Tiger Research: Realism is the Only Answer in the Cryptocurrency Downturn

marsbitPublié le 2026-02-09Dernière mise à jour le 2026-02-09

Résumé

Tiger Research argues that during the prolonged crypto market downturn, only projects grounded in realism and practical problems are demonstrating resilience. The report highlights three key examples: Hyperliquid, which addresses immediate trading friction by building a performant perp DEX; Canton Network, which provides a privacy-focused, compliant infrastructure for institutional finance and real-world assets (RWA); and Kite AI, which is building payment infrastructure for a future AI agent economy. The common thread is a focus on solving real, existing market problems with structurally sound solutions. To assess a project's practical viability, the analysis should focus on three core questions: what specific problem it solves, how the solution is structured, and whether the team has the execution capability. Ultimately, the ability to deliver practical utility is the most powerful fundamental in a bear market.

This report is written by Tiger Research. The cryptocurrency market continues to be in a prolonged downturn. In this environment, the projects that can survive are those that demonstrate pragmatism and a realistic vision.

Core Points

  • Projects that solve real, specific problems can maintain resilience even during market downturns.
  • Hyperliquid, Canton, and Kite target different problem areas, but they share a common trait: providing practical and realistic solutions, rather than abstract narratives.
  • To assess this realism, analysis should focus on three factors: the problem the project aims to solve, the structure of the solution, and the team's ability to execute in practice.

1. Conditions for Survival in a Bear Market: Does it Work in Practice?

Source: Tiger Research

Bitcoin has fallen below $70,000. Among the top 100 cryptocurrencies by market cap, only 7 remain above their 200-day moving average. In contrast, 53 components of the Nasdaq 100 index are still trading above this threshold.

The market conditions are undeniable. Nevertheless, some crypto assets still manage to survive in the harshest environments.

Their resilience cannot be simply attributed to artificial market making or coincidental rebounds. A closer look at their development trajectory reveals a different explanation.

These projects no longer rely solely on vague visions or technical complexity. Instead, they share a common feature: solving core market problems with solutions rooted in practical reality. Their approach typically aligns with three directions:

  1. Do they solve a problem the market is currently facing?
  2. Are they ready for practical application in the near term?
  3. Are they building infrastructure that the industry will rely on long-term?

Ultimately, the ability to solve real problems in practice remains the most powerful fundamental.

2. Three Directions Chosen by the Market

Projects that can answer the above questions have successfully survived. Their approach is: 1) clearly identify a market problem; 2) propose a practical solution that matches a specific timing.

2.1. Hyperliquid: Solving Immediate Trading Friction

Centralized exchanges have traditionally been seen as responsible intermediaries. However, in practice, when problems arise, they often fail to align with investor interests. Decentralized exchanges emerged as an alternative, but poor user experience and performance have caused many investors to steer clear.

In this context, Hyperliquid introduced the concept of a perpetual contracts decentralized exchange (perp DEX). Through its HLP mechanism, it brings the features investors value in centralized exchanges—such as high leverage, fast execution, and stable liquidity—into an on-chain environment.

Early usage was partly driven by demand for the $HYPE token airdrop. However, continued engagement post-airdrop reflects user satisfaction with the platform's performance.

Ultimately, Hyperliquid's resilience stems from solving a persistent, real-world problem: dissatisfaction with centralized exchanges.

2.2. Canton Network: Preparing for the Era of Institutional Finance

Canton proposes a solution for the near future. As interest in real-world assets (RWA) continues to rise, institutions are beginning to see blockchain as financial infrastructure, not a public network. In this context, what institutions need is not complete data transparency, but a selective privacy model that supports regulatory compliance and confidentiality.

The Canton Network was born. Through DAML, Canton can provide configurable data disclosure for each participant.

This allows institutions to maintain transaction confidentiality while sharing information only to the necessary extent. Instead of imposing a provider-driven design, Canton builds infrastructure that aligns with institutional needs.

Another key factor is that Canton, from the outset, has focused on scaling its ecosystem with real-world deployments, supported by early partnerships with financial institutions.

