memory activation

activates

research_neededback to memory

trigger

detail

Tweet by @mkfilko

source tweet

status extracted

created 2026-06-11T19:39:27

score

detail

actionability pending

market confirmation pending

source tier pending

evidence pending

source

detail

@mkfilko / tier A

reliability 88

useful 0 / false+ 0

no source notes

tape

detail

market context unknown

affected IQE, AXTI, SIVE, INTC, TSM, COHR, LITE, MRVL, AVGO, ANET, CIEN, NOK, POET, LWLG

themes co-packaged optics, AI networking, advanced packaging, CPO winner map

signal

detail

signal pending

no tactical trigger

not created

outcomes

detail

0 snapshots

no outcome capture yet

trigger snippet52% confidence

RT @mkfilko: Last post I compared $IQE to the layer below it in the chain ($AXTI, the substrate). Today, the layer above. And the name some of you keep asking about: $SIVE. Substrate → epi → device → module. $IQE grows the epi. $SIVE builds laser devices. Sequential layers of the same AI optical chain. The numbers: $SIVE: ~$2.3B market cap on ~$30M trailing revenue. ~85x sales. EBITDA negative. $IQE: ~$650M market cap on ~$123M trailing revenue. ~5x sales. EBITDA positive. $IQE generates 4x the

The new evidence suggests delays, thermal issues, qualification problems, or yield friction. That does not settle the winner map, but it can push out the timeline for meaningful rack-scale CPO adoption.

tickersIQE AXTI SIVE INTC TSM COHR LITE MRVL AVGO ANET CIEN NOK POET LWLG

horizons1M: Timeline slippage can reduce near-term excitement without resolving the winner map. | 3M: If delays persist, the CPO adoption curve may move out and compress near-term idea urgency. | 6M: Outcome depends on whether delays are Intel-specific or field-wide. | 1Y: Longer-term structural implications need follow-through evidence.

report card

sanitized markdown

# Memory Activation Report Card

- Activation: fa0c5e19-978e-4848-81a1-4c970292f413
- Relationship: activates
- Market action: research_needed
- Evidence status: tweet_only_unverified
- Source: X / @mkfilko
- Source type: tweet
- Source tier: unknown
- Affected tickers: IQE, AXTI, SIVE, INTC, TSM, COHR, LITE, MRVL, AVGO, ANET, CIEN, NOK, POET, LWLG
- Actionability: pending
- Market confirmation: pending

## Trigger
RT @mkfilko: Last post I compared $IQE to the layer below it in the chain ($AXTI, the substrate). Today, the layer above. And the name some of you keep asking about: $SIVE. Substrate → epi → device → module. $IQE grows the epi. $SIVE builds laser devices. Sequential layers of the same AI optical chain. The numbers: $SIVE: ~$2.3B market cap on ~$30M trailing revenue. ~85x sales. EBITDA negative. $IQE: ~$650M market cap on ~$123M trailing revenue. ~5x sales. EBITDA positive. $IQE generates 4x the

## Reasoning
The new evidence suggests delays, thermal issues, qualification problems, or yield friction. That does not settle the winner map, but it can push out the timeline for meaningful rack-scale CPO adoption.

## Claims
- source_claim_only: CPO bottleneck is packaging, thermals, yield, and reliability at rack scale, not just waveguides. [3188f87f-baf4-4aa3-afc6-8bdf3a38f83b]
- source_claim_only: Bubble Boi argues Intel is the only credible CPO winner at rack scale because EMIB avoids large-interposer issues and Intel has internal silicon photonics / optical I/O work. [c96d459d-6985-42eb-9062-c0597114fb91]
- source_claim_only: In this view, Lumentum and Coherent are not the main CPO winners because the bottleneck is system-level packaging and integration rather than standalone optical components. [37c04a80-161d-428c-ade2-8e03921c03d6]

## Conditions
- narrative_shift: If a credible non-Intel CPO platform or customer win emerges for AI switches, it challenges the Intel-only CPO winner thesis. [d5f3ca42-7f95-4cad-9c49-9dd7293fdc52]

## Theses
- Intel-only CPO winner map depends on packaging bottleneck exclusivity [14155f8d-5239-49d9-882d-53c070d45d73]

## Gates
- no gates recorded

## Provider Usage
- openai/gpt-4o-mini memory_cross_reference: allowed, est $0.000581
- openai/gpt-4o-mini tweet_extraction: allowed, est $0.000401

## Signal And Telegram
- Signal: none
- Telegram: live / sent

## Outcomes And Feedback
- Scheduled outcome jobs: 0
- Feedback records: 0

_Private raw text/json and secrets are not included._

feedback

ready

relationship	actionability	evidence	market	notes	created
no feedback yet

shadow payload review

ready

judgment	confidence	reviewer	notes	created
no shadow payload review yet

decision log

ready

type	decision	rationale	created
no decision notes yet

trigger evidence

stored

Tweet by @mkfilkometadata_public

sourceX / @mkfilko

urln/a

published2026-06-11T19:39:08

raw textprivate, not exposed

matched claims

thesissource_claim_only

CPO bottleneck is packaging, thermals, yield, and reliability at rack scale, not just waveguides.

