For the most sensitive issue regarding AI, there is considerable disagreement within Goldman Sachs, but "even if it's a bubble, it will last a long time."

Under the belief in artificial intelligence, tech giants expect that in the coming years, AI capital expenditures will reach an unprecedented scale of $10 trillion. These investments will mainly focus on key areas such as data centers, chips, infrastructure, and the power grid, aiming to lay a solid foundation for the future development of AI.

As global tech giants ramp up their AI investments, a report from Goldman Sachs reveals the uncertain prospects of AI investment returns.

Goldman analysts including Daron Acemoglu released a report on Tuesday stating that apart from efficiency improvements reported within the developer community, the actual benefits of AI technology remain elusive. Even companies benefiting the most in the AI field, such as NVIDIA, have experienced drastic stock price fluctuations, reflecting the market's concerns and uncertainties regarding AI investment returns.

In this report, some analysts hold an optimistic view on the long-term economic potential of AI technology, believing that although the "killer applications" of AI have not yet emerged, it will eventually surpass the current "tools and shovels" stage, generating greater economic returns. However, some analysts are skeptical. They believe that the development of AI technology may not be as rapid as expected, and its cost-effectiveness may not be as attractive as imagined.

Looking ahead to the next decade, skeptics predict that AI can only increase U.S. productivity by 0.5%, contributing only 0.9% to GDP growth. Optimistic analysts, on the other hand, predict that generative AI will ultimately automate 25% of job tasks, increase U.S. productivity by 9%, and boost GDP growth by 6.1%.

Despite significant differences in opinions, Goldman Sachs still believes that even if the fundamental narrative of AI technology ultimately fails to hold ground in the capital markets, the AI bubble may still take longer to burst.

Voices of Doubt: AI Automates Less Than 5% of Tasks

As one of the skeptics, Goldman analyst and MIT professor Daron Acemoglu believes that in the next decade, only a quarter of AI-related tasks will achieve cost-effective automation, meaning that AI's impact on all tasks will be less than 5%.

Acemoglu argues that historical trends of technology improving over time and costs decreasing cannot simply be applied to AI, as advancements in AI models may not progress as rapidly or impressively as many expect.

He also questions whether AI technology can reach the most valuable cognitive abilities of humans, especially considering that AI models are typically trained based on historical data, which may limit their ability to replicate the range of complex cognitive abilities of humans.

Acemoglu predicts that over the next decade, AI will only increase U.S. productivity by 0.5%, contributing only 0.9% to GDP growth.

Jim Covello, Goldman's global head of stock research, is even more pessimistic, believing that AI technology is costly and not necessarily designed to solve complex problemsCovello pointed out that compared to the early days of the internet, the cost of AI technology has not shown a downward trend over time, and he doubts that the cost of AI technology will decrease to a level sufficient to automate a large number of tasks.

Optimistic View: AI Automates 25% of Tasks, U.S. GDP to Increase by 6.1% in the Next Ten Years

Goldman Sachs analyst Joseph Briggs holds a more optimistic view, predicting that generative AI will eventually automate 25% of work tasks and increase U.S. productivity by 9% and GDP growth by 6.1% over the next decade.

Briggs believes that although many AI-related tasks currently do not have cost-effectiveness in automation, the long-term cost reduction potential of new technologies, as well as the possibility of labor reallocation and the creation of new tasks, will drive more AI automation.

Goldman Sachs' Kash Rangan and Eric Sheridan are also enthusiastic about the long-term transformation and return potential of AI. They believe that despite the massive investments in AI infrastructure by large tech companies, there is no sign of irrational exuberance.

Rangan emphasizes that the current capital expenditure as a proportion of income is not significantly different from previous technology investment cycles, and investors will only reward companies that can monetize AI.

Development Bottleneck: Chip and Power Supply Shortages

Goldman Sachs analysts generally believe that shortages in chips and power supply may limit the development potential of AI technology.

Among them, a Goldman Sachs semiconductor analyst believes that chip growth will be limited in the coming years due to shortages of key components such as high-bandwidth memory technology and chip packaging.

A bigger issue is whether the power supply can keep up.

Goldman Sachs utility analysts expect that the widespread adoption of AI technology and necessary data centers will drive a significant increase in power demand. However, U.S. utility companies have experienced almost no growth in power consumption over the past twenty years and are dealing with an aging power grid, so they may not be ready to meet the upcoming surge in demand.

How Long Will the AI Bubble Last?

Despite doubts about the economic benefits of AI technology, Goldman Sachs analysts unanimously agree that even if the fundamental narrative of AI technology ultimately fails to hold ground in the capital markets, the AI bubble may still take longer to burst.**

Goldman Sachs stock strategist Ryan Hammond believes that AI concept stocks have more room to run, and he expects that beneficiaries of AI will continue to expand, not only semiconductor giants like NVIDIA, but also large utility companies.

Looking ahead, Goldman Sachs' chief portfolio strategist Christian Mueller-Glissmann analyzed the impact of AI on the market from a macro perspective.

He found that if AI significantly accelerates economic growth and corporate profitability without exacerbating inflation issues, the long-term return of the S&P 500 index will be above average.

However, he warned that realizing the potential of AI technology as expected by investors is crucial to boosting market returns. Although AI may benefit stocks by increasing productivity growth, the market often anticipates this before actual productivity growth is realized, increasing the risk of overpricing