Know what people believe.
See what the sources show.
Understand what to watch next.

Occlusion follows market stories across filings, news, company pages, market data, archives and online discussion. It separates the public explanation from the evidence and turns it into a brief with the source trail, the likely paths and what would prove the read wrong.

Built for market stories, company shifts, policy changes, technology moves, and online consensus.
Built for
InvestorsFoundersAnalystsOperatorsResearchersIntelligence Teams

Most tools stop at the headline. Occlusion shows the public story and the evidence underneath it. It also shows what would make the read wrong.

Every brief begins with the question most tools avoid: what would make this wrong? Then it shows the public story, the evidence pushing against it, the scenarios worth watching and the sources behind the read. You can see why the conclusion exists and what would break it.

The facts are scattered. The story sounds settled.

A market story can look obvious because everyone repeats the same explanation. The useful read is often somewhere else: in a filing, a route change, a permit, a contract, a price move, a forum thread, or a source that quietly disagrees. Occlusion brings those pieces into one read.

What people repeat
AI is a chip shortageSpaceX IPO is inevitableA peace headline removes oil riskOpen source will crush software marginsStablecoins are mostly crypto
The useful gap
What changes the read
power contractssecondary-market liquidityshipping routesenterprise procurementTreasury reserve data
What Occlusion reads

It reads the story from both sides.

Direct evidence
Filings, company pages, official data, earnings calls, market data, contracts, permits, archives.
Public belief
News repetition, Reddit, YouTube, Hacker News, Wikipedia, X, forums, comments.
Market reaction
Prices, volume, prediction markets, rates, freight, energy, sector movement.
filingsnewscompany pagesarchivesmarket dataRedditYouTubeHacker NewsWikipediaXprediction marketsSECGitHubPolymarket
Product walkthrough
The readAI data-center growth is becoming power-constrained.
Source trail66 sources across filings, market data, news, and public discussion.
Where the read pointsUtilities, grid equipment, cloud margins, data-center REITs.
What would prove it wrongFaster interconnection approvals and signed power agreements.
Watch a read come together

Five stories. Five reads. Same structure every time.

Each demo below shows the same structure every Occlusion brief follows: what people believe, what the evidence shows, where the read points, and what would prove it wrong.

Read in progress
Utility filingsHyperscaler capexGrid operatorNVIDIA IR
What people believe
AI growth is mainly about chips.
What sources show
Power availability, interconnection delays, data-center permits, and utility procurement may become the harder bottleneck.
What most people miss
The constraint can move from silicon to electricity. If that happens, the best-positioned companies may be the ones with power access, grid relationships, and approved sites, not only the ones with GPU supply.
Where the read points
utilities with data-center exposuregrid equipmentpower purchase agreementscloud marginsdata-center REITssemiconductors tied to real deploymentregions with available capacity
What would prove it wrong
01Faster interconnection approvalsWatch: grid queues
02Signed large power agreementsWatch: power contracts
03Deployment timelines with no material delaysWatch: project schedules
Source trail
Utility filingsInterconnection queue backlog at 28 monthsSupports
Hyperscaler capexQ2 power-secured campus investmentSupports
Grid operatorLarge-load study delays increasingSupports
NVIDIA IRGPU demand trajectory reaffirmedPushes back
Source coverage
78%
Evidence movement
64%
Sector sensitivity
82%
The Engine

Search anything. See where the story breaks.

Type a company, market, or public story. Occlusion reads filings, news, community discussion, and market data. Then it shows you where the evidence and the public narrative stop matching.

Real-time source reading across 19 data channels
Fracture score measures story-evidence divergence
Every signal shows what to watch next
The Brief

Occlusion is not here to summarize the headline. It shows the read, the evidence behind it, and the case against it.

Every brief begins with the question most tools avoid: what would make this wrong? From there, it shows the public story, the evidence pushing against it, the scenarios worth watching, and the sources behind the read. You can see why the conclusion exists, and what would break it.

