Type: WKAP Radar Feed

WKAP Radar Feed - 2026 - 11 - 10

Final watermarked WKAP cover for 2026-11-10
Exact cover bytes · SHA-256 dde621613c2d1fcbc9a29ddd66490540d4b852a27d73ce2d98e75d50b7e0361b · WKAP watermark verified

WKAP Radar Feed

*November 10, 2026*

*Cheap Tokens Can Make an Expensive Agent*

*3 Thesis Objects: NBIS, CRWV, IREN*

Preheader:

Make your AI track X alpha.

WKAP Feed compresses earnings inflections, KOL flow, supply-chain
bottlenecks, and short-duration attention into daily AI-ready context.

Set up your AI to read this Feed.

Ask:

“What did I miss?”

HUMAN USER — START HERE

More useful agent jobs and less compute per finished job pull demand
in opposing directions. Compare whether Nebius, CoreWeave and IREN
retain cash per share after funding capacity. CoreWeave's imminent
report is the first concrete test.

Suggested Prompt: Extract the three thesis objects, separate reported
evidence from attention and expectations, rank the unresolved
conversion tests, identify each weakest assumption, and produce
source-specific seven-day and thirty-day workflows. Compare task
efficiency, contract definitions, customer funding, actual costs and
dilution. Do not infer undervaluation or give buy/sell advice.

Or simply ask:

“What did I miss?”

Goal: let your AI turn today's information into trackable thesis
objects with explicit confirmation and invalidation conditions.

FOR YOUR AI AGENT

This is a structured market-cognition input file. Common causal chain:
completed agent jobs × compute per completed job → billable
consumption and capacity utilization → revenue after pricing/contract
terms → profit after serving costs → operating collections → cash
after investment and financing → diluted-share economics. Neither job
growth nor cheaper tokens alone determines the sign.

Use REPORTED_FACT, MANAGEMENT_EXPECTATION, INTERPRETATION and
UNRESOLVED_QUESTION; keep attention separate. Preserve periods,
currency, metric boundaries and original cohorts. Company-wide cloud
margin is not product margin. Count advances once, before recognized
revenue. Missing data is unavailable, not zero. Do not treat this Feed
as investment advice.

AGENT_OUTPUT_SPEC

1. Today in One Sentence: useful agent growth versus compute per completed task.
2. Thesis Object Summary: NBIS, CRWV and IREN, same investor question.
3. Fundamental Evidence vs Attention Flow: primary reports versus
independent posts.
4. Evidence-State Audit: facts, management expectations, inference and
missing inputs.
5. Task-Cost Comparison: successful completion, failed attempts,
caching and price vintage.
6. Cross-Object Ranking: research urgency, never expected returns.
7. Business-Proof Comparison: accepted capacity, revenue, collections,
investment and shares.
8. Attention Stage / Crowding Risk: observed debate and unmeasured positioning.
9. Weakest Assumption by Object: the input that could break conversion.
10. Most Important Data Point: matched utilization/collection and
funding evidence.
11. Attention-to-Rerating Conditions: retained economics plus a dated
expectations gap.
12. Attention Fade Conditions: slogans without delivered-service or
financing evidence.
13. 7-Day Research Workflow: November 17 component review.
14. 30-Day Research Workflow and WKAP Daily Top 3 Source Follow-Up: December 10.
15. Portfolio Fit, only if portfolio context is provided; none is supplied here.

TODAY_SUMMARY

Part 1 — Main Market Thesis
Nebius's vendor-run coding example makes the demand distinction
concrete. GLM completed a task in four calls for about USD 0.008; a
lower-priced-per-token Qwen model used at least 28 calls, spent about
USD 0.055 and did not finish. This is one task, with September 5
pricing, not an independent production benchmark. Better task
execution can reduce consumption per job even while usefulness expands
adoption. Financing and availability remain gates between demand and
shareholder value.

Part 2 — Today’s Thesis Objects
NBIS Fundamental: June-quarter revenue USD 582.3m, group adjusted
EBITDA USD 236.2m and net loss USD 190.4m. Continuing operating cash
USD 2,246.1m was below USD 5,657.4m of cash property/intangible
investment. Attention: a bullish price-elasticity argument, not proof
of production economics. Primary:
https://nebius.com/newsroom/nebius-reports-second-quarter-2026-financial-results
. Task example: https://dev.nebius.com/cookbook/openhands-agent-canvas
.
CRWV Fundamental: June-quarter revenue USD 2,575m and approximately
USD 104bn backlog coexist with USD 679m operating cash, USD 6,422m
cash property/capitalized-software purchases and USD 640m net interest
expense. Attention: neocloud enthusiasm versus equity drawdowns.
Primary: https://investors.coreweave.com/news/news-details/2026/CoreWeave-Reports-Strong-Second-Quarter-2026-Results/
. Its November 9 Wayve agreement is for training, evaluation and
simulation, not proof of exclusively agent-inference demand:
https://investors.coreweave.com/news/news-details/2026/CoreWeave-to-Power-Wayves-AI-Driver-Training-in-Multi-Year-Agreement/default.aspx
.
IREN Fundamental: September-quarter AI-cloud revenue USD 200m
comprises USD 135m services and USD 65m leases. Management moved its
greater-than-USD 4bn ARR milestone to January 31, 2027, possibly
February. Attention: an aggressive capacity/earnings model whose
assumptions remain opinions. Primary:
https://www.sec.gov/Archives/edgar/data/1878848/000187884826000065/irenreportsq1-fy27result.htm
.

