A Note From Cathie Wood, & More

Want more research from the ARK Team? Have feedback on our publications? Click here to help inform our content creation.

1. A Note From Cathie Wood

By: Cathie Wood | CEO/CIO

If you have not tuned into ARK’s "Big Ideas 2025," please visit here and learn more about the technologies that are likely to transform our lives and the investment landscape. With gratitude, as always, to our research, marketing, and compliance teams for bringing "Big Ideas 2025" to life!

2. Deep Research, OpenAI’s AI Agent, Is Pushing The Frontiers Of Artificial Intelligence

By: Frank Downing | Director of Research, Next Generation Internet

Last week, OpenAI launched Deep Research, an AI agent designed to perform multi-step research tasks based on information across the internet. Completing in minutes what typically would have taken hours of human research, the agent is available to Pro users for $200 per month.

Powered by its o3 model, OpenAI optimized Deep Research for web browsing and data analysis. After users query ChatGPT, the agent autonomously searches, analyzes, and synthesizes information from hundreds of online sources, documenting the output with citations, uploaded files, data visualizations, embedded images, and sidebars that track progress during the 5-30-minute research process. Based on our observations, OpenAI's new agent provides deeper research and analysis than Google's Deep Research product launched in December.

Recently, OpenAI’s Deep Research accurately answered 26.6% of the questions on Humanity's Last Exam and created a state-of-the-art score on the GAIA benchmark for real-world AI reasoning.[1] Humanity's Last Exam is a new benchmark that includes the most difficult academic questions possible across many disciplines. Upon the Exam’s release in January, the best performing models were OpenAI's o1 and DeepSeek's R-1, each scoring ~9%.[2] If subsequent model iterations were to follow the success rate of past models relative to SWE-Bench and?ARC-AGI—two particularly challenging AI model benchmark tests that have been solved[3]— Humanity’s Last Exam could be saturated within the next 12 months, effectively surpassing expert level technical knowledge and reasoning capability.

Of course, OpenAI’s Deep Research agent isn’t perfect: it can hallucinate, take a long time to respond, err in formatting results, and make mistakes when distinguishing between credible sources. Updates are likely to add proprietary data sources and integrate with OpenAI’s Operator for automated execution, key features that could unlock greater value for enterprises aiming to accelerate the pace of knowledge work with Deep Research.

3. Stablecoins Are Hitting All-Time High Usage Levels And Reshaping The Digital Asset Space

By: Lorenzo Valente | Director of Research, Digital Assets

As illustrated in ARK’s "Big Ideas 2025" released this week, stablecoins are one of the most transformative and high-growth sectors in the digital asset space. Despite the severe bear market in digital assets in 2022 and 2023, for example, stablecoins grew steadily and hit new highs across most key metrics.

In 2024, stablecoin transaction value surpassed those of payment giants Visa and Mastercard globally, as shown below. Today, metrics like total supply, transaction count, number of holders, and number of stablecoins are at all-time highs.

Tether (USDT) and Circle (USDC), the two largest stablecoin issuers, have caught many financial institutions off guard with their rapid growth and profitability. In 2024, for example, with fewer than 200 employees, Tether hit $13 billion in net income—roughly the same level of income that ~35,000 employees at Mastercard generated—becoming one of the most efficient and profitable businesses in history.[4]

Today, the supply of stablecoins is ~$220 billion, roughly 1% of M2 in the US. According to our research, the supply could hit $1.4 trillion, equivalent to the 14th largest monetary base in the world, by 2030.[5]

Historically, Bitcoin and Ethereum were the primary entry points for newcomers to the digital asset space. Recently, stablecoins usurped that role. They settle more transaction volume across all blockchains than either Bitcoin or Ethereum, highlighting their growing dominance as the foundation of on-chain activity.[6]

4. A Decade Of Innovation

By: Brett Winton | Chief Futurist

A decade ago, ARK Invest started when the innovation platforms around which we centered our research and investing were nascent. The Model S had debuted but was among the most expensive cars on the road, and autonomous robotaxis were nowhere on Tesla’s product roadmap. Humanoid robots often toppled over and rockets never landed. Language was considered off-limits for artificial intelligence, and GPUs were little more than PC gaming chips. Bitcoin’s “market cap” was $6 billion and Coinbase had fewer than a million active customers. Companies focused on CRISPr gene editing were new to the public market scene and, while the cost to sequence a whole human genome had dropped to $1,000, its clinical utility was still questionable.

The capital markets and macroeconomic environment were very different as well, as shown below. In 2014, global nominal GDP[7] had topped $70 trillion, higher than the $50 trillion in global equity market cap, of which disruptive innovation was only ~$6 trillion.