Falsifiers lead every brief, not buried at the bottom
Evidence gap stated in one sentence you can repeat
Named sources with stance: supports, pushes back, unclear
Real scenarios with mechanisms, not percentages pulled from air
The Terminal

Investigate anything yourself. From every angle.

Bring any headline, claim, or market story. Occlusion pulls live evidence, runs adversarial search, finds the counterargument, and tells you exactly what to watch. Run three investigations at once and switch between them.

Four views: the read, the tripwires, the counterargument, the play
Concurrent investigation tabs, like a browser for intelligence
Every source shows what it says and whether it agrees or pushes back
The brief is the product

A read you can inspect, challenge, and trace.

Every serious brief includes the story, the evidence, the possible paths, the affected markets, the source trail, and the conditions that would break the read.

TECHNOLOGY

AI data-center growth is becoming power-constrained

June 2026 · Cracking · 66 sources read
71Signal score
66Sources read
HighEvidence movement
30-90 daysTiming window
RisingConsensus lag
01Interconnection queues shorten across two consecutive quarterly reports
02Hyperscaler capex guidance shifts away from power-secured campuses
03Grid-exposed utility assets stop repricing while the story keeps spreading

AI growth is mainly framed as a chip and GPU supply story. Commentary focuses on semiconductor demand, model scaling, and data-center buildout as a direct extension of compute needs. The constraint is treated as silicon.

Public explanation

AI growth depends on GPU supply. Build more data centers, order more chips.

Sources show

Utility filings show 28-month interconnection backlogs. Power procurement, grid access, and data-center siting may be the binding constraint, not silicon.

The constraint can move from silicon to electricity. If power access becomes the binding limit, the best-positioned companies may be the ones with grid relationships, approved sites, and long-term power contracts, not only the ones with GPU supply.

Utilities with data-center exposureGrid equipment manufacturersPower purchase agreementsCloud marginsData-center REITsRegions with available capacitySemiconductors tied to real deployment
Power becomes the priced constraintBase case

Hyperscalers begin bidding for grid access. Utility capex cycles accelerate. Grid-exposed equities reprice ahead of consensus.

utilitiesgrid equipmentdata-center REITs
Chip narrative holdsStress case

New interconnection capacity comes online faster than expected. Power bottleneck fades before repricing.

semiconductorscloud compute
Both constraints bindFade case

Power and chips become joint constraints. The market splits attention and neither bottleneck resolves cleanly.

diversified infraboth sectors
SourceTitleStance
PJM InterconnectionFilingLarge-load queue study delays reach 28 monthsSupports
ReutersWireHyperscaler signs 1.2GW long-term power agreementSupports
Utility filingsFilingGrid-exposed utilities reprice ahead of broad coverageSupports
NVIDIA IROfficialQ2 guidance reaffirms GPU demand trajectoryPushes back
Data-center forumsDiscourseEngineers describe siting and power constraints firsthandSupports
EIAOfficialElectricity demand forecasts revised upwardSupports

Search gives you documents. News gives you updates.
Occlusion gives you the read.

Search

Useful when you already know what to look for.

News

Useful when the event is already visible.

Chat

Useful when you bring the material.

Occlusion

Useful when the story is still forming and the evidence is scattered.

The difference is the against-case. Occlusion does not only show what supports a read. It shows what would prove it wrong.

Built for people who need to understand the story before it settles.

Investors
See where public belief and evidence are starting to separate before the market fully digests it.
Founders
Track category shifts, competitor pressure, and investor narratives before they show up in meetings.
Analysts
Turn scattered sources into a brief with the source trail and against-case included.
Operators
Watch policy, procurement, supply chains, energy, regulation, and public reaction before they hit planning.
Researchers
Compare the public explanation with the evidence without losing the source trail.
Intelligence teams
Build a shared view of what changed, why it matters, and what to watch next.

A useful read tells you what could break it.

Most summaries sound confident because they leave out the hard part. Occlusion includes it. Every read shows what supports the story, what complicates it and what would prove it wrong.

That is why Occlusion works like a narrative debugger for markets.

Start with one story you care about.

Search a company, sector, market or headline. Occlusion will show what people believe, what the sources show, where the read points and what to watch next.