Part 3 — Attention Flow Today
Rand Group's November 10 discussion contrasts neocloud excitement with
stock outcomes: https://x.com/randgroup/status/2120073114512240910 .
It establishes neither business failure nor cheapness. Opposing
supply/pricing arguments and IREN replies were also inspected;
coverage is bounded, not a trend census.

Part 4 — The Better Question
Surface question: will agents consume more tokens? Better question:
does growth in paid completed workloads outrun compute efficiency, and
who retains the proceeds after capacity, financing and diluted shares?
Use the following worksheets and scenario branches to turn that
question into evidence.

MARKET_REGIME

RISK_TONE: Financing-sensitive AI expansion; no measured whole-market
risk-on/off claim.
MAIN_DRIVER: Agent usefulness versus task efficiency, followed by
capacity-to-cash conversion.
MARKET_CONTEXT: CoreWeave reports after today's U.S. close; its call
starts November 10 at 17:00 EST, November 11 at 05:00 ICT. Nebius
reports November 12 at 08:00 EST. These are scheduled events, not
observed results.
ATTENTION_ENVIRONMENT: Visible bullish and skeptical neocloud
arguments; holdings, consensus and crowding are unmeasured.
WKAP_VIEW: Research candidates only. Current equity values are not
normalized enterprise values or proof of underpricing. No new 30/30
qualification.

STATE_CHANGE_LEDGER

NBIS: Previous state—cheap-token demand hypothesis. New
REPORTED_FACT—vendor task outcomes reveal different calls and
completion, while June financials show substantial
prepayment-supported operating cash and investment. Current
state—mechanism plausible, production elasticity and retained profit
unresolved. Decision impact—request paid-workload growth alongside
efficiency, not token counts alone. Next change—November 12
production, margin, funding and share disclosures.
CRWV: Previous state—large committed-demand narrative. New
REPORTED_FACT—June backlog/revenue supported, but investment exceeded
operating cash; next financial test is imminent. Current state—demand
evidence stronger than retained-cash evidence. Decision
impact—prioritize reconciliation of earning capacity,
delivered-service collections and financing. Next change—actual
November 10 after-close report and filed cash-flow/share schedules.
IREN: Previous state—year-end 2026 commissioning milestone. New
MANAGEMENT_EXPECTATION—January 31, 2027, with possible February
extension after design/supplier changes. Current state—reported
services/lease growth supported; delivery timing and funding remain
conditional. Decision impact—do not convert contracted ARR into
current revenue. Next change—actual acceptance, commissioning and
funding disclosures; the target itself lies beyond this thirty-day
review.

CROSS_OBJECT_EVIDENCE_MATRIX

Worksheet A — completed-task economics
REPORTED_FACT: vendor GLM-5.3-Flash used 46,977 input tokens and 1,028
output tokens, including 452 reasoning tokens, in four calls; its
September 5 tariff was USD 0.15 input/USD 0.50 output per million
tokens. Qwen 30B used 527,265 input tokens and 28+ calls without
completion, about USD 0.055. Repeated conversation/tool context
contributes input; reasoning is already part of output. Do not count
it twice. A displayed UI USD 0.00 reflected a missing price entry, not
free inference.
INTERPRETATION: record successful jobs, failed attempts,
input/cache/output tokens, retries and price vintage before comparing
task costs. UNRESOLVED_QUESTION: production task mix, routing and
demand response are undisclosed. CoreWeave's current GLM product
tariff is USD 0.12 input/USD 0.02 cached input/USD 0.38 output per
million: https://forge.coreweave.com/wandb/inference/coreweave/cw_zai-org_GLM-5.3-Flash
. Different vintages prevent calling this a matched provider
benchmark; tariffs do not reveal margins.

