jakiro
Written article: 0xxz@：
The famous Wall Street investor Catherine Wood and its ARK Investment have released the “Big Ideas” report since 2017, which has always been quite concerned about market attention.
With Big IDEAS 2023, ARK believes that 5 major categories of technologies define our era: public chain, AI, energy storage, robotics, and multiple groups of sequencing.
The “Big Ideas 2024” report is 163 pages and 15 parts.Except for the total view of the TECHNOLOGAL Convergence, the other 14 partsThere are 4 cryptocurrencies and web3s: AI, Bitcoin configuration, and Bitcoin and smart contracts in 2023.
The “Big IDEAS 2024” report was released during the New Year of China in China in 2024. The content of the report, especially the content of encryption, was not completely translated to the SMC network.Bitchain has compiled this part of the content to readers.
In order to take care of the reading experience, we put the Bitcoin configuration, the Bitcoin and smart contracts in 2023, and the AI part at the end.
Bitcoin configuration
Digital assets, such as Bitcoin, are a new asset category
According to ARK’s research, Bitcoin has become an independent asset category, which is worth a strategic allocation in institutional investment portfolios.
Bitcoin performed better than all major assets within a longer period of time
In the past seven years, the average annual return on Bitcoin is about 44%, while the average return rate of other major assets is only 5.7%Essence
Overall, long -term investors holding Bitcoin benefit over time over time
“Time, and choice”: Bitcoin’s volatility may cover its long -term returns.Although a significant appreciation or depreciation may occur in the short term, long -term investment vision has always been the key to investing in Bitcoin.Instead of entangled the “when”, the better question should be “How long does it take?” In history, investors who buy and hold Bitcoin at least 5 years have profit.Essence
The correlation between Bitcoin and traditional assets is very low
Historically, the price trend of Bitcoin is not highly related to other asset categories.In the past five years, the average correlation between Bitcoin returns and traditional asset categories is only 0.27.
Bitcoin can play an important role in maximizing risk adjustment returns
The research of ARK focuses on the volatility and returns of traditional assets, and suggested that a investment portfolio that seek to maximize risk adjustment should be distributed to Bitcoin by 19.4% of funds in 2023.
On the basis of 5 years, in the past 9 years, the configuration of Bitcoin can maximize the risk adjustment and return
According to our analysis, in 2015, within 5 years, the best configuration ratio of maximizing risk adjustment returns was 0.5%.Since then, on the same basis, the average configuration of Bitcoin is 4.8%, and it has reached 19.4%in 2023 alone.
What impact will the best configuration Bitcoin have?
The allocation of Bitcoin from a global investment asset basis of $ 25 trillion will have a significant impact on prices.
The 1%configuration will make the price of Bitcoin reaches $ 120,000; the configuration of 4.8%will make the price of Bitcoin reach $ 550,000; the configuration of 19.4%will make the price of Bitcoin reach $ 2.3 million.
Bitcoin in 2023
Bitcoin prices soared by 155%in 2023, and the market value increased to $ 827 billion
Bitcoin prices break through the average market value on the chain in the past 4 years
The average market value on the chain is an original indicator of ARK. It is a reliable division between risk aversion and risk appetite.In history,When the price of Bitcoin breaks through the average market value, it usually marks the early stage of the bull market.
It provided an important answer to the encryption crisis in 2022 in 2023
During the large -scale closure of banks in the United States, Bitcoin became a shelter
In early 2023, during the historic failure of banks in the United States, the price of Bitcoin rose by more than 40%, highlight
The surge in the number of inscriptions indicates that the Bitcoin network has other functions in addition to transaction settlement
The Bitcoin inscription launched in January 2023 introduced a unique number system that assigned a number for each “Cong” (the smallest unit of Bitcoin) based on its location in the blockchain.Each Cong is recognized and unchanged, allowing users to record data, images or texts on it.Unlike other blockchains that need smart contracts for NFT, the Bitcoin inscription is located at the bottom of the Bitcoin blockchain.
The impact of the inscription on the scale and block space of the transaction has triggered the Debate.We believe that inscriptions are the products of free markets and represent health innovations on Bitcoin.
