Video summary

How Much Should Indians Invest Abroad? | AI Bubble, US Markets & India’s Future ft. Swanandd Kelkar

Main summary

Key takeaways

Finance

Finance-specific takeaways (markets, investing, macro, risk)

1) How much Indian investors should allocate outside India

Core debate: diversification vs. staying fully invested domestically

  • The discussion frames a trade-off between:
    • Diversification (reducing concentration and improving diversification metrics)
    • Staying domestically invested (higher long-run “home” growth potential)

Pro-India long-run argument (as cited)

  • ~11–12% nominal CAGR has been cited for Indian markets historically.
  • The argument is that “building blocks” like double-digit nominal GDP growth remain intact and support long-run returns that can outperform inflation.

Key caution on diversification rationale

  • If an investor has:
    • No foreign-currency liabilities/outflows
    • The ability to stay invested through drawdowns for 15–25 years
  • …then diversification may be less necessary.
  • Behavioral risk is highlighted: many investors struggle to endure deep equity drawdowns (example referenced: 2007–09 ~50% drawdown).

When international diversification can help

For investors (or institutions) that want:

  • Steadier / less-correlated outcomes, and/or
  • Controlled volatility and correlation

…then international equities and multi-asset diversification may help, using metrics such as:

  • Correlation / “un-correlation”
  • Volatility / standard deviation

Personal allocation examples discussed

  • One speaker (Swan): ~20–25% of equity allocation in international equities (personal).
  • Ballpark guidance mentioned: ~25–30% as a personal guideline.
  • Another example: 60/40 India/outside with ~80–90% equity inside the overall portfolio (as described in the conversation).

Recommendation framing

  • “US vs them” patriotism framing is called out as misguided.
  • Final stance: there’s no single set answer—it’s case-dependent (liabilities, time horizon, and volatility tolerance).

2) Framework for global allocation (“alien from space” approach)

Starting point: anchor to a global index proxy

  • Begin with broad global equity exposure using an MSCI ACWI-type fund/index proxy (referred to as “MSCI Acquire/Equi”).

Adjust regional weights relative to the index

  • Determine baseline regional weights from the index.
  • Then adjust using macro/regional research over time.
  • A specific example was given:
    • If MSCI ACWI is ~65% US, then the approach suggests underweighting the US (expecting lower allocation to US than the index over the next ~10 years).

Iteration approach

Step-by-step (as described):

  1. Allocate first to a known global equity index fund (MSCI ACWI / ACWI-like).
  2. Review index composition (e.g., ~65% US).
  3. Make an initial tactical/macro adjustment (e.g., underweight US vs the benchmark).
  4. Reassess and rebalance as understanding evolves through ongoing regional research.

3) AI “bubble” / US markets / South Korea: supply-chain thesis + risks

“Gangotri” (starting point) for the AI rally

  • Thesis: AI capex starts with hyperscalers building data centers (examples: Anthropic, OpenAI, Google).
  • Key estimate:
    • Hyperscaler capex this year: ~$750–800B
    • Expected to reach ~$1T within 2–3 years

Downstream chain described

  • Hyperscalers → chip manufacturers/designers → memory → energy → uranium → “copper” (commodities exposure) → broader infrastructure ecosystem.

South Korea / memory bottleneck

  • Memory bottleneck highlighted:
    • Samsung
    • SK hynix
Risk/reversal argument when valuations look extreme
  • When multiples look stretched (described in the discussion as “5/6/7 EV-to-E … type multiples”):
    • “Experienced people” suggest timing risk—“this is time to sell” is referenced as a typical pattern in deep-cycle memory industries.
What’s “different this time”
  • AI memory demand (for AI servers) is described as different from prior consumer cycles (phones/laptops/notebooks).
  • The supply response is implied to be too slow:
    • No capacity planned quickly enough → potential vertical spike in memory prices.
How the cycle could end (two mechanisms)
  1. Demand destruction
    • Evidence cited: smartphone and notebook shipments down double digits.
    • Higher consumer prices suppress purchases.
  2. New supply
    • Not yet obvious: the bottleneck is framed as caused by fast demand pull and limited supply ramp.
  • Commodities analogy: memory treated like a commodity cycle (demand vs supply response).

