The AI boom shares some similarities with the dot-com era. Companies are investing heavily in infrastructure ahead of fully realised demand, betting that future adoption will justify today’s spending.
The difference is that today’s AI cycle is supported by real revenue growth, strong demand, and investment from some of the world’s largest technology companies. The market is no longer debating whether AI matters. The focus has shifted to whether the scale of investment can generate enough returns to justify current expectations.
That tension is becoming clearer across the semiconductor sector. Investors are still participating in the AI rally, but they are also hedging against the possibility that expectations move faster than earnings.
From growth expectations to earnings proof
Semiconductor valuations have traditionally been driven by familiar factors: sales growth, earnings, inventory cycles, and supply-demand conditions.
Those factors still matter, but AI has introduced a longer-term growth assumption into the market.
Investors are now pricing in several future outcomes:
- Continued AI infrastructure spending from hyperscalers
- Sustained demand for advanced chips
- Wider adoption of AI applications
- Stronger profitability from today’s investment cycle
Many of these outcomes are still developing. The market is paying for future growth, but that growth still needs to translate into measurable returns.
The AI trade through three market signals
Three semiconductor ETFs provide a useful view of how investors are positioning around that uncertainty. Each captures a different part of the AI trade: broad sector participation, speculative conviction, and concentration in AI leaders.
| ETF | What it measures | What it reveals about the AI trade |
| SOXX | Broad semiconductor sector exposure, including AI beneficiaries such as Nvidia, Broadcom, AMD, Micron, and equipment makers | Whether confidence is spreading across the wider semiconductor industry or remaining concentrated in a few names |
| SOXL | 3x leveraged exposure to semiconductor movements | The intensity of investor positioning and willingness to take amplified exposure to the sector |
| SMH | Concentrated exposure to major semiconductor leaders such as Nvidia, TSMC, Broadcom, and ASML | How much the AI rally depends on companies at the centre of infrastructure spending |
The AI trade is not a single bet. SOXX reflects breadth, SOXL reflects conviction, and SMH reflects concentration.
SOXX shows whether confidence is broadening.
A stronger performance suggests AI optimism is spreading beyond a handful of major winners and reaching the wider chip ecosystem. If AI leaders continue rising while SOXX weakens, it would suggest the rally is becoming increasingly narrow.
SOXL shows how aggressively investors are positioned.
SOXL reflects the intensity of the trade rather than its underlying fundamentals.
Because leverage amplifies market moves, strong SOXL performance signals that investors are willing to take larger directional bets. The same mechanism can also accelerate losses when sentiment reverses.
SMH shows where the AI trade is concentrated.
SMH highlights the companies most directly tied to AI infrastructure spending.
Its strength reflects continued confidence in AI leaders but also shows how dependent the broader trade has become on a small group of dominant companies.
Traders can monitor semiconductor ETF price movements through CFDs with the VT Markets app, with access to market data and real-time price tracking.
Memory tests the sustainability of AI demand
The AI infrastructure cycle is not only about processors. Memory demand provides another view of whether spending is spreading through the supply chain.
SK Hynix has increased investment in high-bandwidth memory (HBM), a critical component for AI systems. Strong AI server demand has also supported DRAM pricing. TrendForce estimated conventional DRAM contract prices could rise 58–63% quarter-over-quarter in the second quarter of 2026, while NAND contract prices were projected to increase 70–75% as enterprise storage demand benefited from AI infrastructure expansion.
Memory is an important test because it has historically been one of the most cyclical parts of the semiconductor industry. Strong demand often leads to capacity expansion, which can eventually create oversupply and pressure prices.
The current cycle depends on whether AI creates a lasting increase in memory demand or whether the industry eventually returns to its traditional boom-and-bust pattern.
Investors are hedging AI exposure
The AI trade is showing an unusual combination: strong price performance alongside elevated uncertainty.
