Will the resilience of the US economy and the expectation of a strong US dollar reverse, causing a negative impact on global technology stocks?

Report Summary

AI and big data are the fourth general technological innovation, with the previous reference history being the computer revolution.

Recently, the market has once again focused on the 90s dot-com bubble. What are people discussing?

• NVIDIA has experienced its largest drawdown in 23 years (-25%). The increased volatility in stock prices indicates growing divergence. How do the valuations, performance, and stock price drivers of AI giants compare to those of dot-com giants back then?

• The expectation of a US recession in 24Q3 is rising, with the Fed gradually approaching rate cuts and Japan raising rates. The strong US dollar is showing signs of loosening. Does it resemble the period before and after the bursting of the dot-com bubble in 99-00? (See red arrow in the figure below)

Revisiting the similarities between the 90s dot-com bubble and the current situation: Technological core, economic environment, economic structure, USD liquidity, USD strength, stock market liquidity, policy drivers, industry progress

The dot-com bubble was not built in a day, nor did it burst all at once. Since 1998, a series of negative signals have not reversed the frenzy of the bubble.

The breeding ground for the bubble: Stable economy, growing advantages, strong US dollar

The loosening foundation: This turning point appeared between 1998 and 2000, with inflation leading to rate hikes that suppressed growth. When the expectations of resilient growth and a strong US dollar changed, global capital flows became the straw that broke the camel's back. Although the bubble did not burst overnight, the reversal of expectations triggered a spiral.

Current focus: Will there be a reversal in the resilience of the US economy and the expectations of a strong US dollar (accompanied by similar rate hikes in Japan), leading to an exacerbation of global capital outflows? Will this have a negative impact on global tech stocks?

We understand that there is no definitive answer to this question, but we can try to think from two dimensions: first, whether the size of the bubble has reached the level of the dot-com era; second, whether the force to burst the bubble has reached the level of that time.

• From the perspective of the turning point in economic, inflation, and liquidity expectations, the macro environment is relatively close to the US dot-com period around 1998-2000, which will amplify the volatility of current tech leaders.

• Looking ahead, in the process of the US economic slowdown, the risk of significant downturn (significant recession) is relatively small. Before a consensus on the recession judgment is formed, global capital flows will not be as drastic as in 2000.

• From the perspective of the degree of excess liquidity in the US stock market, the attractiveness of US stocks vs. US bonds to global capital, the stock-bond spread in the US, and the equity risk premium of US stocks, the current degree of US stock market bubble is not as high as it was back then Therefore, whether it is the force that catalyzes the bursting of the bubble (significant recession and weakening of the US dollar) or the actual outflow of fund risks (US stock market liquidity), it is still incomparable to the year 2000.

Comprehensive Comparison of AI Seven Sisters and 90s Tech Giants in terms of valuation/growth/stock price drivers/max drawdown: As the radiation scope of the AI industry has not reached the level of the tech giants in the 90s, looking at indicators such as the market value proportion of the Seven Sisters, valuation, and the slope of the first wave of growth, it has indeed surpassed the representation of the leading tech giants in the 90s, posing a risk of localized bubble formation. This also raises higher requirements for the sustainability of AI performance explosion and the progress of future commercialization.

The figure below shows that the market value of the top 10 tech giants in the 90s was approximately 22% of the US GDP, while the market value of the AI Seven Sisters is approximately 54% of the US GDP.

Stock price evolution rhythm of the leading tech giants in the 90s, and insights from overseas mapping (Japanese technology): Cases such as Microsoft, Intel, IBM show that transitioning from technology to commercialization, especially finding TO-C application scenarios, is a significant driver for the valuation of tech giants. Hardware with difficult cost reduction will gradually fall behind in competition (Japanese Dram semiconductor).

The high prosperity of the tech giants is the catalyst for the bubble, but the high prosperity of the tech giants in the later stages of the bubble has become unsustainable, and stock prices continue to rise.

