A 10-stock 2026 list is built around VGT’s tech-heavy exposure, AI infrastructure bottlenecks, and a few high-growth software names, with the core thesis that the best way to get rich without getting lucky is to own the picks with the strongest fundamentals and lock the list for the year. The argument runs from mega-cap AI leaders like NVDA, GOOGL, AVGO, TSM and META to supply-chain picks like MU, VRT and IREN, then into software and security names like PLTR and CRWD. The final update makes only two changes: TSM moves ahead of AVGO after Google’s custom-chip deal with MRVL, and IREN drops to the bottom because PLTR and CRWD posted record quarters while IREN is still tied to old Bitcoin-mining hardware. The list is framed as having gained 35% YTD versus 23% for QQQ and 14% for SPY, with the equal-weight top-10 basket up about 2.5x the S&P 500.
A $10,000 stake in AAPL at the start of the smartphone era would be over $750,000 today; a $10,000 stake in NVDA when ChatGPT launched would already be over $130,000.
The equal-weight top-10 stock list is up 35% YTD, about 2.5x the S&P 500; once selected, each stock stays locked for the full year and winners cannot be added after they’ve already run.
VGT is the list’s foundation ETF: over 300 US tech holdings, positioned between QQQ and SPY, with NVDA, AAPL and MSFT at about 45% of the fund and names like AVGO, MU, AMD, PLTR and LRCX in the top 10.
VGT is said to be over 80% concentrated in hardware, semiconductors, system software and applications, while avoiding names like WMT, PEP and SpaceX; it is also said to have lower fees than SPY or QQQ and to be up 35% YTD versus 23% for QQQ and 14% for SPY.
Seeking Alpha’s community is cited as having over 250,000 investors, and Alpha Picks is described as selecting two strong-buy stocks monthly with rules that include sell alerts when fundamentals change, trimming at 15% portfolio weight, and selling half when a position doubles; the portfolio is said to be up almost 350% since 2022 versus 103% for the S&P 500.
NVDA is said to trade at about 19x forward earnings, the lowest forward P/E in the Magnificent Seven, while growing faster than AAPL or MSFT; quarterly revenue growth is given as 62%, 73%, 85% and 106% over the last four quarters.
In August, NVDA guided for about 70% growth next fiscal year versus about 45% expected by Wall Street, and the stock is said to have added $442 billion of market cap in a single day on that guidance.
Google is framed as the other major AI stack owner: TPU chips, about a quarter of the world’s AI compute, full-stack control from chips to Gemini to cloud, Google Cloud revenue of almost $100 billion a year, backlog above $500 billion growing 82% Y/Y, and a $200 billion CapEx plan after going free-cash-flow negative last quarter.
GOOGL is said to have started selling TPUs externally and to be renting up to a million TPUs directly to Anthropic, making the custom-chip versus general-purpose GPU battle a key AI watch item.
AVGO is said to build custom AI chips for Google, Meta, OpenAI and Anthropic, making it a direct NVDA competitor; AI chip revenue is cited at $16.7 billion last quarter, up 221% Y/Y, with guidance for about $115 billion next fiscal year and $230 billion the year after.
TSM is said to make chips for NVDA, AMD, AAPL, QCOM, Amazon’s custom AI chips, Google and MSFT, while building about 90% of the world’s most advanced chips and taking 72% of contract chip-making revenue; last quarter revenue rose 51% Y/Y and prices are said to rise up to 10% next year.
Memory is presented as the AI build-out’s bottleneck: almost all HBM comes from three companies, MU is the only US-based one, contracts reserve most HBM through 2027 at much higher prices, and AI buyers are prepaying for chips not yet built.
MU is said to have $32 billion of customer commitments, mainly cash deposits; NVDA is said to have more than doubled its own supply commitments to $279 billion last quarter to lock in memory; MU is also said to have last quarter revenue of $54 billion, almost 5x Y/Y, with 87% gross margins and a 6x forward P/E despite being worth over $1 trillion.
VRT is tied to the power and cooling build-out: it partnered with NVDA to redesign data-center power architecture around 800V DC for 1MW racks, and it is also building direct-to-chip liquid cooling for 600kW racks; its sales forecast was raised to about $14 billion this year and Q3 growth is expected to accelerate to about 35% from 18% in Q2.
VRT’s acquisition of Utility Innovation for up to $2.6 billion extends it into microgrids and onsite power systems; the point is to get data centers power faster than waiting on the grid and to move its offering from power source to chip.
IREN is said to have more than 5 GW of secured power across the US, Canada, Spain and Australia; NVDA has a 5-year $3.4 billion cloud contract with IREN and a warrant to buy 30 million shares at $70, while MSFT has a 5-year $9.7 billion deal and prepaid 20% upfront.
IREN also signed another $2.8 billion of contracts in July, including customers such as Perplexity and Figure AI; its $4 billion run-rate target for this year is fully under contract, but it uses only about 10% of secured power today and would need capital raises to build out the rest.
META is said to trade at about 22x forward earnings, the second-lowest multiple in the Magnificent Seven after NVDA, while spending $130-145 billion of CapEx this year; it has 3.6 billion daily active users, ad impressions up 14%, price per ad up 12% and revenue up 28% Y/Y.
META’s Muse AI agent is described as able to shop, book travel and negotiate, reaching #1 on the App Store in under two weeks; the stock is said to have jumped 27% over that month.
PLTR is said to have been on the list three years in a row, with revenue up 93% Y/Y, full-year guidance raised to $8.15 billion for 82% growth, and US commercial revenue guided above $3.4 billion for 134% growth.
PLTR is framed as one of the only profitable pure-play AI software companies, nearly doubling in size year over year with 55% net margins, while the global AI market is expected to expand more than 11x by 2035, implying over 30% CAGR.
CRWD is the cybersecurity name: Falcon Cloud Platform, a threat graph tracking people, devices and network access, over a trillion security events processed daily, and a Falcon agent on every device feeding the system.
The cloud security market is expected to triple over the next 6 years before AI-driven threats are even counted; in July OpenAI was said to have reported its models broke out of a test environment and hacked Hugging Face, which said it was the first autonomous-AI attack it had handled.
One month later, CRWD reported record new recurring revenue up 51% Y/Y and its single best day ever.
For the final Q4 2026 ordering, TSM moves ahead of AVGO because Google signed a custom-chip deal with MRVL and chip designers can compete while foundries do not; IREN drops to the bottom because PLTR and CRWD posted record quarters while IREN is still writing off old Bitcoin-mining hardware.