Wall Street analysts and the US Treasury issued conflicting assessments of the AI rally's sustainability as major semiconductor companies report record earnings. A recent non-public US Treasury draft report warned that failure to monetize AI investments could trigger dot-com-era financial shocks, while JPMorgan argued that ongoing semiconductor supply shortages present buying opportunities. The divergence comes as bigtech infrastructure spending reached $848B in 2025 and Samsung Electronics and Micron reported historic results. Treasury analysts noted AI's deeper integration into the US economy compared to the dot-com period, citing risks spanning datacenter construction financing, private credit markets, cloud providers, chipmakers, and utilities if productivity gains fail to materialize. Market attention now shifts to upcoming earnings calls from Penguin Solutions on July 7 (local time), TSMC on the 10th and 16th, and ASML on the 15th, where CEO guidance on contract pipelines and future outlook will test investment cycle durability.
US Treasury Warns AI Monetization Failure Could Echo Dot-Com Crash
Political outlet Notus reported on July 7 that US Treasury analysts drafted a non-public report warning of dot-com-era financial system shocks if AI market monetization fails. The report stated AI companies must demonstrate higher productivity and profitability than current levels, or impacts will spread beyond stock markets to datacenter construction financing, private credit markets, cloud providers, semiconductor manufacturers, and power/utility industries. Treasury analysts distinguished current AI firms from dot-com-era speculative companies by noting stronger balance sheets and core business profitability, suggesting financial cushions exist to absorb potential bubble deflation. The assessment comes amid concerns triggered by Apple's recent product price increases and Meta's NeoCloud business entry, which markets interpret as signals of potential AI infrastructure investment slowdowns.
Morgan Stanley Forecasts $1.116 Trillion AI Budget by 2027 Amid Semiconductor Caution
Morgan Stanley projected global AI investment budgets from major hyperscalers will reach $805B in 2026 and $1.116T in 2027, while simultaneously recommending reduced semiconductor stock exposure. The firm argued semiconductor concentration represents a narrow rally peak that has passed, advising short-term hyperscaler overweighting instead. Morgan Stanley cited semiconductor dependence on bigtech investment pace and detected investment slowdown signals from Meta's plans to sell excess AI computing capacity externally. The analysis contrasts with semiconductor demand growth, which increased fivefold since 2023 as five major bigtech companies allocated $848B to datacenter investments in 2025.
JPMorgan Argues Supply Shortages Support Semiconductor Rally Through 2026
JPMorgan characterized recent semiconductor stock declines as market-gifted buying opportunities, projecting global semiconductor stocks will hit record highs during the second half of 2026. The bank's analysis cited persistent robust AI semiconductor demand against production capacity constraints extending through 2028. JPMorgan argued chronic supply shortages enhance pricing power, sustaining semiconductor manufacturer advantages over hyperscalers. Meritz Securities analyst Kim Sun-woo stated the current cycle will see memory supply remain insufficient to match demand growth pace at minimum through Q4 next year. Daishin Securities analyst Ryu Hyung-geun forecast next year's high-bandwidth memory (HBM) average selling prices will rise 91% year-over-year, projecting profitability gap reductions based on absolute advantages.
Penguin Solutions and TSMC Earnings to Test AI Infrastructure Spending
Penguin Solutions will report earnings on July 7 (local time), with guidance and order backlog serving as indicators of whether bigtech companies are executing AI infrastructure construction funding. TSMC will announce June sales on the 10th, coinciding with SK Hynix's US ADR listing the same day. ASML will report earnings on the 15th. TSMC and storage device maker Seagate will release formal quarterly results on the 16th. Analysts emphasized that Q2 semiconductor strong performance was already anticipated, making management's conference call confirmation of specific contracts and future outlook more critical than backward-looking results. Market interpretations of Samsung Electronics' and Micron's record earnings included observations that reliance on price increases rather than volume growth represents typical late-cycle symptoms. Mirae Asset Securities executive director Seo Sang-young stated that if Penguin Solutions earnings demonstrate datacenter construction expansion, semiconductor anxiety could ease, adding that while TSMC results may trigger short-term volatility, significant market contraction probability remains limited.
FAQ
What did the US Treasury warn about AI investments?
A recent non-public US Treasury draft report warned that if AI companies fail to demonstrate higher productivity and profitability, financial shocks similar to the dot-com crash could spread across datacenter construction financing, private credit markets, cloud providers, semiconductor manufacturers, and power/utility industries. Treasury analysts noted AI's deeper integration into the US economy compared to the dot-com period.
When will major semiconductor companies report earnings?
Penguin Solutions will report on July 7 (local time). TSMC will announce June sales on the 10th and formal quarterly results on the 16th. SK Hynix's US ADR listing is scheduled for the 10th. ASML will report on the 15th. Seagate will release quarterly results on the 16th.
Why does JPMorgan recommend buying semiconductor stocks?
JPMorgan argued that AI semiconductor demand remains robust while new production capacity will be difficult to secure through 2028. The bank stated chronic supply shortages enhance pricing power, projecting global semiconductor stocks will reach record highs during the second half of 2026 as semiconductor manufacturers maintain advantages over hyperscalers.
