Cathie Wood's 2026 Macro & Technology Investment Roadmap

Author: Cathie Wood

Compiled by: Bilibili News

This article is not intended as investment advice. Readers are advised to strictly abide by local laws and regulations.

Key takeaway: In her 2026 New Year's outlook, ARK founder Cathie Wood points out that technologies such as AI, robotics, and blockchain are driving record-high capital expenditures, while inflation is declining and productivity is improving, potentially boosting GDP growth in the long term. The article also analyzes the trends and market valuations of gold, Bitcoin, and the US dollar, providing investors with a combined macroeconomic and technological perspective.

Happy New Year to ARK's investors and all supporters!

We sincerely thank you for your continued support.

As I demonstrated in this letter, we truly believe that investors have every reason to remain optimistic! I hope you enjoy the following discussion. From an economic history perspective, we are at a crucial juncture.

A compressed spring.

Despite the continued growth of the U.S. real gross domestic product (GDP) over the past three years, the internal structure of the U.S. economy has undergone a "rolling recession," gradually evolving into a compressed spring that may rebound strongly in the coming years.

In response to the supply shocks related to COVID, the Federal Reserve raised the federal funds rate from 0.25% to 5.5% over a 16-month period from March 2022 to July 2023, an increase of 22 times.

This unprecedented austerity has pushed housing, manufacturing, non-AI capital expenditures, and low- and middle-income groups into recession, as shown in the figure below.

Taking existing home sales as an example, the U.S. housing market fell from an annualized rate of 5.9 million units in January 2021 to 3.5 million units in October 2023, a drop of 40%.

This level was last seen in November 2010, and it has remained near this low level for the past two years.

What further illustrates how tightly the spring is compressed is that current existing home sales are comparable to those in the early 1980s, when the U.S. population was about 35% smaller than it is now.

According to the US Purchasing Managers' Index (PMI), the manufacturing sector has been in contraction for approximately three consecutive years. In this diffusion index, 50 is the dividing line between expansion and contraction, as shown in the chart below.

Meanwhile, capital expenditures, measured by “non-defense capital goods (excluding aircraft)”, peaked in mid-2022, then declined, and have now just returned to that level, regardless of whether technology spending is included.

In fact, this capital expenditure indicator had been difficult to break through for more than 20 years since the bursting of the technology and telecommunications bubble in the 1990s, until 2021, when the supply shock caused by COVID forced digital and physical investment to rise in tandem.

What was once the "ceiling" is now becoming the "floor," as AI, robotics, energy storage, blockchain technology, and multi-omics sequencing platforms have all entered a stage where they can be applied on a large scale.

Following the tech and telecommunications bubble of the 1990s, capital expenditures experienced a 20-year period of repeated peaks around the level of approximately $70 billion.

Now, this phase is giving way to what could be the strongest capital expenditure cycle in history, as shown in the diagram below.

In our view, the AI bubble is still many years away.

According to data from the University of Michigan, consumer confidence among low- and middle-income groups has fallen to levels seen in the early 1980s, when double-digit inflation and interest rates severely eroded purchasing power and dragged the U.S. economy into a prolonged recession.

More notably, consumer confidence has also declined significantly in recent months. We believe that consumer confidence is one of the most tightly compressed springs, and most likely to rebound.

Deregulation and low taxes, low inflation, and low interest rates

Thanks to the combined effects of deregulation and tax cuts (including tariffs), declining inflation, and lower interest rates, the rolling recession in the United States over the past few years may be reversed rapidly and dramatically in the coming year or longer.

Deregulation is unleashing innovation across industries, with AI and digital assets leading the way, spearheaded by David Sacks, the first "Head of AI and Cryptocurrency Affairs".

At the same time, tax cuts on tips, overtime pay, and Social Security are expected to generate substantial tax refunds for U.S. consumers this quarter, boosting real disposable income growth from an annualized rate of about 2% in the second half of 2025 to about 8.3% this quarter.

On the corporate side, accelerated depreciation policies for manufacturing facilities, equipment, software, and domestic R&D spending will reduce the effective corporate tax rate to close to 10%—one of the lowest in the world.

For example, any company that starts construction of a manufacturing facility in the United States before the end of 2028 can realize full depreciation in the first year the building is put into use, instead of having to amortize it over 30 to 40 years as in the past.

Equipment, software, and domestic R&D expenditures can also be depreciated at 100% in the first year. This significant cash flow benefit was made permanent in last year's budget and is retroactive to January 1, 2025.

After stubbornly remaining in the 2%–3% range for the past few years, inflation as measured by the CPI may fall to unexpectedly low levels in the coming years, and may even turn negative.

Over the past few years, inflation, as measured by the Consumer Price Index (CPI), has stubbornly hovered in the 2%–3% range;

However, in the coming years, inflation is likely to fall to unexpectedly low levels, and may even turn negative. There are several reasons for this change, as shown in the diagram below.

First, since its post-pandemic high of around $124 per barrel on March 8, 2022, West Texas Intermediate (WTI) crude oil prices have fallen by about 53%, and are currently down about 22% year-over-year.

