If you pay attention to global technology investment, it is almost impossible to avoid one name - Cathie Wood, who is more familiar to Chinese investors as "Sister Wood".

For the past ten years, she and ARK Invest, the company she founded, have been doing something that is not easy on Wall Street: ignoring short-term noise and betting on long-term, extreme, non-linear technological changes.

ARK's annual research report "Big Ideas" has been published for ten consecutive years. It is not a simple industry outlook, but more like a "technology map for the next ten years."

You can disagree with its conclusions, but it's hard to ignore the questions it raises.

This year's "ARK Big Ideas 2026" has a very eye-catching overall title: The Great Acceleration.

This report focuses on 13 major innovation areas. The core conclusion is that the five major innovation platforms centered on artificial intelligence are accelerating their integration and will trigger a step change in global economic growth by the end of this decade. The actual GDP growth rate in 2030 is expected to reach 7.3%, which is 4 percentage points higher than the 3.1% predicted by the International Monetary Fund.

The most important judgment made by the report is that AI is not another important technological advancement, but a "Central Dynamo" that is driving the acceleration of multiple technology curves at the same time. In the past few decades, technological innovation has mostly presented a linear structure: a technology → an industry → a round of capital cycle. ARK believes that this paradigm has expired. At the current stage, technologies are no longer in a parallel relationship, but are highly coupled and unlocked from each other:

AI’s computing power needs drive the next generation cloud, energy storage and data center revolution; blockchain and digital wallets provide a trusted settlement and execution layer for AI Agents; robots and autonomous driving push AI from the “digital world” into the “physical world”; multi-omics and programmable biology provide AI with high-dimensional life data and reversely accelerate model capabilities.

ARK uses an indicator to describe this state: Technical Convergence Network Strength. By 2025, this indicator will have increased by 35% year-on-year - meaning that the mutual catalysis between different technologies is significantly accelerating. This is why ARK calls 2026: The Great Acceleration.

ARK research shows that reusable rockets launch AI chips into orbit, multi-omics data drives the development of precision therapies, and smart contracts support AI agents to coordinate real-world resources—these seemingly independent innovations are forming unprecedented synergies. The importance of robotics as a catalyst reaches an inflection point in 2025, and energy storage and distributed energy systems have become key drivers of the construction of next-generation cloud infrastructure.

The report stated that the direct impact of this technological revolution is:

The market share of innovative assets will grow from approximately 20% in 2025 to approximately 50% in 2030, and the market value may expand from the current approximately US$5 trillion to approximately US$28 trillion.

Data center system investment is expected to grow from approximately US$500 billion in 2025 to approximately US$1.4 trillion in 2030, with a compound annual growth rate of 30%.

The commercialization process in fields such as self-driving taxis, AI drug development, and household humanoid robots is accelerating, and some fields have entered the stage of large-scale deployment.

However, ARK also made it clear that not all compelling technologies are disruptive. The report takes quantum computing as an example and believes that even at the most radical development rate, the practicality of this technology in cryptographic decryption will not be realized until the 2040s. A truly disruptive technology must meet conditions such as drastic cost reductions, unlock compelling unit economics across multiple industries, and serve as a platform for other technological innovations.

AI leads the era of "great acceleration"

According to the report, ARK named this round of technological change "The Great Acceleration" and believed that the interdependence of the five major innovation platforms of AI, public blockchain, robotics, energy storage and multi-omics is increasing, and the performance improvement of one platform will unlock new capabilities of another platform.

The most striking case in the report is the combination of reusable rockets and AI computing power. Neural networks’ need for next-generation cloud computing power is running into terrestrial scaling limitations, and reusable rockets could be the solution.

At a competitive cost, space-based AI computing power can provide the cloud with computing power that is not limited by ground power and cooling.

ARK's analysis shows that the growth of AI chips may increase the demand for reusable rockets by about 60 times relative to existing models. Based on expected launch costs, space-based computing costs may be 25% lower than ground-based computing.

This convergence of technologies is reportedly giving rise to an unprecedented investment cycle. ARK research suggests capital investment alone could contribute 1.9 percentage points to annualized real GDP growth this decade. The new capital base - self-driving taxis, next-generation data centers and corporate investment in AI agents - should boost return on invested capital. As other innovations begin to impact the growth trajectory, actual growth achieved could exceed consensus expectations by more than 4 percentage points annually.