Most notably, its partnership with DTCC establishes a pathway for assets managed by the traditional financial system to extend into a Canton-based environment. DTCC handles approximately $3.7 quadrillion in transactions annually, highlighting the practical feasibility of the Canton Network's approach.

Ultimately, the Canton Network provides a structural solution designed to meet three institutional requirements simultaneously: privacy protection, regulatory compliance, and integration with the existing financial system.

2.3. Kite AI: Building the AI Economy That Has Not Yet Arrived

Unlike the previous two examples, Kite AI currently has limited practical application. However, from the future perspective of AI agents operating as economic actors, its structural logic remains compelling.

There is broad consensus in both Web2 and Web3 about an agent-driven future. Few question scenarios where AI agents handle tasks like booking hotels or buying groceries on behalf of users.

However, such a future requires infrastructure that allows AI agents to independently initiate and execute payments. Existing transaction systems are designed for transfers between people, and for efficiency among human participants.

Therefore, for AI agents to operate as autonomous economic entities, new mechanisms are needed, including identity verification and automated payment frameworks.

Kite AI is building payment infrastructure for this environment. Its core components include an "Agent Passport" for identity verification and the x402 protocol functionality for enabling automated payments.

The vision proposed by Kite AI cannot be deployed at scale currently, simply because the future it targets has not yet materialized.

Nonetheless, the project's realism stems from a broader premise: when this widely anticipated future arrives, the underlying technology it is developing will be necessary. This alignment with a widely accepted development trajectory gives the project structural credibility, despite its current limited use.

3. Three Key Questions for Assessing Practical Feasibility

Although these three projects have different timelines, they share a common characteristic: real-world feasibility.

Evaluations of the same project often diverge. Some believe it solves a real problem, while others see it as overhyped. To bridge this interpretation gap, at least three core questions must be asked:

Source: Tiger Research

  1. What problem does it solve? Is the problem the project targets real and is there market demand for it?
  2. How does it solve it? Is the proposed solution structurally sound and executable?
  3. Who is executing it? Does the team have the capability and resources to turn the vision into reality?

Since most projects promote optimistic future narratives, answering these questions correctly requires time and effort. Filtering out misleading or incomplete information is not easy. Projects that cannot confidently answer these three questions may experience short-term price increases, but when the next downturn comes, they will likely disappear.

The current state of the cryptocurrency market is clearly unfavorable. But that doesn't mean it's all over. New experiments will continue, and the task is to assess what these efforts truly represent.

What matters most now is realism.

Questions liées

QAccording to the Tiger Research article, what is the common characteristic of projects that survive in a prolonged crypto downturn?

AThe common characteristic is that they provide practical and realistic solutions to real, concrete problems, rather than relying on abstract narratives.

QWhat are the key factors to analyze when evaluating a project's realism, as outlined in the article?

AThe analysis should focus on three factors: the problem the project aims to solve, the structure of its proposed solution, and the team's ability to execute in practice.

QWhat specific market problem does Hyperliquid address, and how does it solve it?

AHyperliquid addresses the friction and lack of alignment with investor interests on centralized exchanges. It solves this by introducing a perpetual contracts decentralized exchange (perp DEX) that brings features like high leverage, fast execution, and stable liquidity on-chain through its HLP mechanism.

QHow does the Canton Network cater to the needs of institutional finance, as described in the report?

ACanton Network caters to institutional needs by providing a selective privacy model that supports regulatory compliance and confidentiality. It uses DAML language to allow configurable data disclosure for each participant, enabling institutions to share information only as necessary while maintaining transaction privacy and integrating with existing financial systems.

QWhat future infrastructure is Kite AI building, and why is it considered realistic even with limited current application?

AKite AI is building payment infrastructure for a future where AI agents act as autonomous economic entities. This includes an 'Agent Passport' for identity verification and the x402 protocol for automated payments. It is considered realistic because its vision aligns with the widely anticipated trajectory of AI development, making its underlying technology a necessary component for that future.

