Thesis: BIP/BIPC are the direct-asset (own-the-infrastructure) AI-power bet — the structural complement to the BAM deep-dive's fee-stream bet. Where BAM owns the management contract on Brookfield's infrastructure funds and earns ~1.2% of fund AUM, BIP owns the actual power-adjacent data-center assets on its own balance sheet and collects the operating cash flow. As of the FY2025 20-F, BIP runs a 150+ data-center, ~2.3 GW platform (1.2 GW operating + 1.1 GW future development), and the Data segment is now BIP's fastest-growing leg — segment FFO $502M in 2025 vs $333M in 2024 (+51% YoY) and $2,220M of 2025 growth capex (67% of the total $3,308M). BIP is the LP yield vehicle (~4.7% distribution, K-1); BIPC is the C-corp twin (1099, ~4.4% dividend) — same economic exposure, cleaner for US/IRA investors, and currently trading at a wider discount to its 52-week high.

This artifact extends the BAM deep-dive rather than repeating it. BAM concluded: "For a pure AI-power play, BIP or BEP are more direct." This is that play, sized.

Direct-asset vs fee-stream — the distinction the task asked for

The task's core claim — "the direct-asset bet is arguably more compelling than the fee-stream bet at current BAM multiples" — holds up directionally, with one important nuance. BAM trades at 22x forward P/E / 15x P/S, pricing in +24% revenue growth and FRE acceleration; there's "no structural rerate path absent multiple expansion" (BAM deep-dive). BIP/BIPC are valued on FFO/distribution and trade at 0.75x P/S (BIP) / 1.4x P/S (BIPC) — you are buying the cash-flowing assets, not paying a premium for a fee annuity. The nuance: BIP's AI-power exposure is diluted inside a diversified-infra body — the Data segment is only ~19% of segment FFO, so this is not a pure-play data-center landlord (that would be EQIX / DLR / a Vantage stand-alone). It's "AI-power inside a defensive-infra cash machine," which is a different risk/reward than the parabolic neocloud cohort (CRWV/CORZ/IREN) in the ai-power-bottleneck perspective.

Tape (2026-05-30 snapshot — narrow single-ticker fetch; returns not yet computed)

Neither BIP nor BIPC was in any watchlist before today (alt-managers covers only BAM/BN/BX/KKR/APO/ARES/OWL/TPG/CG). A narrow single-ticker fetch was run; a shared scan was not run (parallel-fanout collision rule), so RSI/SMA/return fields are not available — latest close + 52-week context only.

Read: BIP sits near its 52-week high (−3.2%) — the market has already re-rated the LP toward the data-center-growth story; entry here is not a discount. BIPC is the more interesting setup — down −19.7% from its high vs BIP's −3.2%, the same economic exposure available ~16pp cheaper relative to its own range. The BIPC discount is partly mechanical (C-corp twins historically swing wider than the LP and have, at times, traded at a premium to BIP when US/index demand was strong; the current discount is a reversal of that), and partly a negative-book-value optical (BIPC P/B is negative, −3.5x, because the exchangeable structure pushes most equity into the LP). The negative book value is a structure artifact, not insolvency — BIPC shares are exchangeable 1:1 into BIP units, so the economic value tracks BIP.

Returns not yet computed — no shared scan run (fanout collision rule). RSI/SMA20/50/200, 7D/30D/3M/1Y not computed in prose per "code does math." A future scan that includes BIP/BIPC should backfill these.

Financials (primary: BIP 20-F FY2025, SEC EDGAR; valuation: market data)

Massive returned "No results found" for both tickers — BIP is a Bermuda-domiciled LP filing a 20-F (not 10-K/10-Q), and Polygon/Massive does not cover it, the same fallback pattern the BAM deep-dive hit for BN and ARM. Fundamentals below are read from the BIP 20-F filed 2026-03-17 (accession 0001406234-26-000002, bip-20251231.htm, period 2025-12-31, CIK 0001406234) — primary source. Valuation multiples are from market-data provider figures.

