UBS: Tech Stock Volatility Hits Highest Level Since Dot-Com Bubble
0xBroomberg
UBS's quantitative unit HOLT reports that global tech volatility has climbed to its highest level since the 2000 dot-com bust, as investors question whether AI-driven hyperscalers and chipmakers can sustain their cash-flow returns.
Hyperscalers are spending big — where's the payback?
The top five hyperscalers — Microsoft, Meta, Alphabet, Amazon, and Oracle — face a combined $227 billion funding gap next year, meaning their cash on hand cannot cover both operations and existing financing commitments.
This means → even though margins are holding for now, massive AI infrastructure spending is dragging down asset efficiency. CFROI (cash-flow return on investment — how much cash each dollar of capital earns) is projected to decline through 2028.
UBS reviewed roughly 650 major capex surges since 1998. In 60% of those cases, returns fell permanently — and companies with the highest starting returns suffered most.
Can semiconductor valuations really hold for five years?
Semiconductor yields have nearly tripled to around 30% — a level fewer than 1% of companies have reached since 1990. In plain terms = the current stock price assumes these firms stay at peak profitability for another five years.
This reflects a gap between valuations and competitive reality — history shows very few companies sustain returns this high before rivals close in.
UBS names Chinese AI firms DeepSeek and Moonshot AI (月之暗面) as examples: these players prioritize market share over profit, suggesting the semiconductor "moat" may not be as deep as priced in.
Software stocks lost 40% of their premium — will it come back?
Over roughly the past eighteen months, software, enterprise data, and services stocks saw aggregate price-to-book ratios drop about 40% — the market is pricing in AI disruption of their business models.
UBS historical data shows 80% of stocks that suffered a markdown of similar magnitude never recovered their prior valuations within a decade. This means → this is not a simple "oversold bounce" setup — the market is repricing the ceiling for an entire sector.
Does hiding in value and low-vol stocks actually work?
UBS reviewed six major tech-driven selloffs since 2004: value stocks outperformed in all six, and low-volatility stocks outperformed in four — making these the two strongest defensive style factors historically.
But UBS flags two limits: since 2023, the traditional link between value stocks and the economic cycle has weakened noticeably; low-vol stocks, unless backed by strong fundamentals, tend to underperform outside selloff windows.
Put simply = the playbooks that worked before may not transfer cleanly — the environment has shifted.
What to watch next?
UBS identifies the core test: whether hyperscalers can keep expanding capex while defending cash-flow returns.
This means → in the coming quarters, the number to watch in earnings is not revenue growth — it is how much real cash each additional dollar of AI spending actually generates.
Content is for reference only, not financial advice.