The AI computing power investment theme, recently battered by fears of a supply glut following reports that Meta is looking to lease or sell its idle AI infrastructure, staged a powerful pre-market rebound in U.S. stocks. This surge was fueled by a U.S. non-farm payrolls report that came in far weaker than economists' forecasts, dramatically cooling expectations for further interest rate hikes from the Federal Reserve.
The latest employment data shows hiring in the U.S. labor market slowed sharply in June, significantly dampening some of the momentum that had been building in job growth this year, despite a drop in the unemployment rate. According to the Bureau of Labor Statistics, non-farm payrolls increased by just 57,000 last month following significant downward revisions to the prior two months' data. This figure is only about half of market expectations and well below the consensus forecast of roughly 110,000 to 115,000. It also falls short of May's revised gain of 129,000 (initially reported as 172,000).
The unemployment rate fell to 4.2% from 4.3% in the prior month, largely due to an unexpected drop in labor force participation. This report indicates that despite signs of robust hiring in recent months, the labor market still faces challenges. While consumer spending data has shown resilience in the face of energy shocks from the Iran conflict, American households remain pessimistic about high prices and wage growth that hasn't kept pace with inflation, which may in turn be causing employers to exercise caution in hiring.
The combined downward revision of 74,000 jobs for April and May suggests previous months' employment strength was clearly overestimated. Following the payrolls release, S&P 500 index futures turned sharply higher, and popular AI-related technology stocks rebounded strongly. Notably, shares of CoreWeave, which was hit hard yesterday, rose by over 3% at one point. U.S. Treasury yields fell significantly as investors and interest rate futures traders dramatically scaled back their bets on Fed rate hikes this year.
Market expectations for Fed tightening have contracted noticeably, with the anticipated number of rate hikes in 2026 being cut from three to just one. The expected timing of the initial hike has also been pushed back significantly from the most aggressive previous forecast of July to December or even early next year.
The newly released weak payrolls data provides a macro-level narrative reversal catalyst for the AI computing power supply chain, which had just been hit by the pessimistic news of "Meta selling excess AI computing resources." The soft jobs report reduces the tail risk of further rate hikes and lowers the discount rate for long-duration assets. Combined with the market's renewed effort to distinguish between the narratives of "computing power oversupply panic" and "AI cloud service monetization capability," this forms the core logic behind the AI computing power chain's pre-market rebound.
Furthermore, Wall Street analysts are unanimously firm in their belief that Meta's sale of computing power, along with SoftBank's establishment of SB Neo to enter the U.S. AI cloud market, is not a bet on "computing power oversupply" but rather a long-term wager on the expansion of AI training and inference demand into the gigawatt-scale infrastructure era.
U.S. Job Market Hits the Brakes: Era of 'Low Layoffs, Low Hiring' Arrives
The slowdown in hiring was primarily driven by the leisure and hospitality sector recording its largest job loss since 2020. Retail trade and information industries also shed jobs, while healthcare and social assistance continued to hire strongly. Thursday's report showed the labor force participation rate—the share of the population that is working or looking for work—fell to 61.5%, its lowest level in over five years. The Bureau of Labor Statistics noted that after accounting for population adjustments, this measure has changed little over the past year.
The participation rate for prime-age workers, those aged 25 to 54, fell to 83.3%, matching the lowest level since 2023. Employment in manufacturing and construction rose in June. Many economists note that while residential construction continues to be suppressed by high interest rates, the data center construction boom expected in 2026 could be a potential driver of demand for construction labor.
Simultaneously, some major tech companies like Meta Platforms Inc. and Microsoft Corp. are implementing layoffs, partly to offset massive spending on artificial intelligence. Employment in the information sector continued to decline, marking the 17th drop in the past 18 months. Employment in financial activities showed little change. This sector is another major employer of white-collar workers, a group seen as among the most vulnerable to automation.
Average hourly earnings rose 3.5% from a year earlier. Economists are closely watching how supply and demand dynamics in the labor market affect wages, especially as inflation begins to outpace wage growth in a range of industries. With the resumption of U.S.-Iran peace talks and a sharp drop in oil prices, global consumer confidence is on a significant recovery path, which may encourage employers to accelerate hiring in the coming months.
Another report released Thursday showed little change in the number of people applying for unemployment benefits last week. Layoffs have remained low in recent years, contributing to what economists call a "low-layoff, low-hire" labor market.
A team of economists from Bloomberg Economics stated after the payrolls release: "The June jobs report sends mixed signals but overall indicates a stabilizing labor market. While job growth was slightly below expectations and prior months' data were revised down, the underlying trend remains strong and is still above most estimates of the breakeven point."
"The June jobs report is clearly disappointing, but this report shouldn't shake anyone's view of the overall economic outlook," said Neil Dutta, chief economist at Renaissance Macro Research, in a note. "The main takeaway from this report is that the labor market is a reflection of the overall economy. Growth is uneven, and therefore the labor market is uneven."
Dual Blows to the Fed's Hawkish Path
For the Federal Reserve's hawkish monetary policy path, the impact comes not only from the payrolls data but also from the reopening of the Strait of Hormuz. On Thursday, four supertankers loading crude oil at Saudi Arabia's primary oil and gas export hub appeared in the Gulf of Oman. This is the largest number of departures since a peace agreement took effect roughly two weeks ago. Furthermore, Saudi Arabia's crude oil exports have surged to near pre-war levels since resuming loadings on tankers within the Persian Gulf. This provides further evidence that supply from producers in the region is recovering following a mid-term peace agreement between the U.S. and Iran.