A decade later, the landscape is different. The economy is bigger—nearly $130 trillion—but capital markets have grown faster and are roughly equal to GDP in size. Innovation assets have appreciated faster still, and at $27 trillion make up more than 20% of the global equity market capitalization.[8]

Now, an electric vehicle is the world’s best-selling car, and autonomous robotaxis have commercialized in multiple cities. SpaceX rockets nearly always land, and humanoid robots that imitate human movement have found their first customers. Language has become the dominant interface for artificial intelligence systems that are surpassing the knowledge of experts across many domains, and AI is driving unprecedented demand for GPUs and computation. The cryptoasset market cap now exceeds $3 trillion, and Coinbase has more customers than all but three US banks. A CRISPr Cas-9 gene-editing cure for sickle cell disease has commercialized, and oncologists are prescribing hundreds of thousands of sequencing-based cancer detection tests annually.

After a remarkable decade of innovation, we believe the next five years will prove more meaningful, as detailed in ARK’s "Big Ideas 2025" released this week. Robotaxis should become a dominant form of passenger transportation, potentially creating an industry worth trillions in annual spending. Humanoid robots will be performing rudimentary chores in households while accelerating the rate of production in factories meaningfully. Artificial intelligence could approach $10 trillion in annualized revenues and command $10s of trillions in market capitalization. Even while still early in their disruption of financial markets, cryptocurrencies and smart contracting protocols also could command $10s of trillions in value. In health, other diseases will have been cured, with many more cures in pipeline, while diagnosing and developing treatments for diseases will likely be much more efficient and more profitable.

If our research on these breakthroughs is correct, the enterprise value of innovators could increase at a stunning rate while that of companies left behind could shrink. According to our forecasts, disruptive innovation could account for more than two-thirds of the global equity market capitalization in 2030. Despite a productivity-driven acceleration in macroeconomic growth, the value of companies not innovating could shrink as technology drives costs down and disrupts existing business models.

We also expect the bull market in innovation to broaden out to many companies. In the past five years, a handful of companies—the so-called Mag 6—have increased from one-third to two-thirds of the equity market's innovation capitalization. During the next five years, while in the aggregate the Mag 6 should continue to perform well, our forecasts suggest that disruptive competitors will grow faster, returning the Mag 6 to its pre-Covid share of the innovation market cap, as shown below.

Yes, those large well-known technology companies could continue to grow rapidly, but our forecasts suggest that more agile, aggressive, and disruptive companies will accrue value more rapidly, creating a historic technological business cycle.

SUBSCRIBE TO ARK'S RESEARCH

[1]?The GAIA Benchmark is a benchmark that challenges AI assistants with simple, everyday questions requiring skills like reasoning, multi-modality, and tool use—tasks that humans handle easily but that are pose struggles for many advanced AIs. The GAIA benchmark scores noted here are available on the benchmark’s website at (https://ai.meta.com/research/publications/gaia-a-benchmark-for-general-ai-assistants/)

[2] The SWE-Bench and ARC-AGI benchmark scores noted here are available on each of the benchmarks’ websites respectively at (https://www.swebench.com/)?(https://arcprize.org/arc).

[3] SWE-bench verified is a human-verified subset of SWE-bench, a challenging benchmark for evaluating an AI agent's coding capabilities by testing whether the agent can solve real-world issues in GitHub code repositories. See OpenAI. 2024. ARC-AGI is a benchmark for general intelligence that challenges AI systems to learn abstract rules from a few examples and solve novel grid-based tasks accurately. These tasks typically are very easy for humans to solve but challenging for AI systems.

[4] Based on data from Hoffman, P. 2024. “The world’s most successful companies by profit per employee.” Best Brokers. Data as of December 31, 2024. On January 31, 2025, Tether released its new data for the second half of 2024, revealing that it earned $13 billion for the full-year. In ARK's Big Ideas 2025, developed before the new data were released, we showed that Tether had earned $5 billion during the first half of 2024. Readers should note the distinction as they read Big Ideas 2025. See Sandor, K. 2025. “Tether Reports $13B Profit for 2024, With Rising Bitcoin, Gold Prices Contributing.” CoinDesk.

[5] Based on data from rwa.xyz (“Stablecoins”) as of December 31, 2024. M2 is a measure of the US money stock that includes M1 (currency and coins held by the non-bank public, checkable deposits, and travelers’ checks) plus savings deposits (including money market deposit accounts), small time deposits under $100,000, and shares in retail money market mutual funds.

[6] Based on data from Artemis Terminal as of December 31, 2024.

[7] Gross domestic product (GDP) measures the final market value of all goods and services produced within a country. It is the most frequently used indicator of economic activity. The GDP by expenditure approach measures total final expenditures (at purchasers' prices), including exports less imports. “Nominal GDP” is not adjusted for inflation. “Real GDP” is adjusted for inflation.

[8] Here and throughout this piece, we include cryptoasset market cap as part of global equity market cap.