After the Google AI conference, some AI tracks continue to decline while others thrive. The market has entered a period of calm after the short-term speculation of AI concepts. Amidst the mixed results, it is clear that not all AI tracks are struggling. So, which tracks continue to flourish in the AI boom and what is the underlying logic behind it?
The Google I/O conference, held on May 14th, showcased a series of significant products in various key areas, leading to a divergence in the AI track. Gemini 1.5 Pro, an upgraded version, was made available to developers, increasing its maximum support for context windows from 1 million tokens to 2 million, and supporting 35 languages. This allows for the analysis of longer documents, code repositories, videos, and audio recordings than before. In response to OpenAI, Google also released multiple multimodal models, including the Image 3 image model and the Veo video model. In the field of education, Google introduced LearnLM, a series of new generative AI models for learning. At the I/O conference, Google also unveiled Gemini Live, a dialogue experience product based on Gemini that allows users to have in-depth voice conversations with Gemini on their smartphones, countering OpenAI’s recent large model dialogue demonstration.
Project Astra, as Google’s first AI Agent product and an innovative part of the Gemini Live technology, aims to create artificial intelligence applications and agents for real-time and multimodal understanding. Additionally, Google is integrating Gemini’s large models deeply into its search engine, indicating a shift from a tool service to an entry point for AI agents. Google has also introduced a range of new AI features on the Android platform, such as “Circle to Search,” which enables cross-app searches and can answer complex questions like math problems and charts. It can even summarize and provide intelligent replies in the user’s email app.
After Google’s release of these products, the AI track has undergone subtle changes. In a sense, due to the deep research and investment required in the AI industry, once the industry barriers are established, it becomes difficult for other projects and teams to break through in the short term. Additionally, the AI track is currently experiencing rapid growth and evolving at a fast pace, leading many projects to constantly update their whitepapers and adjust their design strategies. With the rapid expansion of AI frontrunners, some projects may find it difficult to compete and may essentially declare their demise. However, the strong will continue to thrive, and not all AI projects in the blockchain track will wither away.
AR rises against the trend with its AO system’s advantages. According to CoinmarketCap data, the leading AI projects, WLD and FET, have declined by 17.12% and 35.6% respectively in the past seven days from their peak this year. In contrast, Arweave, an established decentralized storage project that announced its transition into an AI public chain this year, has shown strong performance with a 19.33% increase in the past seven days, reaching its early-year high. So why does Arweave continue to be popular in the market, and what are its technological advantages?
Arweave is a new type of blockchain storage network designed to solve the problem of permanent data storage and access. Its distinctive feature is the use of “permanent storage,” which ensures that information is never lost by storing data on the blockchain. Arweave’s cryptographic economic model incentivizes users to store and transmit data, ensuring the security and reliability of the network. This allows users to securely store important data such as documents, pictures, and videos and access them at any time without worrying about data loss.
The main AI solution proposed by the Arweave team is AO (Actor Oriented). AO is a massively parallel computing protocol that provides a decentralized computing environment, allowing for parallel execution of any number of processes. Compared to previous decentralized computing systems, AO can achieve both large-scale computing and verifiable computing. AO is implemented through three different types of subnetworks and the Arweave base layer: the Messenger Unit (MU), the Scheduler Unit (SU), and the Compute Unit (CU). MU is responsible for receiving and processing information, SU is used for scheduling and sorting information, and CU is used for computation.
Overall, Arweave is a foundational decentralized storage layer, and AO is a decentralized computing protocol built on Arweave. Developers can use the Lua language to develop applications (write smart contracts) in AOS, a specific operating system based on the AO parallel computing protocol, and Lua is a user-friendly high-level language. Mature DApps developed based on AO include the decentralized exchange Bark and the decentralized stablecoin protocol astro. Arweave founder Sam Williams believes that AO is a completely different programming paradigm from smart contract systems and is essentially the best solution for all distributed systems and Web2 era.
With the outbreak of the AI revolution, AI training models are becoming increasingly complex, and the demand for computing power is growing exponentially. Traditional centralized computing may struggle to meet market demands in the future. For example, OpenAI found that from 2012 to 2018, the computational demands of its models doubled every two years and later doubled every three and a half months. This has led to a surge in demand for GPUs, with some cryptocurrency miners even utilizing their GPUs to provide cloud computing services.
Aside from Arweave, many AI projects that focus on decentralized storage have shown relatively strong performance, making decentralized storage the potential biggest winner in the AI boom. Projects like Akash, io.net, iExec, and Cudos are decentralized computing applications that not only provide data and general computing solutions but also offer or will soon offer specific computing resources for AI training and inference. Taking Akash as an example, let’s further explore the significant opportunities in the decentralized storage track in the AI boom.
Akash Network is a decentralized cloud computing platform that aims to integrate underutilized computing resources worldwide by providing a peer-to-peer marketplace. It establishes an open and transparent market where users can freely publish their resource demands and global resource providers can bid in real-time, reducing the cost of cloud services. According to a report by Messari, Akash’s costs for the same hardware are much lower than other cloud providers. Akash was founded in 2015 and launched its mainnet in the Cosmos ecosystem in 2020. Initially focusing on CPU computing, Akash Network completed its mainnet 6 upgrade on August 31, 2023, and started supporting the GPU cloud market.
Akash Network effectively utilizes idle resources by collaborating with multiple large miners. As of May 20, 2024, Akash Network has over 17,800 CPUs and 365 GPUs, and these numbers continue to rise. From the demand side, after introducing GPUs in August 2023, Akash’s daily leasing volume has significantly increased, with over 180,000 leases accumulated and daily revenue from leasing continuing to grow.
In essence, blockchain technology provides a new paradigm for transaction settlement, data storage, and system design, while artificial intelligence revolutionizes computation, analysis, and content production. With the rapid growth of AI technology and its future explosion, the demand for computing power will inevitably continue to rise, making decentralized storage potentially the biggest winner. In addition, Arweave further connects AI with decentralized storage through its AO protocol, which has even greater potential.
For more information, please join us at:
Twitter: https://twitter.com/BitpushNewsCN
Bitpush Discord Community: https://discord.com/invite/QSvv7MZ2tz
Bitpush TG Community: https://t.me/BitPushCommunity
Bitpush TG Subscription: https://t.me/bitpush
This article is from Bitpush. Please indicate the source when reprinting.
Original article link: https://www.bitpush.news/articles/6773337