supportBubble Boi argues Intel is the only company likely to make co-packaged optics work at rack scale because EMIB avoids the reticle/yield/thermal/reliability issues of large CoWoS-style interposers, and because Intel has internal silicon photonics / optical I/O work.

tickersINTC TSM

confidence62%

thesissource_claim_only

Bubble Boi argues Intel is the only credible CPO winner at rack scale because EMIB avoids large-interposer issues and Intel has internal silicon photonics / optical I/O work.

tickersINTC

confidence60%

market_implicationsource_claim_only

In this view, Lumentum and Coherent are not the main CPO winners because the bottleneck is system-level packaging and integration rather than standalone optical components.

tickersLITE COHR

confidence55%

matched conditions

narrative_shiftactive

If a credible non-Intel CPO platform or customer win emerges for AI switches, it challenges the Intel-only CPO winner thesis.

ifA non-Intel player announces or wins a credible co-packaged optics platform for AI switches or rack-scale deployments.

thenThe Intel-only CPO thesis weakens and the CPO winner map may broaden to networking, optical component, or foundry participants.

positiveCOHR LITE MRVL AVGO ANET CIEN NOK POET LWLG TSM

negativeINTC

matched theses

Intel-only CPO winner map depends on packaging bottleneck exclusivityactive

The Intel CPO bull case in this memory cell depends on CPO being constrained by packaging, thermals, yield, and reliability where EMIB and internal silicon photonics create a differentiated path. Credible non-Intel CPO evidence would challenge the winner map.

tickersINTC TSM COHR LITE MRVL AVGO ANET CIEN NOK POET LWLG

watchAI switch customer wins, rack-scale CPO deployment proof, pluggable-versus-co-packaged clarification, thermal, yield, and reliability disclosures, Intel EMIB and silicon photonics roadmap updates

score reasoning

score pending

job history

activation_scoring / stub / succeeded

memory_cross_reference / stub / succeeded

memory_cross_reference / openai / failed_retryable / provider_quota

tweet_claim_extraction / stub / succeeded

tweet_claim_extraction / openai / failed_retryable / provider_quota

research tasks

no research tasks

signal gates

gate	passed	severity	score	reason
no signal gates recorded

provider usage

provider	model	task	gate	cache	cost	reason
openai	gpt-4o-mini	memory_cross_reference	allowed	miss	$0.000581	Market memory memory_cross_reference for evidence bbb1c975-437e-4131-b51a-246982b1f788.
openai	gpt-4o-mini	tweet_extraction	allowed	miss	$0.000401	Market memory tweet_claim_extraction for evidence bbb1c975-437e-4131-b51a-246982b1f788.

market data snapshot

ticker	price	day	premarket	5D	1M	rel vol	VWAP	tape	provider
market data unavailable or pending

outcome snapshots

ticker	horizon	activation price	snapshot price	return	volume change	captured	notes
no outcome snapshots yet

scheduled outcome jobs

ticker	horizon	scheduled	status	attempts	error
no scheduled outcome jobs

telegram payload

live / sent

PUBLIC UNAUTHENTICATED STREAM
UNVERIFIED SOCIAL / TWEET-ONLY / NOT PRIMARY VERIFIED

🧠 POSSIBLE TRADE IDEA - MEMORY CONNECTION
Not a buy/sell signal.

Theme: co-packaged optics / AI switches

New event:
RT @mkfilko: Last post I compared $IQE to the layer below it in the chain ($AXTI, the substrate). Today, the layer above. And the name some of you keep asking about: $SIVE. Substrate → epi → device → module. $IQE grows the epi. $SIVE builds laser devices. Sequential layers of the same AI optical chain. The numbers: $SIVE: ~$2.3B market cap on ~$30M trailing revenue. ~85x sales. EBITDA negative. $IQE: ~$650M market cap on ~$123M trailing revenue. ~5x sales. EBITDA positive. $IQE generates 4x the

Why this matters:
Prior memory from Bubble Boi argued Intel may be the only credible rack-scale CPO winner because packaging, thermal, yield, and reliability bottlenecks may favor EMIB plus internal silicon photonics / optical I/O.

Relationship: moves_timeline_backward
Source evidence status: tweet_only_unverified
Trigger source: tweet / @mkfilko
CPO classification: true_co_packaged_optics
Confidence: Medium-low

Prior claim:
"Intel may be the only credible rack-scale CPO winner because packaging, thermal, yield, and reliability bottlenecks may favor EMIB plus internal silicon photonics / optical I/O."

Possible implication:
The new evidence suggests delays, thermal issues, qualification problems, or yield friction. That does not settle the winner map, but it can push out the timeline for meaningful rack-scale CPO adoption.

Affected tickers: IQE, AXTI, SIVE, INTC, TSM, COHR, LITE, MRVL, AVGO, ANET, CIEN, NOK, POET, LWLG

What to verify next:
- What specifically slipped: packaging, thermals, reliability, or customer qualification?
- Does the delay affect Intel only or the broader CPO field?
- Is the issue tied to lab demo scale-up or production deployment?

Status:
Not a buy/sell signal. Review only.

Dashboard: https://consortium.capital/memory/trade-ideas/a180635b-ec50-458d-81a2-2c008002b6bd