The fundamentals of Bitcoin have no problems in the crisis of 2022, and continue to maintain stable development in 2023
The CME of the American Commodity Futures Exchange surpassed Binance and became the world’s largest Bitcoin Futures Exchange
With the increase in demand for more regulatory and safer infrastructure after the mines in the cryptocry market in 2022, the market developments of Bitcoin turned more to the United States.
Bitcoin is evolving into a reliable hedging asset
With the increase of macroeconomic uncertainty and the decline in trust in traditional “insurance assets”, Bitcoin has become a feasible alternative option.
Bitcoin will face major catalysts in 2024:
Bitcoin spot ETF issuance, Bitcoin halving, institutional acceptance, supervision progress
Smart contract
The smart contract deployed on the public chain provides a globalization, automation and audit alternative, which can replace intermediaries and traditional financial infrastructure that seek rent.
After the “encryption market crisis” in 2022, some digital asset solutions have been favored by the market, including stable currency, tokens, and capacity expansion technologies.
According to the research of ARK, with the increase in the value of financial assets on the chain, market value related to decentralized applications may increase at a rate of 32% per year, which will increase from US $ 775 billion in 2023 to 5.230 trillion in 2030Dollar.
Smart contract is the basis of the Internet financial system
As a new thing, smart contracts are promoting a new Internet native financial system.Due to the largest intelligent contract blockchain, multiple networks are supporting chain activities and competing for market share.
Stable currency highlights the value proposition of the smart contract
In view of the malignant inflation in emerging markets and the increase in global instability, the demand for stable coins to reach the US dollar has soared.In the past three years, the number of active stable currency addresses around the world has increased at a rate of 93% per year, from 171,000 to 1.2 million.In 2023, the stable currency transfer volume exceeded Mastercard.
Traditional financial assets are transferring to the chain
Vigatization enables financial personnel to track, transactions and mortgage funds on public chains than in the traditional financial market.In 2023, the capital funds of the tokens increased by more than 7 times to $ 850 million.Early funds started on the Stellar blockchain, but Ethereum became the largest market for Vinghua Treasury bonds in 2023.
Developers optimized the agreement during the bear market
In the face of the encryption market crisis and its consequences in 2022, core developers promote the technical route chart and strengthen the agreement to support the next round of bull market.Ethereum successfully migrated to the POS consensus mechanism, Solana set a new record for continuous operation time.
The L2 network has expanded the transaction in the Ethereum ecosystem
Since the beginning of 2021, more than 20 L2 networks have been launched, which has enabled Ethereum to expand the average daily transaction volume by 4 times with lower costs.Despite early success, most L2 networks are controlled by centralized.The experience of surge in L2’s surge in users and developers has become complicated.
The lower cost is increasing the participation of the chain
As the cost of transaction decreases, the chain participation (measurement at the ratio of daily active address (DAU) and Moon active address (MAU)) has increased.
Single chains like Solana provide an alternative for vertical extension
Smart contract network design needs to weigh the house.Ethereum ecosystems are decentralized by priority, making it more complicated during the expansion process.Solana has maintained a simple architecture for users and application developers through priority to consider single -layer scalability and has obtained attraction.
Smart contracts can greatly reduce financial service costs
Due to factors such as global economic growth, increasing finance, and equity multiple expansion, global financial assets value expanded from $ 140 trillion in 2000 to US $ 51.0 trillion in 2020.The operating costs of global financial systems have also increased simultaneously with the increase in financial asset value.As a $ 20 trillion financial service industry with a total annual revenue, its rate is 3.3%compared with all financial assets.Smart contracts can greatly reduce economic burden.
By 2030, the intelligent contract network may incur a cost of $ 450 billion
Smart contracts can promote the creation, ownership and management of assets on the chain with a small part of the traditional financial cost.If financial assets are transferred to blockchain infrastructure like the Internet, and the rate of decentralized financial services is one -third of traditional financial services, then smart contracts can incur a cost of more than 450 billion U.S. dollars per year by 2030, creating creation, creating creation.The market value of more than 5 trillion US dollars has increased at a compound annual growth rate of 78% and 32%, respectively.
