple's 154% increase in inventory demonstrates that component inflation is tying up working capital and exerting margin pressure across hardware leaders. This reinforces Intel's earlier disclosures of memory-driven margin drag. The main open question is whether device makers can pass rising memory costs on to end buyers without hurting unit volumes.

    No material update

    Last development
    3 Nov 2026

    Evidence & history →

Assessments are AI-assisted and may be incomplete. Checked Tuesday–Saturday; dates are UTC. A company’s inclusion does not imply positive investment exposure.