From a historical perspective, technological paradigm shifts have repeatedly triggered structural changes in GDP growth rates. ARK data shows that the global real GDP growth rate has gradually increased from 0.037% in 100,000 BC to about 3% currently through the agricultural revolution, industrial revolution and other stages. This round of technological revolution with AI as its core may push this growth rate to more than 7%.

AI infrastructure investment surges

The growth rate of data center system investments is accelerating. Since the launch of ChatGPT, the annualized growth rate of this type of investment has jumped to 29% from the previous 5%.

In 2025, global data center system investment will reach approximately US$500 billion, which is nearly 2.5 times the average level from 2012 to 2023. ARK predicts that this investment scale may grow to approximately US$1.4 trillion by 2030.

The core factor driving the surge in investment is the explosive growth in demand for AI. The cost of inference has dropped by more than 99% in the past year, prompting exponential growth in the use of AI by developers, enterprises and consumers. Taking the OpenRouter platform as an example, since December 2024, the computing demand for large language models has increased by approximately 25 times.

However, compared with the dot-com bubble period, the current valuation of the technology industry is much more rational. While capital spending in the information technology and communications services industry as a share of GDP has reached its highest level since 1998, the technology sector's price-to-earnings ratio is well below its peak during the dot-com bubble.

The average price-to-earnings ratio of six companies, including Nvidia, Google parent Alphabet, Apple, Amazon, Meta and Microsoft, is about a fraction of its historical highs, indicating that the current investment boom is more based on actual application demand rather than speculative bubbles.

The competitive landscape is also changing. Nvidia's early investments in AI chip design, software and networks have enabled its GPU sales share to reach 85% and its gross profit margin to be as high as 75%. But rivals such as AMD and Google have caught up in some areas, particularly in small language model inference.

ARK data shows that AMD's MI355X can handle approximately 38 million tokens per TCO (total cost of ownership) dollar in small model performance, surpassing Nvidia's B200. However, Nvidia's Grace Blackwell rack-mounted system remains the leader in large model inference, powering the most advanced base models.

AI consumer operating system reshapes business model

AI models are converging into new consumer operating systems, fundamentally changing the way people interact with the digital world. Consumer adoption of AI far exceeds the penetration rate of the Internet back then - the penetration rate of AI chatbots among smartphone users reached about 25% within 7 years of its launch, while it took longer for the Internet to reach the same penetration rate among PC users.

This shift is compressing the shopping funnel. It took about an hour to complete a purchase from the pre-Internet era, to a few minutes in the mobile era, and further compressed to about 90 seconds in the AI ​​agent era. AI shopping agents are changing the purchase funnel with unprecedented personalization and speed. Today, 95% of the consumer journey occurs before purchase. Personalization is no longer an option, but a moat.

Underpinning this shift are new protocol standards. Anthropic’s open-source Model Context Protocol (MCP) enables agents to seamlessly access real-time information from across the internet, while OpenAI’s Agent Commerce Protocol (ACP) secures end-to-end transactions. These protocols are simplifying and driving transactions in the AI ​​era.

The scale of the market opportunity is staggering. ARK predicts that global online consumer spending facilitated by AI agents will grow from approximately 2% of online sales in 2025 to approximately 25% in 2030, and may exceed US$8 trillion by then.

AI search traffic share is expected to grow from 10% in 2025 to 65% in 2030, and the annual growth rate of AI-related search advertising expenditures is approximately 50%.

AI agents may generate approximately $900 billion in commercial and advertising revenue by 2030, with the dominant growth factors being lead generation and advertising, far exceeding the contribution of consumer subscription revenue.

Robots: The grossly underestimated GDP engine

If AI is the main engine of the digital world, then robots are its most important "physical export".

The report emphasizes that the rapid progress of AI is transforming robots from specialized equipment for fixed tasks into relatively open and general-purpose platforms, which is the key to unlocking the potential of the industrial and home markets.

ARK estimates that the global robot market has a revenue opportunity of approximately US$26 trillion, divided into two major sectors: manufacturing and home services.

In the manufacturing field, global manufacturing GDP is expected to reach US$32 trillion by 2030. If robotics technology can achieve 100% improvement in labor productivity, it can create approximately US$13 trillion in revenue opportunities based on a 35% share rate for service providers.

In the field of household services, approximately 2.8 billion workers around the world engage in 2.3 hours of unpaid housework every day. Based on the global average hourly wage of US$12 and a time value of 50%, this also corresponds to a market space of approximately US$13 trillion.

ARK particularly emphasizes the macro significance of humanoid robots.