Worksheet B — contract-to-cash funding waterfall
For each company, separate cash received, restricted balances, undrawn
committed facilities, unreceived contracted prepayments and targeted
uncommitted funding. Then match accepted/revenue-earning capacity,
recognized services or lease revenue, actual collections, equipment
payments, financing cash flows and share issuance. Track operating
costs including compensation; reconcile expensed interest with
cash-flow classification rather than subtracting it twice from
operating cash. Do not equate a funding source with investment return.
NBIS REPORTED_FACT: June operating cash included a USD 1,197.0m
deferred-revenue increase; stock compensation USD 102.5m. Management's
early payback and future-prepayment statements are expectations, not
realized returns. CRWV REPORTED_FACT: quarterly net financing cash USD
10,071m and stock compensation USD 165m; backing demand required
substantial external funding. IREN REPORTED_FACT: operating cash USD
529m includes USD 645m deferred-revenue increase; USD 106m stock
compensation and USD 498m ordinary-share proceeds belong in separate
operating/financing analyses.

Worksheet C — comparable periods, boundaries and contrary evidence
NBIS: June quarter; AI-cloud adjusted EBITDA margin 49.7% is
cloud-wide, not TokenFactory product margin. ARR annualizes its last
month's AI-cloud revenue. CRWV: June backlog includes remaining
performance obligations plus estimated committed revenue subject to
delivery/availability; it is not cash or recognized revenue. IREN:
September quarter;86% AI-cloud gross margin excludes depreciation and
precedes site-level costs. ARR annualizes commissioned contracted
GPU-hour pricing plus ancillary charges. Equal-looking ARR or margin
numbers are not comparable evidence of profitability.
Relevant counterexample: SoftBank's November 10 filing records a JPY
597.248bn procurement-revision loss, including non-refundable
advances. Capacity reservations can create obligations before earning
revenue; this does not establish identical terms at these three
suppliers. Primary:
https://group.softbank/media/Project/sbg/sbg/pdf/ir/financials/financial_reports/financial-report_q2fy2026_01_en.pdf
. Separately, the published call transcript records CFO Goto denying a
turbine-order default:
https://stockanalysis.com/quote/tyo/9984/transcripts/704471-q2-2027/ .
This is management's denial, not independent confirmation of every
obligation; the accounting charge and the denial must remain distinct.
Keep the three KOL interpretations separate from issuer measurements.

RANKED_UNANSWERED_QUESTIONS

1. CRWV: did earning capacity and delivered-service collections catch
up with the equipment/funding requirement? Retrieve the actual
report/cash-flow notes through https://investors.coreweave.com/ . This
can most quickly change research priority.
2. NBIS: do paid workloads grow enough to offset compute saved per
successful job, with retained margin after serving costs? Retrieve
November 12 materials through https://nebius.com/investor-events .
Missing disaggregation cannot prove elasticity in either direction.
3. IREN: what capacity has actually been commissioned and accepted,
and which funding is received/unrestricted rather than conditional?
Retrieve reports and filings through
https://www.iren.com/investors/reports and
https://www.iren.com/investors/sec-filings .
4. All three: what current enterprise-value, dilution and comparable
expectations baseline would establish a mismatch? Until defensible,
promotion remains unearned regardless of an appealing mechanism.

SCENARIO_BRANCHES

Confirm: more accepted earning capacity and paid workloads, improving
service collections relative to investment, and retained economics
after financing/share growth. Upgrade that business component; an
opportunity still requires a dated expectations gap. CRWV ranks first
for report urgency, not expected return.
Mixed: backlog/revenue beat but collections rely on new advances,
investment/funding gaps widen or share dilution offsets growth.
Strengthen demand evidence only; keep retained-value status pending.
Efficient tasks with rising adoption but no paid-consumption split are
also mixed, not automatic supplier upside.
Break: documented pushouts, acceptance failures, adverse funding terms
or persistent cost/cash deterioration contradict the specific
conversion claim. Downgrade that component and name the evidence; one
unavailable disclosure is not an invalidation. An IREN milestone
pushed beyond the newly stated window would weaken delivery even if
total contracted demand remained intact.

ATTENTION_TRADE_BOARD
Attention Trade Board

Object | Attention Stage | Attention Source | Why Today | Hard
Evidence | Narrative Gap | Crowding Risk | Likely Window | Fade Signal
CRWV | Earnings test ahead | Rand Group | After-close report | June
demand and funding | Commitment versus retained cash | UNMEASURED |
Actual report/filing | Repeated backlog without conversion
NBIS | Inference debate | Felix Y. | November 12 report ahead | Vendor
task and June financials | Jobs versus compute/job | UNMEASURED |
Report and production disclosures | Cheaper-token slogan only
IREN | Delivery/model debate | Joseph Invests | Revised milestone |
Filed services/lease split | Commissioned ARR versus revenue |
UNMEASURED | Actual acceptance/funding events | Model targets without
delivery

WKAP Attention View
Strongest fresh fundamental change: IREN's milestone revision and
explicit delivery/funding boundaries. Cleanest evidence-to-attention
asymmetry: CRWV's measurable funding waterfall against the headline
backlog debate. Largest optionality/evidence gap: IREN's aggressive
independent long-range model. Most crowded object: UNMEASURED;
counters cannot rank positioning. Highest fade risk: IREN if target
discussion outruns new commissioning evidence. Best durable-thesis
candidate: conditional on demonstrated retained cash, currently
unresolved for all three.