In addition, we can also see different paths of overseas mapping of the tech giants from Japan's TMT industry development in the past (good companies with performance and good stocks, as well as tech stocks without performance).

Report Content

Introduction: Revisiting the Lessons of the 90s Tech Bubble

We are experiencing the fourth general technological innovation, with the previous reference history being the computer revolution. AI technology, as a general technology, can have a wide-ranging impact and empower various industries, bringing about capital deepening and total factor productivity (TFP) improvement. Since the emergence of AlphaGO in 2016, economists around the world have likened the upcoming era of artificial intelligence to the "Fourth Industrial Revolution."

The birth, evolution, bursting, and overseas mapping of the 90s tech bubble have always been a very valuable reference and learning history in global tech stock investments.

In the April 23rd issue of "Dancing with the Wind: Looking at the AI Singularity Moment from Classic Tech Bulls," we compared the A-share and US tech market conditions with the 90s tech bubble from various aspects, and detailed the technical innovation nodes and commercialization rhythm accompanying the main uptrend of the leading tech stocks in the 90s.

In the July 23rd issue of "Japan's Interpretation of Barbell Strategy: 90s Japanese Technology," we focused on the overseas mapping of the computer revolution, especially the development and stock performance of Japan's hardware, software, internet, and other industries In August 24, 20XX, some new global changes have once again drawn investors' attention to the history of the 90s tech bubble, with discussions focusing on—

1. How did the "US recession expectations" and "Fed rate cut trades" in the 90s impact the tech industry? If there is a reversal in expectations of US economic resilience or a strong dollar, would it pose risks to tech stocks?

2. NVIDIA experienced its largest drawdown (-25%) since the AI wave began, followed by reaching new highs. Does the amplified stock price volatility of benchmark companies imply increasing divergence?

3. In comparison to the current tech giants, what are the differences and insights between the current AI "Seven Sisters" in terms of performance, valuation, upward slope, and stock price drivers?

I. Parallels between the 90s Tech Bubble and the Current Situation

During the brewing of the tech bubble, there are significant similarities and some differences between the economic, policy, industry, and technological backgrounds of the US in the 90s and the current US tech market:

US Economic Environment—In the 90s, the US saw stable growth and low inflation, holding a comparative advantage globally, while post-pandemic US economic recovery also leads globally but with volatile economic expectations.

USD Liquidity—In the 90s, there was extreme loose monetary policy and a strong dollar, whereas currently, there is a shift from tight to loose monetary policy, and the dollar remains strong.

Stock Market Liquidity—During the tech bubble era, there was a surge in IPO numbers with ample investor liquidity, while currently, micro-market risk appetite is increasing, but liquidity is not as abundant as before.

Industry Driving Forces—In the 90s, policy played a key role with the "Information Superhighway" guiding top-level design, whereas AI is currently more driven by global technology, with a neutral stance from the Biden administration.

Industry Development Context—The computer revolution in the 90s progressed from semiconductors to PCs to hardware and operating systems to the internet, while AI is currently in the infrastructure construction and hardware demand surge phase.

The burst of the tech bubble in 2000 was actually a concentrated outbreak after turning a blind eye to several negative signals. In fact, before the burst of the "tech bubble" in March 2000, there were continuous negative impacts in terms of profitability, liquidity, policy, and risk appetite:

Negative Signals on Profitability: (1) Starting from the third quarter of 1999, more tech companies began to show declining profitability in their financial reports, and the growth rate of US internet users in 1999 also significantly decreased; (2) After entering 2000, the previously worrisome "Y2K bug" did not have a widespread impact, and the expected large-scale replacement wave did not materialize, continuing to suppress demand for tech products; (3) Financial data released from March 2000 showed poor sales performance of tech products during the 1999 Christmas holiday season, with many companies' profits significantly lower than expected, leading to the bankruptcy of many once-prominent tech companies Negative signals of liquidity: (1) In the fourth quarter of 1998, US inflation began to rise and accelerated in 1999. The US Treasury bond yield also started to rise from October 1998 (the ten-year bond yield increased from 4.2% in October 1998 to 6.4% in March 2000); (2) In the first quarter of 1999, the growth rate of US M2 money supply began to peak and fall.