An easily overlooked fact is that today, a large amount of household maintenance, care, cleaning, and management labor are not included in GDP.

ARK’s calculations show that a single household humanoid robot can convert about $62,000 of hidden labor into explicit GDP per year; if 80% of U.S. households adopt it within 5 years, the annual GDP growth rate may jump from 2–3% to 5–6%

The report believes that this is not a story of "replacement of work", but of converting non-market activities into market activities and releasing time into productivity.

Autonomous driving reaches an inflection point

ARK judges that humanoid robots are about 200,000 times more complex than self-driving cars. This complexity ratio defines the theoretical capabilities required to achieve full autonomy. Nonetheless, by mapping the relationship between the amount of computing required for Tesla's Fully Autonomous Driving (FSD) and performance improvement, ARK predicts that under the conditions of continued AI computing power expansion and hardware advancement, the Optimus humanoid robot may reach human-level task execution capabilities around 2028.

Self-driving taxis are beginning to encroach on the online ride-hailing market share. In the San Francisco operating area, Waymo's market share has put pressure on Uber and Lyft. The cumulative autonomous driving mileage of Waymo, Baidu's Apollo Go, Pony.ai and other companies has reached billions of miles, and the daily autonomous driving mileage is growing rapidly.

Cost reduction will be key to driving demand. ARK predicts that by 2035, the global price per mile of self-driving taxis may drop to US$0.25, which is far lower than the US$2.80 for human-driven online ride-hailing and US$0.80 for private cars in the United States in 2025. In the early commercialization phase, vehicle cost will dominate unit economics, while at scale vehicle utilization will drive cost per mile down.

The market value potential is huge. ARK estimates that self-driving taxis may create approximately $34 trillion in enterprise value by 2030, of which self-driving technology providers will capture approximately 98% of EBIT (earnings before interest and tax) and enterprise value, with automakers and fleet operators accounting for a relatively small share. The main risk to this forecast is whether automakers other than Tesla can expand their self-driving taxi fleets quickly enough.

Autonomous driving logistics also has broad prospects. Fully automated last-mile deliveries—whether via drones or ground-based robots—are already carried out more than 4 million times a year globally. Driverless long-haul trucking has begun in the United States, and operators are planning to rapidly expand routes. ARK predicts that global autonomous delivery revenue may reach $480 billion by 2030, with automation of supervision and back-end loading operations being important limiting factors.

Multi-omics and AI-driven biological breakthroughs

The combination of Multiomics - covering genomics, epigenomics, transcriptomics, proteomics and metabolomics - and AI is creating a flywheel effect in biological innovation. This flywheel includes: generating richer, lower-cost biological data, conducting more accurate tests, generating better biological insights, developing AI-driven medicines, and ultimately achieving cures.

Data generation costs are falling dramatically. The cost of whole-genome sequencing may drop to $10 in 2030, about 10 times lower than in 2015.

This will drive a surge in sequencing demand. The number of next-generation molecular diagnostic tests is expected to grow from less than a million in 2020 to about 7 million in 2030. The amount of token data generated each year may reach about 200 billion, exceeding the scale of 150 trillion tokens used to train cutting-edge language models such as OpenAI, Gemini, Anthropic and xAI.

AI-empowered diagnostic capabilities are reaching an inflection point. Following the launch of ChatGPT, success rates for FDA-approved AI-driven tests and devices hit an inflection point from single-digit percentage levels. ARK's best-fit model shows that the share of AI-driven diagnostics and equipment may expand to about 30% by 2030, and eventually reach close to 100%.

The economics of drug development are being reshaped. AI-driven drug development could reduce time to market by about 40%, from 13 years to eight years, while reducing total drug costs by about four times, from $2.4 billion to $700 million. Combining the two factors of AI acceleration and disease cure, the value of AI-designed drugs in clinical phase I may exceed US$2 billion, while traditional drug assets often only recover capital costs.

The market potential of biological cures is particularly amazing. ARK research shows that the average price to cure a rare disease may currently exceed $1 million, nearly 15 times the lifetime prescription cost to manage the disease. A cure drug that can capture revenue from a large portion of the patient population before patent expiration could be 20 times more valuable than a typical drug and 2.4 times more valuable than a prescription drug to treat a chronic disease.

A more macro perspective is the extension of healthy lifespan. If the U.S. population could live to the theoretical maximum lifespan of 120 years in perfect health but with the risk of accidental death still present, this would result in a gain of 11.9 billion quality-adjusted life years (QALYs). At a valuation of US$100,000 per healthy life year, the potential life gain market opportunity is approximately US$1.2 trillion. The current global biotechnology market only accounts for approximately 0.1% of this potential market.