“Cinderella party” / valuation & risk management cues

  • Warning: don’t extrapolate the “party” indefinitely.
  • Korea bubble-like signs described:
    • Parabolic price action and frenzied participation:
      • SK hynix doubled in a month
      • ~8x in ~a year and a bit
      • Cost up ~3x over a similar timeframe (as claimed)
    • Heavy retail leverage:
      • 2x and 3x leveraged ETFs (no tickers specified)
      • Margin buying from the US and Korea
Positioning/reaction
  • In an emerging markets portfolio, “they own”:
    • Samsung
    • TSMC
    • SK hynix
  • They mention taking some money off because moves were too parabolic, while remaining constructive long-term.

Explicit “exit” / derailment risks for the AI capex-bubble scenario

Key macro risk: liquidity squeeze
  • Rapid interest rate increases / liquidity tightening in US/global markets.
  • Indicator cited:
    • US 10-year Treasury yield hit ~4.5% (“on Friday the US tenure hit 4.5% again”).
  • Bubble logic: bubbles need liquidity/animal spirits.
  • Even if hyperscalers have strong balance sheets, the downstream chain can break.
Additional risks
  • Capex return pressure / RoI divergence
    • Hyperscaler profit/cash flow may diverge from expectations as spending continues (“someone has to pay the piper”).
  • Regulatory/social backlash
    • Reference to prior backlash themes (AI controversy context).
  • Technological breakthrough
    • Memory could be disrupted by new processes reducing requirements.
  • Timing uncertainty
    • “What bubble means / timing it” is described as very hard and not guaranteed.

4) Commodities: is a super-cycle starting + oil near-term view

Commodities framing

Commodities were split into categories such as:

  • Base metals
  • Energy
  • Softs/agri

  • Precious/base/energy/agri categories referenced

  • Silver was described as more speculative.

  • Gold was described as having little fundamental link to the AI/electrification demand theme.

Base metals supported by AI + electrification

Favored metals mentioned:

  • Copper (described as at an all-time high)
  • Aluminum (near record highs; “almost there”)
  • Nickel (“a bit”)

Why:

  • Supply response post-2020 described as weak (limited capacity expansion).
  • Demand appears to be picking up via electrification/renewables/AI data centers.

Geopolitics as an additional driver (fragmentation)

  • A key mechanism: markets are fragmenting.
  • Refineries can’t process oil from any corner of the world.
  • Result: suboptimal refining routes → structural price increases beyond simple global surplus/deficit logic.

Oil (Iran/US war, Strait of Hormuz) — near-term stance

  • “Eye of the storm,” but not expecting oil to return to $60 (explicit comparison).
  • Near-term:
    • Inventories refill likely takes time → oil remains “well bid.”
  • Longer-term:
    • Potentially bearish later, but near-term supportive due to energy security shifts.
  • Defense/energy security logic:
    • Russia–Ukraine increased realization that defense reliance is risky → pushes energy self-sufficiency:
      • renewables, biofuels, possibly coal resurgence.

Investing implication (stocks/indices)

  • Suggested beneficiaries:
    • Defense and electrification/renewables
  • Defense risk screening emphasis:
    • Many defense firms are SOEs (state-owned enterprises) with:
      • single government customer risk
      • delayed/non-commercial payment dynamics
  • Diligence rule for defense investing:
    • Prioritize cash flow and balance sheet first (order book/earnings growth is secondary).
  • Preferred defense company traits (as framed):
    • diversified customers
    • reasonable margins
    • consistent conversion of margins into cash flow

5) India macro & valuation: why foreign sentiment weakened + what could fix it

Listed India “bear case” (as summarized)

  • No listed AI plays (missed “industrial revolution” framing).
  • AI threatens IT services employment, with second-order effects (e.g., real estate/consumption).
  • The “next engine” of manufacturing hasn’t fully materialized:
    • Manufacturing cited at ~17% of GDP.

Valuation/market correction explanation

  • “PE-wise we are not cheap… still 2020 on headline.”
  • PEG framework discussed:
    • If PEG ~ 2x with 9–10% EPS growth, implied P/E around 18–19.
  • Outcome:
    • Forward earnings expectations and price make India look less attractive vs peers.
  • During 2020–2024, India had a growth differential vs EM of ~6–8%, supporting foreign willingness to pay.
  • Now, foreign investors are effectively less willing, contributing to disenchantment.

Timeline / trajectory

  • Before the war: expectation that India EPS growth could return to ~13–15%.
  • With the war: described as delayed rather than denied.
  • Forward P/E cited:
    • From ~22–23 down to ~18–19.