Negative signals of policy: (1) On June 30, 1999, the Federal Reserve began to raise interest rates and raised rates five more times in the following six months; (2) 2000 was the last year of President Clinton's term, and the market began to worry about whether his vigorously promoted new economic policies could continue in the future.

Negative signals of risk appetite: (1) Influenced by the tightening monetary policy, US large-cap stocks began to fluctuate downward from early January 2000, three months ahead of the NASDAQ index; (2) In February and March 2000, US media extensively reported that the Department of Justice's antitrust charges against Microsoft Corporation would face a verdict, stating that a large amount of evidence indicated that Microsoft's monopoly behavior was established, causing market panic.

In comparison, the partial underperformance of leading companies' financial reports, policy uncertainty after the presidential transition, the Federal Reserve's interest rate hike turning into a cut, and escalating recession concerns all bear some resemblance to the situation at that time.

Moreover, Nvidia's more than 25% stock price retreat in August intensified market concerns about global technology stocks.

II. How was the valuation of the tech bubble before and after the US economic recession?

In August 24 years ago, some new global changes brought investors' attention back to the history of the 90s tech bubble. How did the "US recession expectations" and "Fed rate cut trades" in the 90s affect the tech bubble? If there is a reversal in expectations of US economic resilience or a strong US dollar, would it pose a risk to tech stocks?

(1) The breeding ground for the bubble: Stable economy, growth advantages, strong US dollar in the 90s

The 90s tech bubble was the result of the combination of favorable timing, location, and people. However, to delve into its essence, the US economy in the 1990s was in a relatively stable and worry-free development cycle, serving as the breeding ground for the entire tech bubble.

Compared to the 1970s and 1980s, the economic fluctuations in the US in the 1990s were not significant, showing strong stability and maintaining a positive GDP growth for a long time.

During the incubation period of the "tech bubble" from 1991 to 1998, the US economic growth rate was not particularly outstanding, but it had strong stability. After the 1990s, the peak growth rate of the US GDP generally hovered around 4.5%, while in previous economic cycles, the GDP peak was generally above 6%. On the other hand, the economic fluctuations in the US in the 1990s were not significant, showing strong stability. The GDP fluctuated narrowly between 2.5% and 4.5%, which was incomparable to previous economic cycles (where GDP generally experienced negative growth) In addition, at that time, the inflation level and unemployment rate in the United States were both in a continuous downward trend. Therefore, it was a rare period of "stable growth, low inflation, low unemployment", which can be described as the "golden period" of economic development.

This economic structure had a relative advantage globally, attracting global capital inflows and strengthening the US dollar, which also brought a continuous influx of incremental funds to the internet industry. From 1992 to 1998, the advantage of the US economy over other economies around the world became increasingly apparent, continuously attracting overseas funds to purchase US dollar assets, and the US dollar index also entered a clear upward trend starting from 1995.

This relative global growth advantage is very similar to the situation in the United States after the epidemic.

After 2020, due to the resilience of the economy during the recovery from the epidemic and the high interest rate environment driven by high inflation, the US dollar also remained strong. After the epidemic, the trend of US economic growth had a relatively competitive advantage globally, similar to the situation in the 1990s with a strong US dollar.

(II) Loose Foundation: The expected turning point appeared in 1998-2000, with the loosening of the foundation of resilient growth and a strong US dollar

The turning point in the expectations of the US economy and the US dollar trend during the internet era occurred from 1998 to 2000.

From 1998 to 2000, the United States experienced a round of macro expectation shifts: inflation rose in Q3 1998 - the Fed raised interest rates in June 1999 - inflation was contained in Q1 2000 - recession pressure increased in Q3 2000 - interest rates were cut again in January 2001.