Segment FFO — the Data segment is the growth engine

From the 20-F segment reconciliation (FFO attributable to BIP, US$M):

Data segment FFO +51% YoY while every other operating segment was flat-to-down — Data is doing all the marginal growth. Data is ~19% of operating-segment FFO (vs ~13% in 2024), and the mix shift is accelerating.

Growth capex — Data is where the money goes

Data was 67% of 2025 growth capex ($2,220M of $3,308M), up +87% YoY. Per the 20-F, the 2025 data-segment capex jump came largely from "construction progress at our semiconductor manufacturing foundries in the United States" plus the hyperscale build-out. This is a balance-sheet that is deliberately tilting toward data.

Segment assets & revenue (20-F, US$M, BIP share)

Data segment assets grew +46% ($9,358M → $13,622M) in one year and are now the largest single segment by asset base. Data debt attributable to the partnership grew from $6,546M to $10,714M — the segment is being levered up to fund the build, which is the structural risk (below).

Consolidated (market data, for valuation continuity)

(Consolidated revenue/debt include non-controlling interests; the 20-F FFO/assets tables above are the BIP-attributable economics and are the right denominator for the thesis.)

The data-center platform — what BIP actually owns (20-F primary)

The 20-F discloses the data-storage platform in concrete terms (Item 4.B, "Data — Overview"):

• "Over 150 operational data centers, with approximately 1.2 gigawatts (GW) of operating capacity today and an additional 1.1 gigawatts of future development capability" → ~2.3 GW total platform.
• 2025 organic growth: commissioned 220 MW of hyperscale capacity + 200 MW of new billings at the US retail-colocation operation.
• Named assets / JVs in the filing:
  • Vantage Data Centers — BIP's flagship hyperscale platform (Americas + EMEA + APAC), the core of the 1.2 GW operating base.
  • Cyxtera (now "Cyxtera Core Data Center Portfolio") — acquired out of bankruptcy, completed Jan 12, 2024; the US retail/wholesale colocation operation (the "200 MW of new billings"). Held partly through BIPC Holdings per the XBRL tags.
  • Data4 — European hyperscale data-center platform, acquired Aug 2023 for ~$600M; BIP holds an effective 19% interest.
  • Plus a "global data center developer" and US semiconductor manufacturing foundries under construction (the 2025 capex driver).
• The filing explicitly frames the build as "to support the build-out and development of artificial intelligence."
• Broader Data segment also includes ~309,000 telecom towers, ~80,000 km of fiber, 720,000 FTTP connections — the "data transmission & distribution" sub-segment, which is steady-eddy infrastructure, not the AI-power growth story.

Hyperscaler customer mix: the 20-F does not name individual hyperscaler customers (counterparties are confidential under the colocation contracts), but the structure is clear — Vantage is a hyperscale-wholesale landlord (the customer base is the major cloud/AI buyers), and Cyxtera is multi-tenant retail/wholesale colo. This is the "anchor-tenant / build-to-suit landlord" shape that the ai-power-bottleneck perspective formalized in its 2026-05-04 DGXX log entry ("the economics are more like a build-to-suit REIT lease… than a neocloud-shape"). BIP is the investment-grade, diversified version of that shape — the opposite end of the risk spectrum from the DGXX microcap.

Cross-reference: the ai-power-bottleneck landlord thread

The task asked to locate the perspective's "data-center-landlord thread." There is no perspective sub-folder literally named "landlord" — the perspective (2026-05-03-ai-power-bottleneck) is generation-side focused (BE/CORZ/GEV/CEG/VST — who makes the electrons). The landlord/real-estate thread lives as a named sub-category inside log.md, formalized in the 2026-05-04 DGXX/Cerebras entry:

"The 'anchor-tenant colo' sub-category formalizes as distinct from the existing BTC-miner-pivot neocloud cohort… The economics are more like a build-to-suit REIT lease (KRC-shape) than a neocloud-shape."