U.S. CPI data for May showed the energy index rose 3.9% month-over-month, with gasoline up 7.0%. The energy component contributed over 60% of the overall CPI increase for the month. Over the past 12 months, energy prices are up 23.5% and gasoline is up 40.5%, indicating that the previous oil price shock has substantively lifted inflation readings. The EIA, in its latest Short-Term Energy Outlook, also continues to note that higher global crude prices are pushing up expectations for U.S. refined product prices, with wholesale prices for diesel and jet fuel revised significantly higher compared to pre-conflict February forecasts, and gasoline price expectations also raised noticeably.
Therefore, the monetary policy implications of the unexpectedly weak payrolls data, coupled with the gradual reopening of the Strait of Hormuz, lie in significantly reducing the urgency for the Federal Reserve to hike rates further, not in immediately opening the door to significant rate cuts. Only when falling energy prices further lower inflation expectations, and core services, wage, and housing inflation cool in tandem, will the Fed have more sufficient reason to shift from a "highly restrictive rate" stance to a true easing cycle.
Tech Bulls Regain Narrative Control
The disappointing non-farm payrolls data, which caused a major cooling of rate hike expectations, has sounded the charge for a rebound in the AI computing power supply chain. The addition of only 57,000 jobs in June, significantly below the latest survey expectation of 115,000, coupled with the downward revision of May's job gains from 172,000 to 129,000, is sufficient to show labor demand is clearly cooling from its previously strong state. While the unemployment rate fell to 4.2% and layoffs remain low, preventing this data from being a "recession confirmation," it is enough to weaken the Fed's urgency to continue hiking rates.
Additionally, wages rose 0.3% month-over-month and 3.5% year-over-year, not yet giving the Fed a signal that "inflation is completely defeated," but enough to remove the most important pillar supporting further rate hike trades. The pre-market collective rebound in the AI computing power chain reflects the market trading on a clearer logic: U.S. economic and labor market growth trajectories haven't collapsed, employment has clearly cooled, energy risk premiums are receding, the probability of further Fed rate hikes is being suppressed, and the pressure on DCF discount rates for global tech risk assets, led by the AI computing power chain, is being marginally released.
As the weaker-than-expected jobs report eased market concerns about further Fed rate hikes this year, for equity strategy, this means the short-term flow of funds may shift from "guarding against inflation, betting on a rate hike cycle" to an AI computing power thematic trading framework focused on "betting on high-quality growth tech stocks and betting on AI computing power chain leaders with real AI cash flows that are sensitive to interest rate expectations."
The increasing leverage and crowded positioning in the AI semiconductor trade theme, coupled with rising pricing pressure for consumer electronics leaders like Apple, was accompanied by the Philadelphia Semiconductor Index falling as much as 7.9% in a single day and experiencing multiple sharp swings exceeding 5% within a month. This highlights that the AI computing power chain linked to semiconductors has entered a phase of high volatility, extreme leverage and crowded bullish positioning, and high pressure to meet expectations. Combined with Meta's shift to selling computing resources, this is why institutional investors have recently begun emphasizing overly pessimistic bearish narratives like "the AI semiconductor trading frenzy has peaked" and "the AI bubble is gradually bursting."
However, the prominent Wall Street investment firm Nomura released a new research report on Wednesday refuting the "semiconductor peak theory." Nomura's key rebuttal is not simply stating that AI chips will continue to rise, but pointing out that AI cloud infrastructure demand is spreading from a single-point GPU shortage to a systemic component mismatch.
According to Nomura's research framework, AI server revenue is projected to grow 78% and 76% in 2026 and 2027, respectively. The number of global data center projects is expected to increase from 240 to 280, with about 50 being gigawatt-scale projects. New compute capacity deployment in 2027 is forecast at 32GW, with visibility of 23GW already for 2028. However, the real bottleneck is spilling over from GPU capacity and TSMC's CoWoS advanced packaging to wafer-level substrates, AI PCBs, copper-clad laminate (CCL), electronic fabric, MLCCs, glass substrates/ABF substrates, IC substrates, high-end capacitors, power management chips, and high-speed optical interconnect components for data centers.
It is reported that McKinsey's medium- to long-term calculations also support Nomura's emphasis on this direction: by 2030, to meet AI-related demand alone, the global computing power value chain will need to invest approximately $5.2 trillion into data centers, corresponding to a demand for about 156GW of AI-related data center capacity.
This means the main theme of the semiconductor trade is not "peaking," but "rotation of shortage points": from GPUs to HBM, from advanced packaging to substrate materials, semiconductor equipment and raw materials, data center CPUs and optical interconnect systems, to power, liquid cooling, networking, and cloud orchestration software. Earnings revisions and pricing power expectations may still be the strongest catalysts for the core hardware chains related to AI computing power.
On a deeper level, Meta and the SoftBank Group founded by Masayoshi Son may be betting on long-term scarcity of computing resources as AI workloads shift from training centers to inference centers. McKinsey estimates that by 2030, global data centers will require about $6.7 trillion in investment to meet computing power demand, with about $5.2 trillion in data center capital expenditure related to AI inference processing loads. The International Energy Agency expects global data center electricity consumption to double to about 945 terawatt-hours by 2030, with electricity use from AI-driven accelerated servers growing at an annual rate of about 30%. Wall Street giant Goldman Sachs also expects U.S. data center power demand to rise from 31 gigawatts in 2025 to 66 gigawatts in 2027.
Comments