After 1995, the Fed initiated the second round of easing cycle, followed by the third round in 1998. In 2000, it confirmed the turning point from fighting inflation to recession expectations dominating the easing, during which the peak of the Nasdaq bubble was seen.

The third monetary cycle began with a rate cut on September 29, 1998 (from 5.5% to 5.25%), and the Nasdaq index entered a rapid rise phase from early October, acting as a direct catalyst;

In the first half of the tightening phase of monetary policy (2000), it failed to stop the expansion of the bubble, but the turning point from rate hikes to rate cuts (2000-2001) corresponded to the peak of the bubble.

Coincidentally, Japan also had a rate hike point in 2000, and the relative change in expectations between emerging markets and the US economy determined the global rebalancing of fund allocation

This is similar to the situation in the United States from 2020 to 2024: 40 years of unprecedented inflation—harsh interest rate hikes—inflation easing—rising recession expectations—about to start rate cuts.

With the Federal Reserve maintaining high interest rates, orderly cooling of inflation in the United States is expected in 2024, returning to the "2% range." In addition, in recent months, the US labor market has significantly cooled down, with non-farm data repeatedly falling below expectations and previous data being continuously revised downward. The unemployment rate has risen to 4.1%, triggering the "Sam Rule." On August 23, at the Jackson Hole meeting, Powell mentioned that the US labor market is undeniably cooling down, and he does not want to see it weaken further. Therefore, the market has started trading "recession expectations" and "rate cut expectations."

The current question to consider is whether we are currently at a turning point of similar expectations? Or, have the foundations of resilient US growth and a strong US dollar loosened?

Looking at the longer cycle, from 2009 to 2023, there has been a major tightening cycle by the Federal Reserve, which has also been a driving force behind the continuous rise of the Nasdaq. The rate hike cycle in 2022-2023 only briefly halted the upward trend of the Nasdaq, more so reflecting the relatively strong fundamentals and interest rate center of the United States.

This upward movement of the long-term interest rate center has been accompanied by global funds flowing into the United States, and now we are at a turning point of the short-term interest rate center.

(Three) Current Issue: Will the resilience of the US economy + expectations of a strong US dollar reverse? Will it trigger global fund volatility?

The continued trend of funds flowing into US stocks raises the question of whether there will be a phase of rebalancing that we need to focus on. By observing the net flow changes of 14 typical and large-scale US stock ETFs (including broad-based ETFs and technology ETFs) in the US stock market, we can see that in the first half of July, funds were mostly flowing in, but after the Bank of Japan's rate hike on July 31st, funds began to flow out until early August. After mid-August, the trend stabilized again.

(Four) Funds are pouring into US dollar assets at an accelerated pace, but in the 90s it was inflows into equities, this round is increasing holdings of US bonds

The anchor of risk-free assets, namely the trend and position of US bond yields, is somewhat different—looking at the medium-term perspective of 10-15 years, US bond yields in the 90s continued to decline and are currently at a high point in nearly 15 years.

From a medium to long-term perspective, US bond yields in the 90s were in a continuous channel of fluctuating decline, with differences in the extent of liquidity abundance—during the "dot-com bubble" brewing period from 1991 to 1998, US Treasury yields dropped from 8% to 4.2%, and the federal funds target rate dropped from 6.75% to 5% The level of discount rate and the direction of marginal changes in the discount rate will affect the theoretical valuation center of US technology companies, as well as the asset price ratio between US bonds and US stocks.

In the 1990s, there was a significant inflow of equity assets: the scale of US mutual funds expanded by 7 times, while the scale of equity funds expanded by 20 times. The 401k plan allowed individuals to invest part of their wages in the stock market, leading to the rise of stock investments in residents' asset allocation, surpassing bonds. The proportion of US stocks held by residents through mutual funds continued to increase.