BIP slots directly into that thread as the diversified, investment-grade landlord — distinct from (a) the neocloud cohort (CRWV/CORZ/IREN — diversified-tenant + spot-pricing risk), (b) the anchor-tenant microcaps (DGXX — single-counterparty binary risk), and (c) the generation layer (BE/GEV/CEG). BIP is the lowest-beta way to own the physical real estate the GPUs sit in, with the AI-power exposure cushioned by utilities/transport/midstream cash flow. The BAM deep-dive already cross-cited BIP as "direct-asset-on-AI-power"; this deep-dive ratifies that with the 20-F segment data.

Perspective mutation note (do NOT auto-apply — producer-only): the landlord thread currently has no investment-grade-REIT-landlord representative; BIP/BIPC would be the natural addition. Flagged as a follow-up, not applied. → Proposal filed 2026-06-05 as perspective log-entry 2026-06-05-bip-data-center-landlord-key-ticker-candidate (names BIP as the key_ticker symbol, BIPC as the discounted exchangeable vehicle; surfaces the ~19%-of-FFO dilution caveat → reviewer may prefer watch_for / landlord-thread tag). README key_tickers still unedited — awaiting reviewer sign-off.

BIP vs BIPC — which vehicle (the task's "cleaner US-investor" question)

BIPC was created (2020) specifically to be the cleaner vehicle for US/index investors who can't or won't hold a K-1. Because the two are economically identical (1:1 exchangeable, same distribution per share), the rational preference is: hold whichever trades cheaper at the moment. Right now that is decisively BIPC — same assets, ~16pp cheaper relative to its own 52-week range, and no K-1 headache. The historical pattern is that BIPC oscillates between a premium and a discount to BIP based on index-demand flows; buying the discounted leg and (optionally) exchanging captures the spread. For the AI-power-direct-asset thesis, BIPC is the recommended entry vehicle.

(One caveat: BIPC's lower float / higher beta / negative-book optical means it can de-rate faster in a risk-off tape — the −19.7% drawdown already shows this. That cuts both ways: more downside captured, more upside on a re-converge.)

Setup

Conviction: medium. The asset quality and the Data-segment growth trajectory are real and primary-source-verified; the thesis is not a pure-play (Data is ~19% of FFO), and BIP itself is near its 52w high, so the edge is in vehicle selection (BIPC) and entry discipline, not in a screaming valuation gap. Distribution/dividend growth + Data-segment FFO compounding is the return engine; rate normalization is the multiple tailwind (beta ~1.0–1.3, lower than BAM's 1.25 / BN's 1.85 — this is the defensive Brookfield expression).

Primary recommendation: BIPC (cleaner US vehicle + currently the discounted leg).

BIPC setup:

• Entry zone: $38–42 (current $41.53 is at the upper end; a pullback toward the $38 area — roughly the mid-point between current and the $34.18 52w low — is the better-risk entry; the $34–35 52w-low zone is the conviction-add).
• Stop: $33 (below the 52w low of $34.18; a break there says the AI-data-center-FFO-growth thesis is being repriced, not just the rate environment).
• Target (base): $50 (~+20% from current; re-converge toward BIP's relative strength + Data-segment FFO continues +30–50% → consolidated FFO/unit +6–8%, distribution grows ~6%, multiple holds). This roughly closes the gap back to its own 52w high ($51.72).
• Target (bull): $62 (~+49%; Data segment keeps compounding at +50%, hyperscale commissioning accelerates, rate-cut cycle compresses the infra discount rate and marks the asset base up, BIPC re-rates to a premium-to-BIP as index demand returns).

BIP setup (companion — the LP yield vehicle):

• Entry zone: $35–38 (current $39.04 is near the 52w high $40.32 — wait for a pullback; the SMA-support / $34–35 region is the add zone, though SMAs are not yet computed here).
• Stop: $33 (below the structural $29.63–34 base).
• Target (base): $46–48 (~+18–23%; FFO/unit + distribution growth; less upside than BIPC because BIP starts near its high).
• Conviction: medium — same assets, but you're paying near-high and taking the K-1; prefer BIPC unless you specifically want the LP yield in a taxable account.