RADAR_OBJECT_INDEX

THESIS_OBJECT_1: NBIS
THEME: Agent task efficiency and retained inference economics
STATUS: active_trackable; research candidate, production elasticity pending
POSITION_CONTEXT: [not provided]
PRICE_AT_PUBLISH: USD 138.97, Nov 10, 2026, 8:05 AM EST premarket;
observed during the 20:00 ICT release check via
https://stockanalysis.com/stocks/nbis/ . Thin-session dated
observation, not execution price.
DATE_FIRST_ADDED_TO_RADAR: 2026-11-10, prospective only if released
SETUP_TYPE: Mechanism-to-cash source test
ATTENTION_STAGE: Competing elastic-demand arguments
ATTENTION_WINDOW: November 12 report and subsequent primary disclosure
KEY_QUESTION: Does paid-workload growth exceed task efficiency and
retain cash per share?

THESIS_OBJECT_2: CRWV
THEME: Agent-serving capacity, customer funding and cash conversion
STATUS: active_trackable; research candidate, retained economics pending
POSITION_CONTEXT: [not provided]
PRICE_AT_PUBLISH: USD 48.29, Nov 10, 2026, 8:06 AM EST premarket;
observed during the 20:00 ICT release check via
https://stockanalysis.com/stocks/crwv/ . Thin-session dated
observation, not execution price.
DATE_FIRST_ADDED_TO_RADAR: 2026-10-01, preserved earliest published card
SETUP_TYPE: Imminent financial conversion test
ATTENTION_STAGE: Neocloud theme versus equity debate
ATTENTION_WINDOW: November 10 after-close report and actual filing
KEY_QUESTION: Is demand being financed or becoming cash retained after
capacity costs?

THESIS_OBJECT_3: IREN
THEME: Contracted GPU capacity, delivery and financing conditions
STATUS: active_trackable; research candidate, commissioning/funding pending
POSITION_CONTEXT: [not provided]
PRICE_AT_PUBLISH: USD 20.99, Nov 10, 2026, 8:06 AM EST premarket;
observed during the 20:00 ICT release check via
https://stockanalysis.com/stocks/iren/ . Thin-session dated
observation, not execution price.
DATE_FIRST_ADDED_TO_RADAR: 2026-11-10, prospective only if released
SETUP_TYPE: Delivery and definition counterexample
ATTENTION_STAGE: Model optimism meets revised timetable
ATTENTION_WINDOW: Actual commissioning/funding updates; target beyond
thirty days
KEY_QUESTION: Which commitments become accepted service and cash
without eroding per-share value?

THESIS OBJECTS

THESIS_OBJECT_1 — NBIS
CARD_ID: WKAP-RADAR-2026-11-10-NBIS
CARD_TITLE: Better Agents Can Use Less Compute Per Finished Job
TYPE: trackable_wow
THEME: Agent task efficiency and retained inference economics
STATUS: active_trackable; research candidate, production elasticity pending
POSITION_CONTEXT: [not provided]
PRICE_AT_PUBLISH: USD 138.97, Nov 10, 2026, 8:05 AM EST premarket;
observed during the 20:00 ICT release check via
https://stockanalysis.com/stocks/nbis/ . Thin-session dated
observation, not execution price. Separately, provider prior-close
equity value approximately USD 38.77bn; not current premarket
enterprise value or consensus.
DATE_FIRST_ADDED_TO_RADAR: 2026-11-10, prospective only if released
ATTENTION_STAGE: Competing elastic-demand arguments
ATTENTION_WINDOW: November 12 report and subsequent primary disclosure

WHY_A_READER_CARES_NOW: A cheaper successful task can improve adoption
while reducing compute per job; the supplier benefits only from the
net paid workload and retained economics.
PREVIOUS_STATE: Cheap-token demand narrative without a production bridge.
CURRENT_STATE: Vendor task mechanism and historical cloud growth
supported; production elasticity and product profit undisclosed.
DECISION_OR_MONITORING_PAYOFF: Request the paid-workload/efficiency
bridge and use the funding waterfall before extrapolating cash per
share.

THESIS_SUMMARY
June financials show real growth but major capacity investment.
Continuing operating cash minus disclosed cash property/intangible
investment is negative USD 3,411.3m before other investing activities.
This is a specifically defined spread, not a company-labeled
free-cash-flow measure.