The Cboe Semiconductor ETF Volatility Index, which tracks expected volatility for SMH, reached 64.57 at its recent peak and closed around 61.21. The broader VIX was around 17.05 during the same period. Semiconductor investors are pricing in significantly larger potential moves than the broader equity market.
Normally, volatility rises when prices fall. Recently, volatility has also remained elevated during periods of strength. This suggests investors are not abandoning the AI trade, but they are managing the possibility of sharper moves in either direction.
Options activity and concentrated positioning may amplify short-term price swings, but the underlying drivers remain unchanged:
- AI infrastructure spending
- Semiconductor demand
- Earnings growth
- Interest rates
- Valuations
Market positioning can accelerate a move, but it does not replace fundamentals.
The proof points to decide AI’s next phase
The next stage of the AI cycle will depend less on how much companies spend and more on whether that spending creates measurable returns.
Hyperscaler AI infrastructure spending
Cloud companies remain the foundation of AI infrastructure demand.
Continued investment growth across multiple companies would support the current cycle. A slowdown in infrastructure plans would suggest spending is moving ahead of expected returns.
Alphabet raised its 2026 capital expenditure outlook to approximately $195–205 billion, up from its previous range of $180–190 billion.
Recent estimates have also suggested that combined AI infrastructure spending ambitions from major cloud companies could exceed $700 billion in 2026.
The market will be watching whether this investment continues translating into revenue growth.
AI chip demand and semiconductor supply trends
AI chip demand remains strong, but the quality of that growth matters. A healthy cycle would see demand expanding across a wider customer base, with sustained orders and firm pricing.
Signs of weaker demand concentration, slowing orders, or rising inventory would suggest supply is catching up.
Enterprise AI adoption and AI monetisation
Infrastructure spending ultimately depends on whether businesses turn AI tools into measurable value. Wider commercial adoption would strengthen the case for continued investment.
A longer period of experimentation without large-scale deployment would raise concerns that monetisation is taking longer than expected.
Semiconductor profit margins and AI investment returns
The next phase of the AI cycle will be judged by earnings, not investment alone. Companies maintaining or expanding margins despite higher spending would show that AI demand is translating into financial returns.
Margin pressure would suggest that infrastructure costs are rising faster than the revenue they generate.
Outlook on AI
Participating in the rally while hedging against downside risk reflects a more cautious approach to a market where future returns remain uncertain. Dot-com’s fatal flaw was a one-way conviction. Almost nobody was hedged against their own story, so when the story broke, there was nothing underneath to absorb it.
This market is doing the opposite, and that might be the real difference this time. Holding positions on both sides of a move is a familiar approach in derivatives markets, where participants use different strategies to express views or manage exposure.
The market has not reached a verdict on whether AI represents a bubble or a new productivity cycle. What is becoming clearer is that the next phase will be decided by one factor: whether today’s spending can become tomorrow’s earnings.
Interested in tracking price movements in the semiconductor ETFs mentioned? Take positions on both rising and falling markets on them, now available as CFDs on VT Markets.
Tap for TL;DR
Is the AI rally still supported by fundamentals?
Yes. AI demand, semiconductor investment, and hyperscaler spending remain strong, but investors are watching whether future earnings can justify current expectations.
Why are investors hedging AI exposure despite strong market performance?
Hedging reflects caution around valuations, concentration risk, and uncertainty over how quickly AI spending will translate into returns.
What do SOXX, SOXL, and SMH reveal about the AI trade?
SOXX shows whether AI optimism is spreading across semiconductors, SOXL reflects market conviction, and SMH highlights dependence on AI leaders.
Is AI infrastructure spending creating a sustainable growth cycle?
The next phase depends on whether chip demand, memory growth, and enterprise AI adoption can turn investment into measurable profits.
What will determine the next move for AI semiconductor stocks?
Hyperscaler spending, AI chip demand, adoption rates, and semiconductor margins will show whether the AI cycle can deliver long-term returns.
Start trading now — click here to create your real VT Markets account.