According to ICI, in 1990, the scale of US mutual funds was 1.1 trillion, with the scale of equity funds only at 0.2 trillion, accounting for 22.4% of the total mutual fund scale. With the continuous inflow of domestic and foreign funds into US stock equity funds, the asset size of US stock equity funds increased from 0.2 trillion US dollars in 1990 to 4.0 trillion US dollars in 2000, expanding nearly 20 times, with its proportion of the total mutual fund scale also rising to 57%.

Driven by the dual factors of economic resilience and technological trends after this round of the pandemic, global funds continue to flow into the US stock market. However, compared to the growth of stock assets from 1990 to 2000, the current growth rate is relatively moderate. According to US balance sheet data, the equity assets held by the US private sector expanded by 5 times from 90-99, with an annual growth rate of 17%, while in the 20-plus years from 2000 to the present, it only expanded by 4 times, with an annual growth rate of 8.3% from 19-23, indicating that although US stocks still attract global funds, their growth rate has significantly slowed down.

According to the US International Capital Flow Report, from 2021 to 2023, global capital continues to flow into the US bond market, while the net purchase amount of US stocks is not at a high level. Against the backdrop of global deglobalization, especially influenced by the pandemic and geopolitical uncertainties, US bonds, as risk-free assets, have further enlarged their certainty premium, attracting more global capital inflows into the US bond market.

(V) The degree of stock market liquidity is not as excessive as before, and the asset price ratio and risk appetite are not as high as before.

From 1995 to 2000, an average of 250 tech IPOs were financed each year, with a small percentage of profitable companies, all achieving significant gains. In the 1990s, the financing scale of tech IPOs in the US stock market surged. In the three years from 1998 to 2000, the number of IPOs related to the US tech sector reached 744, while in the 5 years after 2000, the total number of IPOs was only 168 Over the past three years, technology-related IPO companies in the United States have raised over $84.1 billion, with a three-year CAGR of 73.69%. The decrease in the number of profitable IPO companies indicates an increase in risk appetite and liquidity abundance. Investors also no longer require profitability from listed companies, with only 14% of US stock IPO companies being profitable between 1999 and 2000.

If a company's suffix is ".com", it will attract a large number of investors regardless of its fundamentals. (Guangfa Computer Group, "The Duality of New Technology Development from the Perspective of Industry Chain ROI")

After 2021, the primary market of US stocks cannot be considered prosperous, and the sentiment on the first day of IPOs is generally average, indirectly confirming the abundance of funds. The US stock IPO market has sharply contracted in the past two years in a high-interest-rate environment. Compared to the IPO boom during the loose monetary policy of the Federal Reserve in 2020-2021, the number and amount of US stock IPOs have sharply declined in the past two years under the backdrop of the Federal Reserve's continuous rate hikes and an overall economic downturn. The IPO first-day return rate can better measure market sentiment and micro liquidity levels, and the current listing returns of US stock IPO companies are also at historically extreme levels.

Looking from another perspective, based on the asset price index of US stocks/US bonds, the current situation has not reached the imbalance seen in 2000.

In 2000, stocks were significantly less attractive compared to bonds (stock-bond yield spread imbalance at +2X, indicating higher stock risk); in 2020, US stocks were very attractive compared to US bonds; currently, the attractiveness of US stocks and bonds is similar, without reaching the extreme imbalance seen in 2000.

Based on the existing information, it is also difficult to make a judgment on whether the United States is at significant "recession risk".

The main support items for the US economy in Q2 are consumption (goods improving, services resilient), manufacturing construction spending, and equipment investment.

In this economic cycle in the United States, the driving force of each sub-item structure on the total is mismatched, not rising and falling together. In the past two years, the core support items of the US economy have experienced the following sequence: government spending (since the second half of 2022) -> construction investment (since the end of 2022) -> goods consumption (since early 2023) -> residential investment (since the second half of 2023) -> service consumption, equipment investment (since the end of 2023) Looking ahead, in the process of economic slowdown, the risk of significant ups and downs (significant recession) may be relatively small. During the economic slowdown, due to the alternating effects of supporting factors, the overall economic performance is relatively stable, reducing the risk of significant recession.

