How to Build Your AI Stock Portfolio for MAXIMUM Investment Returns
Overall Thesis
Bhakdi outlines his AI-era portfolio construction framework with six themes: embodied AI (Tesla), hyperscalers (Microsoft/Google/Meta/Amazon/Apple), organizational AI (Palantir/Snowflake), AI infrastructure (Nvidia/AMD/Broadcom/Micron/Intel), age of capital (Robinhood/SoFi/Coinbase), and crypto (BTC/ETH/MSTR/Bitmine). He runs concentrated diversification with ~5-9 positions and targets 36% compound annual returns. Current positioning is heavy TSLA (50%), heavy crypto, with new Meta addition.
Narratives
Nvidia is a very low-risk investment with high growth as the world's largest corporation and AI chip leader with CUDA moat, connectivity backbones, and massive demand. Every chip producible today through 2028 is already sold out. But because the business is so transparent with known revenue streams, the alpha opportunity is smaller than Tesla.
Key Arguments
- CUDA software ecosystem is Nvidia's most powerful moat that AMD struggles to catch up to
- Every single chip producible even in 2028 is already sold out
- Hyperscalers pull demand from NVIDIA into their super AI cloud
- Growth business with low bear risk
Tesla is Bhakdi's top position at 50% of portfolio because it is vastly undervalued compared to other AI stocks given the robotaxi and Optimus opportunity. The disconnect between the future as it will happen and what the market believes is far greater for Tesla than for Nvidia, creating alpha. Tesla is the only real embodied AI play with uncontested leadership in autonomous transport and humanoid robots.
Key Arguments
- Tesla is the only real embodied AI play in the western world
- Humanoid robots addressable market: billion robots x $20,000 = 20 trillion
- Tesla will have tens and hundreds of millions of robotaxis very soon
- Market misunderstands the future trajectory more for Tesla than Nvidia, giving more alpha
- Tesla is a perceived higher risk with way higher growth versus Nvidia
Meta is Bhakdi's favorite underrated hyperscaler pick because the market misunderstands Meta's AI capex and has not recognized that Yann LeCun leaving frees their amazing AI team. Low P/E plus Zuckerberg's aggressive entrepreneurship should drive cool AI innovation in the next 12 months while the stock is underappreciated.
Key Arguments
- Meta has over 3 billion users
- Zuckerberg is scaling AI very aggressively
- Yann LeCun leaving unchains Meta's AI team to actually innovate
- Lowest P/E among hyperscalers - best price per company quality
- Next 12 months will see cool AI stuff from Meta
Coinbase is in the age-of-capital theme but is a little bit behind Robinhood, essentially copying Robinhood's strategies. It will invade Robinhood's territory more and more but is less interesting given Robinhood's aggression.
Key Arguments
- Coming from crypto side, invading Robinhood area
- Will converge with Robinhood and compete
- Part of the trio aiming to become largest financial companies
Bitcoin is the most reliable hedge against inflation and money printing driven by AI-induced mass unemployment and government fiscal crisis. Limited supply makes it even better than gold. Forms one of Bhakdi's three portfolio legs, independent from equities.
Key Arguments
- Best store of value due to limited supply, better than gold
- Money printing guaranteed to go through the roof in age of AGI
- Low correlation with equities - independent portfolio leg
- Bitcoin at ~2T vs gold at ~10T leaves massive room
AMD is a smaller AI infrastructure play with chips that are sometimes cheaper and better performance per dollar, but struggles catching up with CUDA. Only interesting because so much smaller than Nvidia at 400B vs 4T, giving it easier growth runway.
Key Arguments
- Some chips cheaper with better performance per dollar
- CEO Lisa Su is related by blood to Jensen Huang - great Taiwan heritage
- Smaller at 400B vs NVDA 4T - can grow more easily
Apple is the weakest of the hyperscalers and is falling behind. Bhakdi thinks Apple will probably not look good in the next 10 years - it's lame and he's not looking at it right now.
Key Arguments
- Falling behind other hyperscalers
- Weakest of the five hyperscalers
- Lame - Bhakdi is not looking at it
Micron Technology is the American-made memory stacked onto GPUs, a super amazing business in the AI infrastructure theme. One of the great things.
Key Arguments
- American-made memory for GPUs
- Super amazing business
Google is one of the great leaders in frontier AI and very strong in research. Killing it now with Gemini 3.
Key Arguments
- Great leader in frontier AI
- Very strong in research
- Killing it with Gemini 3
Robinhood is Bhakdi's favorite age-of-capital pick because CEO Vlad Tenev leads much more aggressively and innovatively than SoFi or Coinbase. Robinhood has a comprehensive vision, is tokenizing everything going into retail, and is executing disruptive strategies despite its $70 billion market cap.
Key Arguments
- CEO Vlad Tenev has critical edge of aggressiveness as entrepreneur
- Leading Robinhood like a startup even at $70B
- Tokenizing everything and disrupting personal finance
- Aiming to become largest financial company on earth
SoFi is in the age-of-capital theme with a comprehensive financial base. A great company that has gone up like crazy but is not done, aiming to become one of the largest financial companies.
Key Arguments
- Comprehensive financial basis
- Gone up like crazy but not done
- Part of age-of-capital thesis
Ethereum is the rails on which future financial industry is being built - from BlackRock to JP Morgan to Robinhood and Coinbase. Number one choice for tokenizing all financial assets on blockchain.
Key Arguments
- Rails for financial industry tokenization
- Dragged into sunlight by BlackRock, JP Morgan, Robinhood and Coinbase
- Number one choice for tokenizing financial assets
Palantir best represents the organizational AGI transformation theme - turning companies from people in cubicles into cybernetic intelligent organisms. Has the tightest grip on this vertical. Every company on earth needs to go all-in on this transformation or be knocked out of business within 3 years.
Key Arguments
- Tightest grip on organizational AGI transformation vertical
- Margin booster of insane proportions - can reduce workforce to near zero
- Every company on earth needs this transformation or dies
- Proctor and Gamble example: fire 90% of employees, real-time intelligence
Broadcom delivers custom-built chips for hyperscalers that circumvent NVIDIA's CUDA moat because hyperscalers have their own software teams. Over a trillion market cap with growing share.
Key Arguments
- Custom chips for hyperscalers that don't need CUDA
- Cuts middleman costs
- Over a trillion market cap
Intel is the 'Apple of AI infrastructure' - a little low tech, a little behind, many problems, but still in the game. Bhakdi isn't looking at it now but keeps it in the universe.
Key Arguments
- Still in the game despite problems
- Don't discount them entirely
Microsoft under Satya Nadella is one of the great new AI companies leading the frontier, with massive Windows distribution and a key stake in OpenAI. Strategically placing bets.
Key Arguments
- Satya Nadella strategically placing bets
- Owns good chunk of OpenAI
- Massive Windows distribution
Amazon is the secret execution master applying AI both internally for profit generation in core retail and externally as a hyperscaler providing cloud systems. Can become absolutely dominant in global e-commerce with AI streamlining.
Key Arguments
- Secret execution master
- AI applied internally for retail profit generation
- Hyperscaler providing cloud systems
- Can dominate global e-commerce with AI