Code-does-math caveat: entry/stop/target levels are framed from 52-week range structure (the only computed reference available without a shared scan run). RSI/SMA-confirmed levels should be re-derived once BIP/BIPC are added to the standard scan universe.

Bull case

• Data segment is a genuine step-function grower inside a defensive body — FFO +51% YoY, 67% of growth capex, segment assets +46%. The AI-data-center build is showing up in the numbers BIP discloses, not just the narrative.
• 2.3 GW platform with 1.1 GW of future development already in hand — the growth runway is contracted/committed, not speculative. 220 MW of hyperscale commissioned in 2025 alone.
• Diversified-infra cushion — utilities/transport/midstream (~80% of FFO) are inflation-indexed, regulated, contracted cash flows that fund the data build and smooth the cycle. This is AI-power exposure you can hold through a drawdown.
• Rate normalization is a double tailwind — lower discount rates mark the asset base up and lower BIP's ~6% non-recourse borrowing cost. The BAM deep-dive's rate-cut thesis (Fed −75bp since Jan 2026, ~50bp more priced) applies directly to the asset values here.
• BIPC entry discount — same assets ~16pp cheaper relative to its own range, no K-1, exchangeable into BIP. A clean structural arb if the legs re-converge.
• Capital recycling track record — Brookfield's "sell mature, recycle into growth" model (BIP sold the Indian-tower stake, US gas pipeline interest, etc.) funds the data build without diluting unitholders excessively.

Bear case

• Not a pure-play. Data is only ~19% of FFO; if you want the AI-power-landlord theme undiluted, EQIX/DLR or a Vantage stand-alone is more direct. BIP's AI exposure is real but cushioned/diluted.
• Leverage is heavy and concentrated in the growth segment. Data-segment debt attributable to the partnership grew $6.5B → $10.7B in one year; consolidated total debt is $69B (BIP). The build is debt-funded at ~6%; a higher-for-longer rate scenario squeezes the FFO/unit math and the asset marks simultaneously.
• BIP trades near its 52-week high — no valuation cushion at $39.04 (−3.2%); the LP has already re-rated on the data story. Entry discipline matters.
• FX / EM exposure — material Brazil (regulated gas, ~7% local rate), India, UK assets introduce currency and regulatory drag; the 20-F repeatedly flags Brazilian-rate and FX headwinds to FFO.
• K-1 / structural friction on BIP — the LP is genuinely awkward for many US investors (UBTI, K-1 timing); part of why BIPC exists. Structure complexity is a real (if surmountable) deterrent.
• Net income is thin and volatile — BIP TTM net income $0.77B on $24B revenue (consol.); BIPC posted a TTM net loss (−$0.24B) with negative book value. The economics are an FFO/distribution story, not a GAAP-earnings story; investors who anchor on P/E (27–32x fwd) will find it expensive on the wrong metric.
• Re-contracting risk — the 20-F flags that some contracted revenues face re-contracting risk at finite contract expiry; colocation/hyperscale lease renewals are not guaranteed at current pricing.

Catalysts

• Q2 FY26 results (~August 2026) — Data-segment FFO trajectory + hyperscale-commissioning cadence will set the next leg of the data-growth multiple debate.
• New hyperscale-MW commissioning / customer announcements — the 20-F's "future 1.1 GW" converting to operating capacity is the concrete, trackable catalyst; named hyperscaler anchor announcements (à la the MSFT-BEP PPA on the renewable side) would validate the customer-mix question the filing leaves blank.
• US semiconductor-foundry construction milestones — the 2025 data-capex driver; ramp/commissioning of these is a 2026–27 FFO catalyst.
• Federal Reserve cadence — next FOMC mid-June 2026; the infra-discount-rate compression is the multiple tailwind shared with the whole alt/infra complex.
• Capital-recycling events — a mature-asset sale recycled into data/AI capex re-rates the growth mix (Brookfield's standard playbook).
• BIPC↔BIP spread re-convergence — if index/ETF demand returns to BIPC, the −19.7% discount-leg closes; a structural arb catalyst independent of the underlying.