Reusable rockets open up space economy

SpaceX’s reusable rocket technology is pushing the economy into the space age. In 2025, the mass put into orbit each year reaches an all-time high, with SpaceX dominating. The company has more than 9,000 active Starlink satellites, accounting for approximately 66% of all active satellites in Earth orbit.

Launch costs continue to fall. According to Wright's Law, every time the cumulative launch mass doubles, the launch cost should decrease by about 17%. In the 17 years since 2008, taking advantage of Falcon 9's partial reusability, SpaceX has cut costs by about 95%, from about $15,600 per kilogram to less than $1,000. ARK research shows Starship can continue this trajectory to under $100 per kilogram at scale.

Satellite bandwidth costs are also falling. According to Wright's Law, satellite bandwidth costs should fall by approximately 44% per doubling of cumulative orbital gigabits per second (Gbps), allowing satellite connectivity to supplement cell towers and provide ubiquitous mobile coverage throughout the United States.

Comparison shows that in 2001, the monthly mobile connection fee for American consumers was about US$90 (2025 US dollars), which only included 0.001GB of data, covering about 1% of the US land area; in 2025, the monthly fee was about US$100, providing unlimited high-speed Internet, covering about 86% of the land area; by 2030, it is expected to achieve 100% coverage at the same price.

The size of the market opportunity is considerable. Thanks to cost reductions and performance improvements, scaled satellite connectivity could generate more than $160 billion in annual revenue, accounting for approximately 15% of ARK's global communications revenue forecast. This forecast is based on the relationship between constellation bandwidth capacity and revenue opportunities, showing exponential growth potential.

Distributed energy supports AI computing power needs

Energy is becoming more and more efficient in driving economic growth. While there were concerns about energy intensity during the dot-com boom, economies have actually become more energy-efficient, and the same dynamics may reappear in the AI ​​era. The energy intensity (kWh per dollar of GDP) of major economies such as China, the United States, Japan, India, and Germany has continued to decline over the past three decades.

The cost of multi-omics data has plummeted. Solar and battery costs continue to fall following Wright's Law, and nuclear cost declines were interrupted by regulatory changes in the 1970s, but recent U.S. executive orders should push nuclear energy back on its previous cost trajectory. Historically, solar and nuclear energy costs (in megawatts) and battery costs (in megawatt-hours) have fallen significantly each time cumulative capacity doubled.

Electricity prices are expected to resume their downward trend. In line with Wright's Law, ARK research shows that, with the exception of World War II, U.S. electricity prices declined steadily from the late 19th century to 1974, before being interrupted by rising nuclear energy construction costs due to increased regulation. Without increased regulation, ARK research suggests today's electricity prices could be around 40% lower than they actually are. As low-cost generation scales up to serve power-hungry AI data centers, retail electricity prices should begin to fall again after 50 years of stagnation.

The demand for investment is huge. Given ARK's rapid GDP growth forecast, cumulative global capital expenditure in power generation must expand approximately 2-fold to approximately $10 trillion by 2030 to meet global electricity demand. Therefore, stationary energy storage deployment needs to be expanded by about 19 times. Data centers are expected to account for approximately 5% of total power generation investment between 2026 and 2030.

The digital asset market shows an evolutionary trend

Stablecoin activity will see significant growth in 2025 due to the regulatory framework that may be brought about by the GENIUS Act. Some companies and institutions have announced stablecoin-related plans, and BlackRock disclosed that it is preparing an internal tokenization platform. Stablecoin issuers and financial technology companies such as Tether, Circle and Stripe have launched or supported Layer 1 blockchains optimized for stablecoins.

Data shows that the market value of tokenized real-world assets (RWA) will grow by approximately 208% in 2025, reaching approximately $18.9 billion. BlackRock’s BUIDL money market fund has a size of approximately $1.7 billion and is said to account for 20% of the approximately $9 billion U.S. Treasury tokenization market. Tether’s XAUT and Paxos’ PAXG have a tokenized commodity market size of approximately US$1.8 billion and US$1.6 billion respectively.

ARK predicts that the scale of tokenized assets may grow from US$19 billion to approximately US$11 trillion by 2030, but there is considerable uncertainty in this prediction. While sovereign debt currently accounts for a major share of the tokenization market, the future path remains to be seen.