WKAP_ANGLE
The surface-level frame: falling token prices necessarily mean
explosive compute demand.
The alternative frame: usefulness and efficiency pull billable
consumption in different directions.
The key research question: which production workloads expand after
both effects and retain profit after serving and capital costs?

CORE_THESIS
Reported production-inference growth does not disclose TokenFactory
margin or agent-specific revenue. June unrestricted cash USD 8,042.1m
provides substantial liquidity against the funding gap. Quarterly ATM
issuance 12.7m shares and stock compensation require a per-share
bridge with matched share denominators. Sources: financial statements
and shareholder letter linked from
https://nebius.com/newsroom/nebius-reports-second-quarter-2026-financial-results
.

ATTENTION_TRADE_FRAME
Attention Source: https://x.com/FelixYangA8/status/2119310604343333300
Why Today: November 8 argument directly challenges the inference-price
bearish case before November 12 results.
Attention Stage: Independent bullish hypothesis, not company guidance.
Attention vs Evidence
Hard evidence: task results and reported cloud/cash/share data.
Primary sources: issuer report above and vendor cookbook in TODAY_SUMMARY.
Attention / interpretation: cheaper tokens stimulate demand while
supply constraints support rentals; production elasticity remains
unproven.
Attention Path: usefulness → paid jobs → net compute → retained margin
→ funded capacity → cash per diluted share.
Attention Asymmetry: completion economics are less visible than
per-token tariffs.
Crowding Risk: UNMEASURED; visible replies do not establish holdings
or consensus.
What Could Sustain Attention: disclosed paid usage, efficiency,
utilization and product economics.
What Could Make Attention Fade: new tariffs or generalized agent
enthusiasm without conversion evidence.
Attention-to-Thesis Conversion: production support for the mechanism
plus a defensible dated expectations gap.

WEAKEST_ASSUMPTION: demand expansion exceeds efficiency gains at
economics the supplier retains.
MOST_IMPORTANT_DATA_POINT: comparable paid workload, compute intensity
and retained serving margin.
NEXT_DATA_POINT: November 12 Q3 release/call via
https://nebius.com/investor-events ; record actuals only when
available.

THESIS_OBJECT_2 — CRWV
CARD_ID: WKAP-RADAR-2026-11-10-CRWV
CARD_TITLE: A Backlog Can Be Financed Before It Becomes Cash
TYPE: trackable_wow
THEME: Agent-serving capacity, customer funding and cash conversion
STATUS: active_trackable; research candidate, retained economics pending
POSITION_CONTEXT: [not provided]
PRICE_AT_PUBLISH: USD 48.29, Nov 10, 2026, 8:06 AM EST premarket;
observed during the 20:00 ICT release check via
https://stockanalysis.com/stocks/crwv/ . Thin-session dated
observation, not execution price. Separately, provider prior-close
equity value approximately USD 27.01bn; not current premarket
enterprise value or consensus.
DATE_FIRST_ADDED_TO_RADAR: 2026-10-01, preserved earliest published card
ATTENTION_STAGE: Neocloud theme versus equity debate
ATTENTION_WINDOW: November 10 after-close report and actual filing

WHY_A_READER_CARES_NOW: Today's report can separate evidence of
committed demand from the cost and financing required to deliver it.
PREVIOUS_STATE: June backlog and large expansion supported, cash
retention pending.
CURRENT_STATE: No new report assumed; historical financials provide
the comparator.
DECISION_OR_MONITORING_PAYOFF: Classify the next result as demand
support, cash conversion, mixed or adverse rather than treating a
backlog beat as all four.

THESIS_SUMMARY
June revenue and backlog support customer demand. Adjusted operating
margin 5%, net loss USD 626m and the funding requirement qualify the
economic conclusion. Net interest expense is not a substitute for cash
interest; depreciation and stock compensation cannot disappear merely
because adjusted EBITDA is high.

WKAP_ANGLE
The surface-level frame: a valuable AI theme or falling stock
establishes cheapness.
The alternative frame: earning capacity, service collections and
financing terms control retained value.
The key research question: does incremental delivered revenue reduce
dependence on external capital without unacceptable dilution?

CORE_THESIS
Compare the USD 5,743m June operating-cash/equipment-software spread
with like-defined actuals, using Worksheet B's one-count advance rule.
Active power 1.5 GW and contracted power 3.7 GW are different stages,
neither alone proving accepted revenue. Strong backlog is meaningful
contrary evidence to a simple demand-collapse story, but
delivery/availability conditions and funding remain material. Primary:
https://investors.coreweave.com/news/news-details/2026/CoreWeave-Reports-Strong-Second-Quarter-2026-Results/
.