Furthermore, from the perspective of the balance sheet, since 2008, the main leveraged entities in the United States have been government departments, while the balance sheets of residents and enterprises are relatively healthy. Historically, recessions have occurred when the leverage ratios of residents or enterprises have reached high levels.

Finally, in this high interest rate environment, the damage to the profitability of listed companies is not significant. On one hand, this corresponds to the resilience of macroeconomic data, and on the other hand, it comes from the boost of emerging artificial intelligence technologies.

Currently, the ROE levels of the S&P 500 and NASDAQ are at historical highs, while the debt-to-equity ratios are at historical lows.

The change of presidents in the 90s was a clear negative for the tech sector, shifting from the strong support of the Clinton administration to the relatively cautious approach of the Bush administration. In 1993, the Clinton administration issued the "National Information Infrastructure Action Plan," planning to invest $400-500 billion over 20 years.

After Bush took office, the first budget proposal included a significant slowdown in the growth rate of technology funding compared to the past few years, with funding increasing by only 1.4 percentage points, while the average annual growth rate in the past few years was over 6%; in addition, the Department of Commerce's "Advanced Technology Program" was also announced to be terminated.

The Democratic government pays more attention to privacy and security in AI policies, so the Biden administration's attitude in the AI field appears more neutral, and the election results in the future will have an impact on policy coherence, but not to the extent of a significant deterioration. For the Biden administration, although there are relevant policies in place, overall, artificial intelligence is not a core issue of its governance. It remains to be seen whether the Republican Party will adjust its stance on privacy and security during the election process.

In conclusion: From the turning points of the economy, inflation, and liquidity, it is closer to the dot-com era of U.S. stocks around 2000, which may amplify the volatility of current technology leaders.

However, looking ahead, in the process of the U.S. economic slowdown, the risk of significant ups and downs (significant recession) is currently relatively small, and the degree of liquidity flooding and the speed of liquidity are not as high as before, which will influence the subsequent global capital's degree of rebalancing of the total amount of U.S. stocks.

III. Performance of Tech Giants Before and After the Expectation of US Recession

(1) The Information Technology Revolution Gives Birth to Giants of the Era, Creating Stocks with 50x and 100x Returns

General technological innovation will give birth to giants of the era, great companies standing at the peak of the wave, producing stocks with 10x, 50x, and 100x returns.

However, stock prices will fluctuate in this process, and the rhythm is crucial: For example, from 1983 to 1991, Intel's stock price only doubled in nearly 8 years, but in 1992-1993, within 2 years, the stock price nearly tripled, and from 1994 to 2000, the stock price further increased by nearly 20 times.

Transitioning from technology to business, and opening up TO-C usage scenarios, are key points for tech giants to enhance their valuations—

First: Transitioning from technology to products (finding commercialization paths). Microsoft, an operating system developer, and Intel, a processor manufacturer, are in irreplaceable positions. Key technological breakthroughs correspond to the main uptrend of stocks: for example, Intel provided every PC with a Pentium chip, from producing low-performance microprocessors, to the rapid expansion of market share with the development of 32-bit microprocessors in 1986, to the Pentium processor in 1993, establishing technological dominance and widening the gap in stock prices with other companies.

Second: Opening up TO-C usage scenarios. IBM's initial customer base was government departments, the military, banks, and research institutes. Missing out on and lacking experience in operating TO-C consumer products, IBM gradually fell behind in the competition in the 1980s and 90s, leading to losses in 1993 and a decline in stock prices. On the TO-C side, Apple and Microsoft were relatively successful. For example, in 1984, the first Apple Mac computer became an affordable and user-friendly personal computer for the general public.

