The U.S. CPI downturn has not reached a turning point yet! China Securities Co., Ltd.: Under normal circumstances, AI and oil price disturbances continue to be monitored, with attention to the base year switch next year.

China Securities Co., Ltd. released a research report stating that AI has a stronger transmission capability downstream, but for now, the increase in oil prices has a greater impact on inflation. The core CPI increased by 0.2% year-on-year, while the total contribution of AI and energy combined was close to 0.5%, explaining all the inflationary pressures, with the remaining sub-items overall trending downwards. Looking ahead to the second half of the year, under the baseline scenario, while the year-on-year contribution of oil prices may decrease, it will still remain significant; AI's contribution may double or more; the overall impact of both will be maintained at a relatively high level. Expectations for a decline in inflation require: a decrease in oil prices, deflation in core commodities, a switch in the base year for oil prices. Abstract The rising concerns about inflation in the United States in 2026 are mainly driven by two cost factors: the significant increase in energy prices and the soaring prices of AI-related raw materials. How much have these two factors contributed to the recent increase in core inflation, and how significant will their ongoing impact be? I. Overall situation of the upstream costs driven by AI and energy Energy prices have seen a significant increase, driven by the rise in oil and electricity prices. In the past year, WTI crude oil prices have risen by nearly 65%, while electricity prices have increased by 3.57%, reflecting a simultaneous increase in energy-related components in the CPI. However, with the recent significant drop in oil prices, these concerns have alleviated significantly. The increase in AI-weighted costs has been moderate, mainly due to significant differentiation in the internal structure. While storage prices have soared, their weight is low, with hardware prices, which have the highest weight, running at low levels. II. Transmission of AI and energy to core inflation Sub-items directly related to AI and energy downstream have shown signs of increase this year, indicating some level of transmission. AI-related sub-items such as computers and peripheral devices and financial services (with significant investment in AI technology by institutions) have seen price increases, while some electricity-related sub-items such as laundry services, personal care, and accommodation have also shown signs of price hikes. However, examining the cost increases of AI and energy and their comprehensive diffusion in core inflation only by tallying highly correlated sub-items is not enough. By using a multivariate regression model incorporating unemployment rates and the federal funds rate, we estimated the portion of core CPI that can be explained by AI and energy. (1) Overall level: AI transmission capability > energy, but momentum pull higher energy > AI First, we investigated the overall transmission of AI and energy to core CPI, including transmission coefficients and momentum pull. Transmission coefficients: AI is greater than energy. This metric measures the ability of each cost factor to spread downstream under comparable price increases. The results of the regression analysis show that the transmission coefficient of the recent increase in oil prices to core inflation downstream is only 0.5%, while the coefficient for AI price increases is as high as 4.6%. Overall momentum pull: temporarily, energy > AI, but both factors can explain the entire increase in core inflation; with the drop in oil prices, energy-induced inflation may decrease significantly in the second half of the year. In the May year-on-year increase in core CPI, energy contributed by approximately 0.31% (with a one standard deviation range of 0.22% to 0.4%), while AI contributed only 0.12% (with a one standard deviation range of 0.06% to 0.18%), totaling about 0.43%. Energy has a larger contribution, primarily due to significant increases in oil prices. Compared to the end of last year, the year-on-year increase in core CPI is around 0.2%, with the oil price-related portion increasing by 0.39% and the AI-related portion by 0.08%, totaling 0.47%, well above 0.2%, meaning that AI and energy together can basically explain all of this year's increase in core CPI, with other remaining components showing a slowdown in growth, possibly due to negative feedback from rising costs. Negative feedback from cost increases on downstream consumer prices is more visible in the impact of oil prices. Airfare prices are significantly affected by aviation fuel prices, and the direct transmission of rising oil prices is evident. Calculations based on the increase in airfare sub-items and their weights show that airfare contributed 0.35% to core CPI in May, nearly 0.36% higher than at the end of last year, surpassing the overall impact of energy. As a result, energy may have a restraining effect on consumer goods other than airfare, leading to a slowdown in price hikes, or even deflation. Historical data shows that during