ATTENTION_TRADE_FRAME
Attention Source: https://x.com/randgroup/status/2120073114512240910
Why Today: November 10 theme-versus-stock debate precedes the actual
financial test.
Attention Stage: Independent market interpretation; historical
percentage claims are not adopted.
Attention vs Evidence
Hard evidence: June report, current GLM tariff and issuer calendar.
Primary sources: linked financial release;
https://investors.coreweave.com/ for actual new disclosures.
Attention / interpretation: excitement can coexist with poor equity
performance; no verified explanation of price moves follows.
Attention Path: contracted demand → available capacity → accepted
service → collected cash → financing/dilution → per-share value.
Attention Asymmetry: the contract headline is clearer than the funding
waterfall.
Crowding Risk: UNMEASURED; post engagement is not positioning.
What Could Sustain Attention: earning-capacity growth with improving
collections and funding terms.
What Could Make Attention Fade: commitments repeatedly outpace
deliverable economics.
Attention-to-Thesis Conversion: matched cash/share improvement and a
current expectations mismatch, not a low-looking equity value alone.

WEAKEST_ASSUMPTION: large commitments convert into profitable service
before financing costs absorb their value.
MOST_IMPORTANT_DATA_POINT: delivered-service cash conversion relative
to investment and capital claims.
NEXT_DATA_POINT: actual November 10 after-close result; call 17:00
EST/November 11 05:00 ICT. Calendar:
https://investors.coreweave.com/news/news-details/2026/CoreWeave-Announces-Date-of-Third-Quarter-2026-Financial-Results-and-Conference-Call/default.aspx
.

THESIS_OBJECT_3 — IREN
CARD_ID: WKAP-RADAR-2026-11-10-IREN
CARD_TITLE: Commissioned ARR Still Has to Become Revenue
TYPE: trackable_wow
THEME: Contracted GPU capacity, delivery and financing conditions
STATUS: active_trackable; research candidate, commissioning/funding pending
POSITION_CONTEXT: [not provided]
PRICE_AT_PUBLISH: USD 20.99, Nov 10, 2026, 8:06 AM EST premarket;
observed during the 20:00 ICT release check via
https://stockanalysis.com/stocks/iren/ . Thin-session dated
observation, not execution price. Separately, provider prior-close
equity value approximately USD 9.09bn; not current premarket
enterprise value or consensus.
DATE_FIRST_ADDED_TO_RADAR: 2026-11-10, prospective only if released
ATTENTION_STAGE: Model optimism meets revised timetable
ATTENTION_WINDOW: Actual acceptance/funding updates; January/February
target beyond thirty days

WHY_A_READER_CARES_NOW: A revised commissioning date and mixed
services/lease revenue change how an apparently comparable AI-cloud
metric should be interpreted.
PREVIOUS_STATE: Year-end milestone and broad funding shorthand.
CURRENT_STATE: New timetable, explicit ARR definition and funding
perimeter disclosed.
DECISION_OR_MONITORING_PAYOFF: Test capacity and financing stages
separately before adopting a long-range earnings model.

THESIS_SUMMARY
AI-cloud revenue growth and approximately USD 20.3bn backlog support
demand. Neither establishes exclusively agent inference or fully
burdened profit. Total September revenue USD 228m includes USD 28m
Bitcoin revenue; the company is not a pure agent business.

WKAP_ANGLE
The surface-level frame: high gross margin and fully funded capacity
settle the thesis.
The alternative frame: margin exclusions, conditional funding and
acceptance timing still control cash.
The key research question: does commissioned contracted pricing become
recognized service/lease revenue after all costs and capital claims?

CORE_THESIS
“Fully funded” refers to the 0.3 GW program supporting the
greater-than-USD 4bn target and includes restricted cash, undrawn
facilities and contracted advances not yet received. The broader
remaining FY2027 plan includes USD 6bn targeted funding subject to
market conditions. September equipment/intangible payments total USD
4,889m across three disclosed categories versus USD 529m operating
cash, a negative USD 4,360m specified spread before other investing
uses. Do not label it company free cash flow. A particular contract's
prepayment supports that project, not every project or
replacement/dilution costs. Primary:
https://www.sec.gov/Archives/edgar/data/1878848/000187884826000065/irenreportsq1-fy27result.htm
.

ATTENTION_TRADE_FRAME
Attention Source: https://x.com/Joseph_Invests/status/2119579585784262668
Why Today: November 9 independent model makes capacity, financing,
contract-mix and valuation assumptions inspectable.
Attention Stage: Bullish long-range model, not issuer guidance or consensus.
Attention vs Evidence
Hard evidence: filed quarter, ARR definition, delay and funding conditions.
Primary sources: filed exhibit above; actual issuer reports/SEC links below.
Attention / interpretation: selected forward multiple, annual
additions and cheaper borrowing are assumptions. Model return targets
are not adopted.
Attention Path: equipment → power/commissioning → customer acceptance
→ recognized revenue → retained cash → diluted shares.
Attention Asymmetry: annualized contracted capacity is easier to
publicize than actual recognition and funding conditions.
Crowding Risk: UNMEASURED; bullish replies are not a position census.
What Could Sustain Attention: accepted capacity, disclosed cash
receipt and attractive funded delivery economics.
What Could Make Attention Fade: milestone slippage or repeated model
targets without new primary evidence.
Attention-to-Thesis Conversion: actual conversion and a credible
expectations baseline, not comparable-looking ARR alone.