Open questions / follow-ups

• BIP/BIPC watchlist coverage decision — neither is in any of the 41+ watchlists. They fit ai-infra (where the perspective routes), and BIP/BIPC are the investment-grade-landlord representative the ai-power-bottleneck landlord thread currently lacks. Producer flag only — describe, do not apply. (Companion to the BEP/BEPC follow-up the BAM deep-dive filed.)
• BEP / BEPC deep-dive — still open from the BAM deep-dive (2026-05-26). BEP is the renewable-PPA direct-asset twin to this infrastructure deep-dive (MSFT $10B / 10.5 GW PPA). Same direct-asset-vs-fee-stream frame; should be done to complete the Brookfield direct-asset triad (BIP + BEP + BN-balance-sheet).
• Hyperscaler customer mix — the 20-F redacts named counterparties; a follow-up via Vantage/Cyxtera press releases or sell-side channel checks would pin down concentration (single-customer vs diversified) — material to the re-contracting risk.
• Technical backfill — once BIP/BIPC are in a watchlist, run a scan to backfill RSI/SMA/returns so the entry/stop/target framing can be re-derived on technicals rather than 52-week-range structure.
• BIPC negative-book / exchange mechanics — confirm the exact exchange ratio terms and any tax consequence of a BIPC→BIP exchange before recommending the spread arb for execution.

Sources

• Tape: yfinance via narrow single-ticker fetch the desk's own tooling / fetch BIPC (2026-05-30 pull) → info, info. No watchlist coverage prior to today (filed as coverage-gap follow-up). Shared precompute not run (parallel-fanout collision rule); returns/RSI/SMA not computed.
• Fundamentals (primary): BIP 20-F filed 2026-03-17, SEC EDGAR accession 0001406234-26-000002 (bip-20251231.htm, period ending 2025-12-31, CIK 0001406234) — segment FFO, growth capex by segment, segment assets, data-storage operating/development capacity, named data-center assets (Vantage, Cyxtera, Data4). Verified via EDGAR full-text search (efts.sec.gov) + direct filing fetch.
• Fundamentals fallback (valuation only): yfinance info dict (2026-05-30) for market cap, EV, fwd P/E, P/S, P/B, dividend/distribution yield, beta, shares out, consolidated revenue/EBITDA/debt. Massive returned "No results found" for both BIP and BIPC (Bermuda LP / 20-F FPI filer, not in Polygon coverage) — same fallback pattern as BN/ARM in prior deep-dives; provenance recorded here per SOP.
• Sibling artifact (extends, does not repeat): 2026-05-26-bam-deep-dive — fee-stream side; this is the direct-asset side.
• Perspective context: 2026-05-03-ai-power-bottleneck (generation-side README + log.md); landlord/anchor-tenant thread is the 2026-05-04 DGXX/Cerebras log entry ("anchor-tenant colo sub-category… build-to-suit REIT lease shape"). BIP/BIPC = the investment-grade-landlord expression of that thread.
• Comp framing: BAM deep-dive comp-set context; pure-play data-center landlords (EQIX/DLR) referenced as the undiluted alternative — not independently re-priced here.

Note on data fallback: BIP is a Bermuda-domiciled LP reporting on a 20-F/6-K cadence (not 10-K/10-Q). Massive/Polygon's financials endpoint returned "No results found" for both BIP and BIPC. Primary segment economics were read directly from the FY2025 20-F on SEC EDGAR; yfinance supplied the valuation overlay. This is the same deliberate-fallback pattern used for BN (2026-05-26) and ARM (2026-05-25). All figures verify against the 20-F as filed; re-verify against the primary filing before any trade execution.