Secondly, since peaking in October 2022, the sales price of newly built detached houses has fallen by about 15%;

Meanwhile, the inflation rate of existing single-family home prices (calculated on a three-month moving average) has fallen sharply from a year-on-year peak of about 24% in June 2021 after the pandemic to about 1.3% currently, as shown in the figure below.

In addition, in the fourth quarter, efforts were made to reduce the inventory of nearly 500,000 newly built detached houses, a level last seen in October 2007, just before the global financial crisis.

Three major residential developers have significantly reduced prices compared to the same period last year:

Lennar saw a 10% year-on-year decrease, KB Homes a 7% decrease, and DR Horton a 3% decrease. These price reductions will gradually be transmitted and reflected in the CPI over the next few years with a lag.

Finally, nonfarm productivity, one of the most powerful hedges against inflation, remained strong amid a rolling recession, growing 1.9% year-over-year in the third quarter.

Compared to a 3.2% increase in hourly wages, productivity gains have suppressed unit labor cost inflation to 1.2%, as shown in the chart below. This figure shows no trace of the "cost-push inflation" seen in the 1970s.

Further evidence of this improving trend is that the inflation rate, as measured by the Truflation indicator, has recently fallen to 1.7% year-on-year, as shown in the chart below, which is nearly 100 basis points (bps) lower than the official inflation data based on CPI from the U.S. Bureau of Labor Statistics (BLS).

Productivity Boom

If our research findings on technology-driven disruptive innovation hold true, non-agricultural productivity growth could accelerate to 4%–6% in the coming years, thereby further depressing unit labor cost inflation.

The integration of AI, robotics, energy storage, public blockchain, and multi-omics technologies will not only drive productivity to a new level but also create enormous wealth.

Increased productivity may also alleviate significant geoeconomic imbalances in the global economy. Companies can allocate the productivity dividend along one or more of the following four strategic directions:

Increase profit margins, increase R&D and other investments, raise employee compensation, and lower product prices.

In China, higher productivity, which corresponds to higher wages and higher profit margins, helps the economic structure shift from excessive investment, which has long accounted for about 40% of GDP—almost twice that of the United States—to a more balanced development path (as shown in the figure below).

This investment ratio has remained high since China joined the World Trade Organization (WTO) in 2001.

Increasing workers' income will drive China's economy to shift from investment and "involutionary competition" to consumption-driven growth, which is in line with President Xi Jinping's goal of "anti-involution."

At the same time, American companies may further enhance their competitiveness relative to Chinese companies by increasing investment and/or lowering prices.

However, it should be noted that in the short term, the productivity gains enabled by technology may continue to slow job growth in the United States, causing the unemployment rate to rise from 4.4% to 5.0% or even higher, thereby prompting the Federal Reserve to continue cutting interest rates.

Following this, deregulation and other fiscal stimulus measures are expected to amplify the effect of low interest rates, significantly accelerating GDP growth in the second half of 2026.

At the same time, inflation is likely to continue to slow. This is not only because of lower oil prices, housing prices, and tariffs, but also because it is these same technological forces that are driving productivity growth and lowering unit labor costs.

Surprisingly, according to some benchmarks, AI training costs are decreasing at a rate of about 75% per year, while AI inference costs (i.e. the cost of running AI application models) are decreasing by as much as 99% per year.

This unprecedented decline in technology costs will drive an explosive growth in the quantity of related products and services.

Therefore, we are not surprised that the nominal GDP growth rate of the United States may remain in the 6%–8% range in the next few years, driven by factors including:

Productivity growth of 5%–7%, labor force growth of about 1%, and inflation of -2% to +1%.

The deflationary effects brought about by AI and the other four major innovation platforms will continue to accumulate, creating an economic context similar to the last major technological revolution that occurred in the 50 years leading up to 1929, driven by the internal combustion engine, electricity, and telephone communication.

During that period, short-term interest rates generally moved in tandem with nominal GDP growth, while long-term interest rates responded to the deflationary undercurrents that accompanied the technological boom. As a result, the yield curve was inverted by an average of about 100 basis points, as shown in the figure below.

Other New Year's Thoughts

Gold prices rise vs. Bitcoin prices fall

In 2025, the price of gold rose by 65%, while the price of Bitcoin fell by 6%. Many observers attribute the rise in the price of gold from $1,600 to $4,300 (a cumulative increase of 166%) since the end of the US stock market bear market in October 2022 to concerns about inflation risks.

However, another interpretation is that the rate of global wealth creation, such as the 93% rise in the MSCI World Equity Index, has exceeded the annualized growth rate of approximately 1.8% in global gold supply.

In other words, new demand for gold may be outpacing its supply growth. Interestingly, during the same period, while Bitcoin's supply grew by only about 1.3% annualized, its price surged by 360%.

In this comparison, a key difference lies in how gold miners and Bitcoin "miners" react to price signals in completely different ways.

Gold miners can respond to rising prices by increasing production, while Bitcoin cannot do so.

Bitcoin's supply growth is strictly limited by mathematical rules: its annual growth rate will be approximately 0.82% over the next two years, and will then slow further to approximately 0.41%.