WEAKEST_ASSUMPTION: delivery, replacement spending and financing allow
modeled capacity to create cash per diluted share.
MOST_IMPORTANT_DATA_POINT: commissioned/accepted capacity reconciled
with recognized revenue and actual funding availability.
NEXT_DATA_POINT: actual updates through
https://www.iren.com/investors/reports and
https://www.iren.com/investors/sec-filings ; no promised January
milestone completion within December 10 review.

CROSS_OBJECT_ATTENTION_COMPARISON
Cross-Object Attention Comparison

Rank | Object | Attention Asymmetry | Evidence Quality | Catalyst
Clarity | Crowding Risk | Attention Window | Conversion Potential
1 | CRWV | Backlog versus funding waterfall | Historical primary; new
result pending | Imminent report | UNMEASURED | November 10/next
actual filing | Conditional, not expected return
2 | NBIS | Completed jobs versus compute/job | Vendor example plus
group financials | November 12 report | UNMEASURED | Report/production
disclosures | Production bridge unresolved
3 | IREN | Contracted ARR versus accepted revenue | Filed definitions
and delay | Updates, target beyond horizon | UNMEASURED |
Acceptance/funding events | Timing and per-share funding unresolved

Cleanest Attention Trade: CRWV for a concrete next-report question,
not valuation. Most Evidence-Backed Attention Trade: IREN's
metric/timetable correction is the freshest filed change. Most Crowded
Attention Trade: UNMEASURED. Highest Fade Risk: IREN if long-range
model excitement lacks delivery evidence. Best Candidate to Become a
Durable Thesis: no unconditional selection; require conversion and
price/expectations support. Rankings concern research priority and can
differ from business confidence.

7_DAY_RESEARCH_WORKFLOW

NBIS — 7-Day Checks
Read November 12 actual report, letter and call through
https://nebius.com/investor-events . Compare paid workload, compute
intensity, serving economics, cash investment and diluted shares.
Classify absent production detail UNAVAILABLE. Do not convert a
workload-growth statement into measured demand elasticity or
substitute cloud-wide margin for product profitability.

CRWV — 7-Day Checks
Through https://investors.coreweave.com/ retrieve the actual November
10 report/filing. Match quarter definitions, backlog/RPO,
active/accepted capacity, advances, delivered-service collections,
investment, cash interest and financing/share schedules. A revenue
beat with a wider funding gap is mixed. Preserve strong demand
evidence rather than recast every negative cash spread as demand
failure.

IREN — 7-Day Checks
Follow actual reports/SEC navigation for commissioning, acceptance,
services/lease recognition, cash received and available facilities.
Keep the January 31/possible February target in 2027. Short- and
long-duration contract pricing cannot be compared as equivalent annual
price improvements. A documented new delay changes delivery status;
silence alone does not.

Cross-Object — 7-Day Checks
November 17 is this dated cohort's seven-calendar-day review. Freeze
business component verdicts before inspecting returns. QQQ is
precommitted; baseline is the first completed U.S. regular close after
actual publication, still future during preparation. Retrieve baseline
and endpoint together using the original
adjusted-price/corporate-action method and disclose any alternate
provider with both endpoints from that provider. Preserve original
cohorts and benchmarks; CRWV's first-added date does not reset them.
Missing prices remain unavailable, never zero. No return attribution
or factor-neutral alpha claim.

Primary-source retrieval registry: NBIS:
https://nebius.com/investor-events ; CRWV:
https://investors.coreweave.com/ ; IREN:
https://www.iren.com/investors/reports . Follow actual dated report
and filing links, never invented future URLs.
Attention-source retrieval registry: NBIS:
https://x.com/FelixYangA8/status/2119310604343333300 ; CRWV:
https://x.com/randgroup/status/2120073114512240910 ; IREN:
https://x.com/Joseph_Invests/status/2119579585784262668
Evidence-state snapshot: NBIS=vendor mechanism/historical cloud
supported,production elasticity pending;CRWV=committed demand
supported,retained cash pending;IREN=services/lease growth
supported,delivery/funding
conditional;classification=trackable_wow;rule=component business
status separate from price;source-rule=primary facts,management
expectations,interpretation,unresolved questions,attention separate.