AI
AI models that show super-human levels in various tests, such as GPT-4, should give birth to an unprecedented productivity explosion.ChatGPT’s “iPhone moment” makes companies be caught off guard, they are now scrambling to use the potential of AI.
Due to the rapid decline in costs and the emergence of open source models, AI can achieve far exceeding the promise of improving efficiency.If by 2030, the productive forces of intellectual workers can be turned over, as we believe, in the next five to ten years, the actual GDP growth may accelerate and break the record.
ChatGPT satisfies consumers and amazed by the company
Since Google invented the TRANSFORMER architecture in 2017, after years of progress, ChatGPT catalyzed the public’s understanding of the generation AI.It is no longer just a tool for developers. The simple chat interface of ChatGPT enables people who speak any language to use the powerful features of large language models (LLM).In 2023, the enterprise scrambled to understand and deploy the generation AI.
AI has significantly improved productivity
Code assistants like GitHub Copilot and REPLIT AI are early successful cases that improve the productivity and work satisfaction of software developers.The AI assistant is improving the performance of intellectual workers. Interestingly, they are more beneficial to employees with poor performance than high -performancers.
Basic models are being improved across the field
With larger training data sets and more parameters, GPT-4 is significantly better than GPT-3.5.More and more basic models have become “multi -modal” -the support text, images, audio and videos -not only more dynamic, easier to use users, but also higher performance.
Wenshengtu model is reshaping the graphic design
For the first time, researchers at the University of Toronto launched the modern Wensheng model for the first time, the output of the image model is now comparable to the works of professional graphic designers.Human designers can spend hundreds of dollars to create a image within a few hours, such as a group of elephants through green grass.Wenshengtu model can generate the same image with a few cents in a few seconds.Professional applications such as Adobe Photoshop and consumer applications such as Lensa and ChatGPT integrate image models into their products and services.
The cost of creative text content has plummeted
In the past century, the cost of creating written content has been relatively stable in actual value.In the past two years, the cost has plummeted with the quality of LLM’s writing.
AI training performance is rapidly improving
AI researchers are innovating across training, reasoning, hardware, and model design to improve performance and reduce costs.
Training costs continue to decrease by 75% per year
According to WRIGHT’s law, the improvement of hardware to compute hardware should reduce the production cost of AI relative computing units (RCUs) by 53%per year, and the improvement of algorithm models can further reduce the training cost by 47%.In other words, the integration of hardware and software can promote the cost of AI training in 2030 at a rate of 75% per year.
With the emergence of production cases, the focus of AI is moving to the cost of reasoning
After the researchers were initially focused on the optimization of LLM training costs, the cost of reasoning was now used as a priority.Based on enterprise -based use cases, the cost of reasoning seems to decline at about 86% of the annual growth rate, and even decreases faster than training costs.Today, GPT-3 related to GPT-4 Turbo is lower than the GPT-3 a year ago.
Open source community is competing with private models
In order to challenge the closed source models of OpenAI and Google, the open source community and its corporate giant Meta are democratizing the access to the generation AI.All in all, the performance improvement of the open source model is faster than the closed source model. Recently, some models in China have also helped the power.
The performance improvement of language model requires meticulous technology
The GPT-4 is significantly better than ordinary humans in the standardized education test from SAT to senior winemakers.However, according to Winnography’s measurement, it lags behind human level in common sense reasoning.Stanford University’s framework -the overall evaluation of the language model (HELM) is one of the most comprehensive and updated evaluation methods. It has tested more than 80 models and combined evaluation for 73 scenes and 65 indicators.
Will the language model run out of the data, to limit its performance?
Computing power and high -quality training data seem to be the main contribution of model performance.As the model becomes larger and larger, more training data is needed, will the data insufficient cause the model performance stagnation?EPOCH AI estimates that by 2024, high -quality language/data sources, such as books and scientific papers, may be exhausted, although there are still a large number of unopened visual data.
Customized artificial intelligence products should have greater pricing power
With the appearance and cost of open source alternatives, software suppliers that customize AI to final use applications should be able to make profit more easily.
