Dot-Com Bubble 2.0 Incoming? Why Tech Fund 'Aggressively' Sold Shares | Stephen Yiu
Speakers
Stephen Yiu
— CIO and Co-Founder, Blue Whale Growth Fund
Summary
Stephen Yiu of Blue Whale Growth Fund discusses whether tech is in a dot-com-style bubble, arguing valuations are elevated but business models are more robust and AI is more real. He prefers AI infrastructure suppliers Nvidia, Broadcom and Vertiv over AI application/services companies and is underweight the other Magnificent 6 due rising AI capex and falling ROIC. He also likes idiosyncratic names Flutter and Philip Morris, biologics equipment makers Danaher and Sartorius, and expects Chinese EV makers to dominate while avoiding Tesla. He sees a dot-com-style tech crash as very low probability, though quantum computing is a long-term disruptive theme to monitor.
- Stephen Yiu says tech is not in a 2000-style bubble despite higher concentration and valuations.
- He favors AI infrastructure, especially Nvidia, Broadcom and Vertiv, over unclear AI application winners.
- He sold Microsoft and is underweight the Magnificent 6 ex-Nvidia due ROIC pressure from AI capex.
- He prefers Broadcom over Nvidia for near-term AI upside.
- He likes Flutter and Philip Morris as idiosyncratic holdings and is cautious on consumer staples ex-Philip Morris.
- He has healthcare exposure via Danaher and Sartorius for biologics manufacturing.
- He expects Chinese EV makers to dominate and views Tesla valuation as expensive.
- He views a decade-long Nasdaq tech crash as very unlikely and quantum computing as a 5-10 year watch item.
Ideas
Tech still best area for returns.
Despite bubble comparisons and high tech concentration, Stephen expects technology companies to remain the area where most money is made over the next couple of years. Digital adoption and AI are real, tech business models are robust and highly profitable, and valuations are elevated but less extreme than in 2000. He says investors must differentiate within tech but still favors the sector over non-tech.
Microsoft AI capex threatens ROIC.
Stephen aggressively sold Microsoft and pushed it out of the fund's top 10 because he expects Microsoft's return on invested capital to decline. Microsoft must spend heavily on AI infrastructure and GPUs from Nvidia/Broadcom for Copilot, unlike its prior high-margin software/cloud transition where it did not pay external suppliers. Core Office/cloud remains robust and he might buy after a 20-30% decline, but current incremental AI spend pressures returns.
Prefer AI infrastructure over applications.
He draws a major line between AI infrastructure and AI applications/services. He prefers infrastructure, especially Nvidia, Broadcom and Vertiv, because every company must spend money with these suppliers before it can compete in AI; they receive the capex rather than spending it. AI application startups face a bubble-like VC market with unclear ultimate winners.
Prefer AI infrastructure over applications.
He draws a major line between AI infrastructure and AI applications/services. He prefers infrastructure, especially Nvidia, Broadcom and Vertiv, because every company must spend money with these suppliers before it can compete in AI; they receive the capex rather than spending it. AI application startups face a bubble-like VC market with unclear ultimate winners.
Google AI costs pressure ROIC.
Alphabet's AI search/Gemini is more expensive to serve than traditional Google Search because it requires training and inference, while search is a pure ranking algorithm. Google must monetize Gemini better, and Stephen expects Alphabet's ROIC to decline due to AI infrastructure spending. Core search is robust but incremental AI capex weighs returns.
Underweight Big Tech excluding Nvidia.
Stephen is significantly underweight the Magnificent 7 except Nvidia and is negative or cautious on the other six. They are the customers of AI infrastructure, so their capital intensity is rising and their free cash flow is shifting to Nvidia/Broadcom, lowering ROIC. He does not predict a crash but prefers not to own them.
Nvidia outperforms, but less explosively now.
Stephen still likes and holds Nvidia. It is a capital-light AI infrastructure winner that outsources manufacturing to TSMC, earns high margins, and should benefit as enterprises such as banks and Johnson & Johnson buy GPUs because they will not build in-house. But upside is more moderate than before: he sees Nvidia outperforming an S&P 10% annual return with maybe 15%, not another 100-200%, because it is already a $3T+ company and depends on big tech/enterprise spending and TSMC execution.