Earnings growth vs employment impact

  • The “employment/income engine” isn’t strictly IT anymore:
    • Financial services cited as a larger employer in listed space.
  • Example of consumption/operation strain:
    • delivery/gig (Swiggy/Zomato) struggling to find delivery staff

  • IT services hiring:

    • peaked around 2021, then became anemic
    • India avoided a severe contraction (“no huge NPSPA cycles” referenced)

  • Reskilling:

    • Some cuts are not expected structurally; retooling/reskilling expected.

Sector tilts (portfolio behavior)

  • Avoidance noted:
    • “Staples pay 50p for 5–7% growth” (low-growth staple framing)
  • Portfolio tilt:
    • More toward consumer discretionary than staples (not heavily into staples).

Oil/rupee risk considerations

  • Risk concern: inward-looking sectors (banks/financial services; gig/consumption categories) could be affected if oil stays high.
  • Counterpoint:
    • Consumption impact depends on duration/magnitude.
    • Oil scenario recalled from 2013 context:
      • current account deficit ~5% of GDP
      • $90B in a smaller economy
  • Rupee/currency:
    • India appears reasonably cheap on real effective exchange rate (macro).
    • Near-term moves driven by exporter/importer expectations (a behavioral reversal dynamic).
    • Tools exist (govt/RBI) and it’s not expected to become a “currency spiral.”
  • Capital account reality emphasized:
    • FDI net weak: “single digit billion” last year.
    • FIIs selling recently.
    • If current account deficit is ~2%, balancing framed as needing roughly ~$80B capital inflow.

Tax policy discussion (FIIs / gains)

  • Would removing capital gains tax for FIIs fix capital inflows?
    • Uncertain—“never know the counterfactual.”
    • It might help, but the larger driver is future expected equity returns.
  • Counterargument raised:
    • Domestic SIP/investors have been primary support over the last 5 years.
    • Question posed: why do foreign investors get tax breaks but domestic savings do not?

6) Commodities / investment holdings explicitly mentioned

  • Commodities via stocks only:

    • Uranium (uranium stocks in emerging markets; specific tickers not provided)
    • National oil companies in Brazil and China (no ticker names provided)

  • Metals held:

    • Copper
    • Aluminum
    • Uranium (via stocks)

7) Tickers / instruments / assets explicitly mentioned

Countries / indices / instruments

  • MSCI ACWI (referenced as “MSCI Aqui/Acquire”)
  • 2x / 3x leveraged ETFs (generic; no specific tickers)
  • Emerging markets portfolio / India fund / emerging markets fund (no tickers named)

Equities / companies

  • Samsung
  • SK hynix
  • TSMC
  • Nvidia (mentioned in context/comparison)
  • BYD
  • AI/hyperscalers named:
    • Anthropic
    • OpenAI
    • Google
  • Food delivery named:
    • Swiggy
    • Zomato
  • TCS mentioned (in context; not as a specific ticker recommendation)

Commodities / asset classes

  • Copper
  • Aluminum
  • Nickel
  • Silver (speculative)
  • Gold (little fundamental linkage claimed)
  • Oil
  • Uranium

Rates / bonds

  • US 10-year yield cited around 4.5%

8) Key explicit performance / valuation numbers mentioned

  • India market long-run: ~11–12% CAGR (historical claim)
  • Patience required: holding through volatility for 15–25 years
  • Behavioral difficulty: enduring equity drawdowns of ~50% (example: 2007–09)
  • Hyperscaler capex:
    • ~$750–800B this year
    • ~$1T within 2–3 years
  • Memory cycle evidence:
    • Smartphone and notebook shipments down double digits
  • South Korea bubble-like moves:
    • SK hynix doubled in a month
    • ~8x in ~a year and a bit
    • Cost up ~3x
  • US rates:
    • US 10-year yield around ~4.5%
  • India valuation:
    • PEG example: PEG ~2x with 9–10% EPS growth → ~18–19 P
    • P/E moved from ~22–23 to ~18–19
  • India macro claim:
    • ~$4T economy growing ~10% nominal, creating roughly ~$400B GDP per year

Disclosures / disclaimers

  • Breakout Capital says they:
    • advise/manage with no product sales in India (stated: “we have nothing to sell in India”).
    • “personal view not a recommendation” is stated when discussing defense company selection.
  • A general “not a recommendation” style disclaimer is present (though not quoted in full).

Presenters / sources (named)

  • Swanandd Kelkar (interviewee; managing partner at Breakout Capital; described as managing around $1.7B in advised assets, up from $1B)
  • Host/interviewer: Neil
  • Additional names referenced:
    • Kevin Walsh (regarding Fed governor confirmation)
    • Arun Jaitley (referenced for capital gains tax timeline)

Original video