The tipping point of technology turning into products (Microsoft's Windows 3.0 in the 90s, Intel's Pentium processor in 1993), and the opening of user, especially TO-C user scenarios (Apple's first PC computer for individuals in 1984, Yahoo's creation of a portal website for TO-C users in 1994), are all catalysts for the main uptrend of stock prices in the tech wave.

(2) Strong Performance is the Catalyst for Bubbles, but Leading Performance has Started to Decline in the Late Stage of the Bubble

The weakening of the US economic cycle and the slowdown of the industrial cycle have correspondingly affected the prosperity of tech giants. However, we can see that the slowdown in prosperity does not directly lead to a decline in stock prices. In the late stage of the bubble, the ROE and performance growth rates of tech giants have already begun to decline **

During the period of industrial breakthrough, the leading companies had ROE levels of 30-40%, but the prosperity of tech companies began to decline in the late stage of the bubble. As shown in the chart below, Cisco, Intel, and Microsoft reached their highest ROE levels before entering the "bubble period" - their ROE annualized levels in the third quarter of 1997 reached 44% and 39% respectively, but then began to decline. It wasn't until the third quarter of 1998 that the ROE of these three companies started to rise again. However, the good times didn't last long, as the ROE of these companies started to decline again in the second half of 1999, with a downward shift in the center.

In the 90s, the compound annual profit growth rate of leading tech companies could reach 30-50%, but it also began to decline in the late stage of the bubble. Dell, Microsoft, and Intel all experienced varying degrees of profit decline; Ericsson and HP saw negative profit growth starting in 1998.

(III) Typical companies' valuation breakthroughs from performance-driven to valuation-driven

In 1998-1999, as the profits of tech companies began to decline, their stock prices soared significantly, leading to a substantial increase in their valuation levels, far exceeding the previous fluctuation center, entering a stage of valuation bubble.

Before 1999, Intel's PE fluctuated between 8 times and 20 times, but after the second half of 1999, it broke through 20 times PE, reaching a peak of 55 times; Microsoft's PE fluctuated between 17 times and 30 times before 1999, but after the second half of 1999, it also broke through the center, reaching over 80 times at its peak; Cisco's valuation peak exceeded 200 times.

(IV) Using Japan as an example, the overseas reflection of the tech bubble

Regarding the performance of the Japanese tech industry throughout the entire tech bubble process, we have detailed it in last year's report "Japan's Interpretation of Barbell Strategy: 90s Japanese Tech".

In the 90s, influenced by global tech innovation, the development trend of Japan's manufacturing industry shifted towards the TMT industry, marking a vibrant industrial trend during the economic downturn.

During this process, Japan's relatively industrially advantaged hardware sector experienced rapid development (electronic components, semiconductors). Good companies emerged (with high compound annual growth rates), and they were also good stocks (the upper part of the table shows stock prices, most of which saw significant increases in various years)

However, Japan's software and internet industry lacks competitiveness in nurturing, with companies being valuation-driven without performance support. As the bubble of the science and technology network bursts, the stock price performance is short-lived.

The strategic transformation of Japan's software industry is slow, still stuck in the era of large computers, mainly focusing on software outsourcing and IT services, failing to catch up with the software innovation wave of the PC era. Japanese companies use their hardware development experience to develop software, pursuing quality, stability, and zero defects, but their corporate strategies are not well-suited.

Under overseas mapping, Japanese computer, software, and internet companies can still rise significantly, mainly being driven by overseas mapping to boost valuation, but with negative performance contribution. The market is mainly driven by valuation. By the peak of the science and technology network bubble, the PE valuation of software services reached 192X at the end of 1999 and 332X in the first quarter of 2020.

IV. A Comparison between the Current AI "Shovel Sellers" and the 90s Science and Technology Giants

(1) The completeness and coverage of the AI industry chain are not as good as the 90s science and technology network, with a narrower investment scope.