The Gold Price in Perspective

Measured by the ratio of gold market value to M2 money supply, this ratio has only been higher than it is now in one period over the past 125 years: the Great Depression in the early 1930s, when the price of gold was fixed at $20.67 per ounce, while the M2 money supply plummeted by about 30%, as shown in the figure below.

Recently, the gold-M2 ratio has broken through its previous historical high, which occurred in 1980 when inflation and interest rates both soared to double digits. In other words, from a historical perspective, the current gold price is at an extreme level.

It is also worth noting that, as can be seen from the chart below, this ratio tends to correspond to good returns in the stock market during long-term downward phases.

Since 1926, according to research by Ibbotson and Sinquefield, the long-term compound annual return on stocks has been approximately 10%.

After the ratio hit two long-term highs in 1934 and 1980, stock prices, as measured by the Dow Jones Industrial Average (DJIA), rose by 670% and 1,015% in the 35 years leading up to 1969 and the 21 years leading up to 2001, respectively, corresponding to annualized returns of 6% and 12%.

It is worth noting that small-cap stocks achieved annualized returns of 12% and 13% in these two phases, respectively.

Another crucial factor for asset allocators is that since 2020, Bitcoin's returns have shown very low correlation with gold and other major asset classes, as illustrated in the table below.

What's even more interesting is that the correlation between Bitcoin and gold is even lower than the correlation between the S&P 500 and bonds.

In other words, in the coming years, Bitcoin is expected to become an important diversification tool for asset allocators to improve "return per unit of risk".

The Dollar's Outlook

In recent years, a widely circulated narrative has been that "American exceptionalism" is coming to an end. Representative evidence for this view includes:

The dollar's decline in the first half of a year was the largest since 1973; on a full-year basis, it was the most significant annual drop since 2017.

Last year, measured by the trade-weighted dollar index (DXY), the dollar fell 11% in the first half of the year and 9% for the whole year.

If our assessments of fiscal policy, monetary policy, deregulation, and US-led technological breakthroughs are correct, then the US rate of return on invested capital will rise relative to the rest of the world, thereby driving the dollar higher.

The Trump administration’s policies are echoing those of the early Reaganomics of the 1980s, when the dollar nearly doubled, as shown in the chart below.

AI Hype

The AI wave is driving capital spending to levels not seen since the late 1990s, as shown in the chart below.

By 2025, investment in data center systems (computing, networking, and storage devices) will grow by 47%, approaching $500 billion.

It is projected to grow by another 20% to reach approximately $600 billion by 2026, far exceeding the long-term trend of $150 billion to $200 billion annually in the decade before ChatGPT's emergence.

Such a huge investment naturally raises a key question: where will the returns on these investments come from, and who will receive them?

Beyond semiconductor and large cloud computing companies in the public market, unlisted AI-native companies are becoming significant beneficiaries of this growth and investment returns. AI companies are among the fastest-growing in history.

According to our research, consumers are adopting AI at twice the rate they did when they adopted the internet in the 1990s, as shown in the figure below.

By the end of 2025, OpenAI and Anthropic reportedly had annualized revenue run rates of $20 billion and $9 billion, respectively, representing a 12.5-fold and 90-fold increase from $1.6 billion and $100 million in just one year.

Market rumors suggest that both companies are considering IPOs within the next one to two years to fund the massive investments needed to support their product models.

As Fidji Simo, CEO of OpenAI's Applications division, stated:

"The capabilities of AI models far exceed what most people experience in their daily lives, and the key in 2026 is to close that gap. The leaders in the AI field will be those companies that can translate cutting-edge research into products that are truly useful to individuals, businesses, and developers."

This year, substantial progress is expected in the user experience aspect, making it more intentional, intuitive, and highly integrated.

An early example is ChatGPT Health—a dedicated section within ChatGPT designed to help users manage their health and medical care based on their personal health data.

On the enterprise side, many AI projects are still in their early stages, constrained by bureaucratic processes, inertia, and the reality that organizational restructuring and data infrastructure must be carried out before AI can truly realize its value.

By 2026, companies will likely realize that they must train models based on their own data and iterate rapidly, or they will be left behind by more aggressive competitors.

AI-driven applications will bring instant and superior customer service, faster product launches, and startups that can "do more with less."

High valuation of the market.

Many investors are concerned about stock market valuations, which are currently at the high end of their historical range, as shown in the chart below.

Our own valuation assumption is that the price-to-earnings ratio (P/E) will fall back to the average level of the past 35 years—about 20 times.

Some of the most significant bull markets have evolved precisely during periods of exponential contraction. For example:

From mid-October 1993 to mid-November 1997, the S&P 500 index had an annualized return of 21%, while the P/E ratio dropped from 36 to 10 during the same period.

From July 2002 to October 2007, the S&P 500 index had an annualized return of 14%, while the P/E ratio contracted from 21 to 17 during the same period.

Given our forecast of accelerated real GDP growth driven by productivity and slowing inflation, this dynamic is likely to reappear in this market cycle—and may even be more pronounced.

My heartfelt thanks to the investors and other friends who have supported ARK as always, and also to Dan, Will, Katie and Keith for helping me complete this lengthy New Year's address!