Lectures associées

Google et Amazon investissent simultanément dans un concurrent, la logique commerciale la plus absurde de l'ère IA devient réalité

En l'espace de 4 jours, Amazon a annoncé un investissement supplémentaire de 25 milliards de dollars et Google jusqu'à 40 milliards de dollars dans Anthropic, une startup d'IA. Ces deux concurrents directs ont ainsi engagé plus de 65 milliards de dollars dans la même entreprise. Cet investissement massif ne relève pas d'une logique capitalistique traditionnelle, mais marque le début d'un nouveau round dans la guerre du cloud. L'enjeu pour Google et Amazon n'est pas la simple possession d'une part d'Anthropic, mais la sécurisation de contrats de pré-achat de puissance de calcul (ou "pré-vente de calcul"). Les fonds investis sont conditionnés à ce qu'Anthropic les dépense en services cloud et puces des investisseurs. Amazon a obtenu un engagement de dépense de plus de 1000 milliards de dollars sur AWS sur dix ans, et Google fournira environ 5 gigawatts de puissance de calcul. La concurrence dans le cloud a changé : les entreprises ne choisissent plus un fournisseur pour son prix ou sa stabilité, mais pour savoir "quel modèle d'IA de pointe tourne sur son cloud". Le modèle détermine le choix de la puissance de calcul. OpenAI étant déjà étroitement lié à Microsoft, Anthropic et son modèle Claude sont devenus la seule cible stratégique disponible pour Google et Amazon pour ne pas perdre des clients entreprises. Anthropic, avec un revenu annuel récurrent (ARR) de 30 milliards de dollars, est devenu un acteur infrastructurel incontournable. Cependant, cette dépendance financière et technique envers deux géants concurrents pose des défis pour son indépendance future, sa narration autour de la sécurité et sa pression pour une introduction en bourse. La situation contraste avec celle de la Chine, où des investissements comme celui d'Alibaba et Tencent dans DeepSeek (modèle open-source) suivent une logique différente, moins centrée sur un verrouillage par la puissance de calcul et plus ouverte. Le paysage de l'IA évolue ainsi vers une structure à plusieurs pôles fermés (modèles privateurs liés à un cloud spécifique), tandis que les modèles open-source offrent une alternative cruciale pour l'écosystème.

marsbitIl y a 4 h

Google et Amazon investissent simultanément dans un concurrent, la logique commerciale la plus absurde de l'ère IA devient réalité

marsbitIl y a 4 h

Capacité de calcul limitée : pourquoi DeepSeek-V4 est-il open source ?

Face à des contraintes de calcul, DeepSeek a choisi de rendre son modèle DeepSeek-V4 open source, tout en proposant une version Pro aux capacités avancées mais limitée en débit en raison de ressources de calcul haute performance insuffisantes. Le modèle exploite une architecture MoE (Mixture of Experts) avec 1,6T de paramètres au total mais seulement 49B activés lors de l'inférence, permettant une gestion de contexte longue (1 million de tokens). Une version Flash, avec 284B de paramètres totaux et seulement 13B activés, est conçue pour une adoption massive sur du matériel moins performant, visant ainsi les PME et développeurs. Le modèle excelle dans des tâches exigeantes comme la génération de code et le raisonnement complexe, rivalisant avec des modèles fermés leaders. DeepSeek a également collaboré avec des fabricants de puces chinoises (Huawei, Cambricon, Hygon) pour optimiser l'exécution sur du matériel local, bien que des défis persistent en termes de performance pure et d’approvisionnement. Cette stratégie intervient dans un contexte de concurrence intense sur le marché chinois des modèles de langage et de départs clés dans l’équipe R&D. DeepSeek cherche également à lever des fonds à une valorisation élevée, et cette version open source constitue une démonstration de sa résilience technologique et de son orientation pragmatique vers une IA accessible malgré les limites actuelles du calcul.

marsbitIl y a 5 h

Capacité de calcul limitée : pourquoi DeepSeek-V4 est-il open source ?

marsbitIl y a 5 h

Trading

Spot
Futures
活动图片