The complete development context of the previous computer wave: semiconductors, personal computers, hardware and operating systems, revolutionary software with radiation, the internet. The internet began to deeply integrate with traditional industries, giving rise to new models like "internet + retail" and "internet + payment," a revolutionary scenario from scratch.

Currently, AI is in a period of infrastructure laying and explosive hardware demand, yet it has not achieved a revolutionary application end. There are expectations for large models and applications in 24 years, but it has not reached the stage of commercial landing.

(2) The market value concentration of the "Seven Sisters" in the US stock market is higher, with the first wave of increase slope surpassing the science and technology network of the 90s.

Due to the lack of radiation and diffusion of targets, the market value proportion of the seven sisters in the US stock market is higher than that of the 90s science and technology giants. Before the 2000 science and technology network bubble, the top ten technology companies with the highest market value on the Nasdaq accounted for about 22% of the US GDP; currently, the market value of the seven sisters in the US accounts for about 51% of the US GDP.

The proportion of the current seven sisters in the total market value of the top 3000 stocks in the US stock market is also higher than the peak of the bubble in 2000.

If epoch-making explosive products are taken as the starting point, it can be seen that compared to the science and technology giants of the 90s, NVIDIA has a steeper first wave of increase slope. Taking the launch of revolutionary products as a reference point, NVIDIA's current stock price increase slope has surpassed Intel in 95 and Microsoft in 98

(3) AI Performance Explosive, High Valuation Implies Expectations of High Growth in the Coming Years

The fierce rise of leading AI stocks in the US stock market is directly related to the explosive performance and optimistic guidance expectations.

Compared to K-Net, this round of explosive performance in the AI "shovel-selling" industry is stronger, but compared to K-Net's compound high-speed growth of over 5 years, the sustainability of this round is currently unknown. The performance of A-share and US stock core targets in the AI industry has already shown explosive growth, with ROE also at high levels (many exceeding 30%), but whether there will be sustained performance explosion depends on breakthroughs in the supply side bottlenecks, as well as incremental demand brought by emerging applications and new business models.

If benchmarked against hardware companies Intel/Cisco, the valuation pulse of this round of NVIDIA is higher, and the current valuation level is relatively high (even considering dynamic valuation metrics).

In the first half of the K-Net bubble, despite the introduction of revolutionary products, the valuation levels of hardware companies remained stable at around 30-40 times, while NVIDIA's valuation in this round has indeed risen more steeply.

One possible explanation is that due to the narrow scope of investable targets in the current AI industry, the convergence effect of global funds will push up valuations and exacerbate valuation fluctuations.

Considering the related companies in the A-share AI industry chain, the nearly 2-year rise in stock prices can be divided into valuation vs. performance, with currently most companies driven by both performance/valuation, with some companies being more valuation-driven.

As shown in the chart below, during the K-Net period (blue dots), valuation contributed more, while in the current AI period (yellow dots), performance/valuation contributions are basically equal.

(4) K-Net leaders also experienced 20-30% retracements in the past, the key is whether the subsequent performance guidance can continue

Since the outbreak of the AI industry in 23, NVIDIA's stock price has experienced 4 retracements of more than 10 points, with the retracement in August 24 exceeding 25%, marking the longest adjustment period, causing market concerns.

In fact, K-Net leaders such as Microsoft, Intel, often experienced 20-30% stock price retracements during the main uptrend in the 90s, and high fluctuations in the stock prices of tech leaders are not uncommon. The duration of these retracements is 2-5 months, with retracement levels generally within 30% However, it is worth mentioning that the corrections of the leading stocks in the science and internet sector between 1992 and 2008 were not due to the falsification of industry trends or problems in company operations, so after high volatility, they were quickly able to continue to reach new highs.

During this period, companies like Cisco, Intel, and Dell in the hardware and equipment sector did not experience performance guidance misses due to hardware shortages, which may be a variable to observe in the subsequent rounds.

Analysts:

Liu Chenming: SAC License No.: S0260524020001

Zheng Kai: SAC License No.: S0260515090004