the rapid increase in oil prices, there was indeed a slowdown in the growth of consumer spending on energy-related commodities other than airfare. (2) Structural level: AI has a broader diffusion in the service industry, while energy is concentrated in the transportation sector AI not only has a higher overall diffusion level downstream, but also has a wider diffusion breadth, especially in some service sub-items. As more companies invest in digitization, the pricing impact of AI may be more sustainable. AI not only has a noticeable pushing effect on consumer goods related to computers, but also shows high transmission intensity in professional services, telecom services, financial services, and educational communications sub-items, likely related to increased investments in digitalization by companies. Therefore, the impact of AI on prices may be more sustainable (with increased participation from more companies and more purchases) rather than being a one-time cost push. Energy, on the other hand, is still largely concentrated in the transportation sector, where crude oil costs are high, making its impact relatively narrow and more of a one-time shock. The transmission intensity of oil prices to various sub-items of core CPI is not extensive, with significant impacts mainly seen in airfare, public transportation, and transport services, while other sub-items show significantly lower transmission intensity. Notably, sub-items such as new cars and used cars respond negatively to oil prices, consistent with the logic of consumer feedback and price suppression noted earlier. (3) Historical comparison: This round of AI influence is weak, while energy is similar to historical levels, indicating that the intrinsic momentum of inflation is not strong Historically, there have been many periods similar to this year with increases in AI costs and oil prices. We compared the strength of transmission in these cases. Out of six chosen time windows, some show instances of isolated oil price increases or AI increases, while others show synchronized movements of oil prices and AI. The results indicate that the transmission of energy to core CPI in this round is roughly in line with historical levels, weaker than in 2022 and 2017; whereas the transmission of AI in this round is relatively weak compared to historical levels. The reasons behind this may be related to the overall weak economic conditions in the United States this year, particularly the weak consumer demand, leading to significant resistance in the transmission of upstream costs to downstream consumers. The overall intrinsic momentum of inflation is not particularly strong. III. Prospects for inflation with falling oil prices and high AI costs In the baseline scenario, the pull from oil prices may decrease in the second half of the year, but the pull from AI may increase, leading to a marginal decrease in the overall pull on core CPI. Assuming oil prices remain stable and rise to $75 per barrel by the end of the year; in terms of AI costs, assuming storage prices continue to rise by 30% and hardware costs drop by around 3%, while other costs remain the same, the overall increase is estimated to be about 1.8%. Despite the drop in the centrum of oil prices in Q4 last year, the year-on-year contribution may decrease slightly but remain significant. On the other hand, the contribution from AI is expected to continue to rise, approaching a doubling by May. Short-term risks calculated suggest a bias towards an upper trend, with particular attention to the sustainability of AI impacts; while downside risks mainly involve further decline in the remaining segments under demand pressures. On one hand, the subsequent development of the US-Iran conflict remains uncertain, and the stability of oil prices at lower levels needs to be observed. On the other hand, as AI penetration continues, it may lead to price increases in more industries. Since the impact of AI is not necessarily a one-off disturbance, it remains a key focus for the future. Looking ahead, the anticipation of substantial year-on-year decline in inflation may require: a significant drop in oil prices, a return to deflation in core commodities, or waiting for the switch in the base year to occur next year. Risk warning Unexpected increase in US inflation, stronger-than-expected economic growth in the US leading to continued tightening of monetary policy by the Federal Reserve, substantial appreciation of the US dollar, rise in US bond yields, continued decline in US stocks, commercial bank bankruptcy crisis, and currency and debt crisis in emerging markets. Unexpected recession in the US economy leading to liquidity crisis in financial markets, forcing the Fed to switch to a more accommodative policy. Energy crisis in Europe escalating beyond expectations, deep recession in the Eurozone, global market turmoil, shrinking external demand, and policy facing dilemmas. Escalation of global geopolitical risks, worsening US-China relations beyond expectations, uncontrollable factors affecting commodities and shipping, further deepening of deglobalization, continuous disruption in supply chains, and worsening competition for resources.