30_DAY_RESEARCH_WORKFLOW

NBIS — 30-Day Checks
Validate comparable paid-workload and cost disclosures. Upgrade the
conversion component only with retained profit/cash evidence after
investment and shares. Downgrade on adverse utilization, economics or
financing evidence. Invalidate the specific efficiency-adjusted growth
claim if actual comparable production data contradict it; absent
disaggregation leaves it pending.

CRWV — 30-Day Checks
Validate accepted earning capacity and the one-count
collections/funding waterfall. Upgrade on improving retained
conversion with sustainable financing. Downgrade for delivery or
financing deterioration. Invalidate the specified per-share conversion
thesis on persistent documented economics inconsistent with it, rather
than a stock decline alone.

IREN — 30-Day Checks
Validate recognition, acceptance and actual available funding against
the filed definitions. Upgrade on delivered economics without
value-eroding dilution. Downgrade on new delays, adverse terms or
recognition shortfalls. Invalidate the specific commissioning/funding
claim if primary evidence contradicts it. December 10 cannot judge
completion of a January/February 2027 milestone prematurely.

Cross-Object — 30-Day Checks
On December 10 retain the same claims, QQQ baseline and adjusted-price
method. Report confirmed, weakened, invalidated, pending and
unavailable business components separately from stock/benchmark
returns. NO_NEW_DATA after a bounded issuer search preserves known
facts and schedules the next source check within existing wakes.
Operating success cannot replace an adverse stock outcome, or vice
versa.

WKAP DAILY TOP 3

Three market sources worth feeding into today’s market chat. Not
required reading — WKAP has already extracted the signal.

1. @FelixYangA8 — Cheaper Inference Can Expand Demand
URL: https://x.com/FelixYangA8/status/2119310604343333300
WKAP signal: A serious opposing argument links lower token costs,
demand expansion and constrained supply. It is a hypothesis, not
measured production elasticity.
Why it matters today: Nebius reports November 12; a two-sided
task-efficiency test is more useful than assuming either demand
explosion or collapse.
Themes/tickers: Agent inference, paid workload, NBIS.
Question to ask: Which comparable production disclosures would show
adoption outgrowing compute savings without losing retained margin?

2. @randgroup — A Valuable Theme Can Still Hurt Equity Holders
URL: https://x.com/randgroup/status/2120073114512240910
WKAP signal: The theme-versus-stock debate points toward capital
requirements and the price paid; the post itself proves neither cause
nor valuation.
Why it matters today: CoreWeave's actual after-close report can turn
the debate into a collections, investment and financing comparison.
Themes/tickers: Neocloud funding, capacity conversion, CRWV.
Question to ask: Would a new commitment change delivered-service cash
or merely finance more capacity before delivery?

3. @Joseph_Invests — Make the Capacity Model's Assumptions Visible
URL: https://x.com/Joseph_Invests/status/2119579585784262668
WKAP signal: Separate assumed capacity additions, cheaper borrowing,
contract mix and forward earnings multiple from issuer guidance. No
model return target is adopted.
Why it matters today: IREN's revised 2027 timetable, services/lease
split and conditional funding must constrain any long-range
projection.
Themes/tickers: Commissioned ARR, recognized revenue, financing, IREN.
Question to ask: Does the model still create cash per diluted share
after realistic commissioning, replacement spending and funding
conditions?

Edition_version: 1
Cover_image_source_url:
https://pbs.twimg.com/media/HWwWWYNbgAAUwW-?format=jpg&name=medium
Cover_image_sha256:
dde621613c2d1fcbc9a29ddd66490540d4b852a27d73ce2d98e75d50b7e0361b
Cover_logo_verified: true

Agent-readable facts

artifact_type
radar
market_date
2026-11-10
canonical_url
https://wkap.ai/radar/wkap-radar-feed-2026-11-10.html
content_sha256
99c054dfb0f0ee6235a5e50db4afae3748f763d3805e97cc152e27484e06119f
content_sha256_covers
sha256 over UTF-8 text: radar\n{market_date}\n{title}\n{body_text}\n{edition_version}\n{cover_image_sha256}\n{cover_logo_verified}\n{challenge_invitation}. It does not cover rendered HTML, proof panels, manifests, or OpenTimestamp status.
github_file_url
https://github.com/wkapai/wkap-ledger/blob/main/radar/wkap-radar-feed-2026-11-10.html
github_commit_sha
af8af02d17ac6caed62749030f335df901cd0e5f
manifest_url
https://github.com/wkapai/wkap-ledger/blob/main/manifests/radar-85.json
opentimestamp_status
stamped
opentimestamp_proof_url
https://github.com/wkapai/wkap-ledger/blob/main/timestamps/radar-85.json.ots
title
WKAP Radar Feed - 2026 - 11 - 10