Broadcom preferred AI infrastructure upside.
Broadcom is a recent AI infrastructure addition and his preferred pick over Nvidia. Big tech customers want to diversify away from Nvidia and are working with Broadcom on in-house GPUs, including Meta, Microsoft and rumored OpenAI/Apple work, so Broadcom is smaller and could have more short-term upside, possibly 50% in 12-18 months.
Vertiv benefits from data-center cooling.
He likes Vertiv, a lesser-known AI infrastructure company supplying liquid cooling systems for data centers. GPU compute is more energy intensive, so data centers need liquid cooling to cut energy costs and improve efficiency.
Flutter is idiosyncratic US sports gaming play.
He likes Flutter as an idiosyncratic US sports gaming play. Sports gaming legalization happens at the state level, so it is not dependent on Trump or broad macro policy; he considers it new and differentiated.
Philip Morris smokeless transition drives growth.
Philip Morris is his only consumer staples holding and he likes it for its transition to smokeless products, IQOS expansion/return in the US, and Swedish Match acquisition in oral nicotine/chewing tobacco. It has performed better than Microsoft and he calls it one of the best consumer staples companies.
Quantum computing is long-term disruptive watch.
Quantum computing is still early innings. Google's Willow breakthrough is helpful but not yet a real application. If it does break through, it could disrupt Nvidia GPUs and transformer/LLM computing and become the next level after AI, but it is likely 5-10 years away. Nvidia/Broadcom investors should monitor it closely.
Tesla valuation not justified by fundamentals.
Stephen has never owned Tesla and cannot justify the valuation. He estimates a large premium is required for Elon Musk to make autonomous driving a game-changer; current fundamentals, deliveries and cash flow do not support the market cap, with roughly $5B FCF against over $1T market value. He thinks Tesla could be one of the bubbles to bet against, but avoids shorting because Musk could make it work.
Chinese EV makers to dominate; Europe lags.
If he had to bet, Chinese EV makers have the best chance to dominate global autos. EV is a new technology, Chinese manufacturers started earlier, have scale and strong charging infrastructure in tier 1/2 cities, and can remain competitive even with US/European tariffs because of low-cost technology. European manufacturers have not cracked EV and face harder adoption/charging economics; outside Tesla, Western competition is limited.
Chinese EV makers to dominate; Europe lags.
If he had to bet, Chinese EV makers have the best chance to dominate global autos. EV is a new technology, Chinese manufacturers started earlier, have scale and strong charging infrastructure in tier 1/2 cities, and can remain competitive even with US/European tariffs because of low-cost technology. European manufacturers have not cracked EV and face harder adoption/charging economics; outside Tesla, Western competition is limited.
Staples ex Philip Morris face deflation risk.
He is very cautious on consumer staples outside Philip Morris and skeptical of personal/household goods. These companies raised prices during the pandemic/inflation, but consumers now face weak discretionary spending, high rates and inflation, and are trading down to cheaper supermarket private-label products, creating deflationary pressure.
Biologics equipment benefits from AI drug discovery.
Within healthcare he is selective and only invests in biologics. He owns Danaher and Sartorius, which supply manufacturing equipment for biologics used by biotech companies. He believes AI will accelerate drug discovery, and over about 10 years personalized drugs will grow versus chemical/synthetic drugs; the sector has lagged but should do well.
Up Next
This The David Lin Report video, published January 08, 2025, features Stephen Yiu discussing XLK, MSFT, AIQ, AI-SECTOR, GOOG, Magnificent 6, NVDA, AVGO, VRT, FLUTTER, PM, QUBT, TSLA, Chinese EV manufacturers, European auto manufacturers, XLP, DHR, SRT3.DE. 17 trade ideas extracted by AI with direction and confidence scoring.
Speakers: Stephen Yiu · Tickers: XLK, MSFT, AIQ, AI-SECTOR, GOOG, Magnificent 6, NVDA, AVGO, VRT, FLUTTER, PM, QUBT, TSLA, Chinese EV manufacturers, European auto manufacturers, XLP, DHR, SRT3.DE