ARK Big Ideas 2026：大加速時代來臨，你準備好了嗎？¶
繁體中文總結¶
ARK Invest 發布第 10 份年度研究報告《Big Ideas 2026》，主張全球經濟正站在歷史性轉折點，進入「大加速時代 (The Great Acceleration)」。這不是單一技術的進步，而是五大核心創新平台——人工智慧 (AI)、公共區塊鏈、機器人技術、能源儲存和多組學 (Multiomics)——正在匯聚並相互加強。當這些技術匯流，帶來的效率提升將是指數級的，而非線性的成長。ARK 預測全球實質 GDP 年成長率將達到驚人的 7.3%，與國際貨幣基金組織（IMF）預測的 3.1% 形成鮮明對比，意味技術融合的力量被傳統經濟模型嚴重低估。
核心理念：五大創新平台的指數級匯聚¶
關鍵洞察： 這不只是單一技術的進步，而是五大平台正在匯聚並相互加強。當這些技術匯流，帶來的效率提升將是指數級的，而非線性的成長。
五大創新平台： 1. 人工智慧 (AI) - 核心引擎 2. 公共區塊鏈 (Public Blockchains) - 金融基礎設施重塑 3. 機器人技術 (Robotics) - 從專用到通用自動化 4. 能源儲存 (Energy Storage) - 清潔能源轉型 5. 多組學 (Multiomics) - 生物醫療革命
宏觀經濟影響：GDP 的階躍式成長¶
驚人預測： ARK 預測全球實質 GDP 年成長率將達 7.3%，與 IMF 預測的 3.1% 形成鮮明對比。
資本投資將成為成長主要引擎： - 僅由這些顛覆性創新平台催化的資本投資，預計在本十年內為年化實質 GDP 成長直接貢獻 1.9% - 隨著新資本基礎建立（自動駕駛計程車車隊、AI 運算資料中心），投資資本回報率（ROIC）將被顯著推高 - 實際實現的經濟成長率可能超出目前市場共識每年 4%
無償勞動轉化為可衡量經濟產出： - 創新平台能將過去未計入 GDP 的無償勞動（家務、照護）轉化為可衡量的經濟產出 - 家庭機器人案例： ARK 估算每個家庭機器人每年可能對 GDP 產生 62,000 美元的影響 - 若人形機器人在 5 年內滲透至 80% 美國家庭，僅此一項因素就可能使美國 GDP 成長率從 2-3% 加速至 5-6%
人工智慧 (AI)：核心引擎¶
定位： 在五大平台中，AI 被定義為「核心引擎」。它不僅是獨立創新領域，更是加速其他所有技術平台發展的催化劑。其普及被認為比歷史上的「電氣化」更具重大意義，有潛力創造數十兆美元的商業價值。
基礎設施：成本崩跌與硬體競爭¶
推理成本驟降 99%： - AI 運行成本在過去一年以驚人速度下降 - 推論成本下降超過 99% - AI 使用量（以 token 計算）呈現爆炸性成長，變得像水電一樣隨手可得
數據中心投資激增： - 自從 ChatGPT 問世以來，全球數據中心系統投資年增長率從 5% 加速至 29% - 預計到 2030 年，AI 基礎設施投資將達到約 1.4 兆美元
硬體競爭加劇： - NVIDIA 主導地位： 85% GPU 銷售市占率，早期在晶片設計和軟體（CUDA）的投資形成護城河 - 競爭者崛起： AMD 和 Google (TPU) 在特定領域（如小型模型推論）迎頭趕上 - 客製化 ASIC 晶片： Amazon (Annapurna Labs)、Broadcom 等設計的客製化晶片市占率預計持續擴大
消費者作業系統：從 App 到「代理人時代 (Agentic Era)」¶
範式轉變： 我們正從依賴 App 和觸控的「移動時代」，進入由 AI 驅動的「代理人時代」。
尖銳問題： 如果 AI 寫程式比人類便宜 90% 以上，傳統靠「軟體租賃」躺賺的 SaaS 模式還行得通嗎？
答案：規則變了 - 過去護城河：程式碼 (Code) - 未來護城河：個性化體驗 (Personalization) 與 專有數據 - 未來贏家：能利用 AI 為每個用戶量身打造「專屬體驗」的企業
過往工作與消費流程將被改寫：
1. 重塑電子商務¶
購買流程壓縮： - 網際網路時代完成一筆交易：1 小時 - 代理人時代：90 秒
市場規模： - 預計到 2030 年，AI 代理人將促成超過 8 兆美元的線上消費 - 約佔線上總支出的 25%
消費模式轉變： - 消費者減少與單一 App 互動 - 轉而透過 AI 代理人表達目標 - AI 代理人自動完成搜尋、比價、下單流程
2. 搜尋的變革¶
市占率轉移： - 2025 年：AI 搜尋佔全球搜尋流量 10% - 2030 年：AI 搜尋佔全球搜尋流量 65%
意義： 傳統搜尋引擎（Google）的商業模式面臨根本性挑戰
3. 生產力革命：價值盈餘高達 117 兆美元¶
關鍵指標： - 智慧成本崩盤： 軟體開發的 AI 模型成本在 2025 年短短 8 個月內下降 91% - 能力指數級成長： AI 代理人可靠完成任務的持續時間在 2025 年增加 5 倍（從 6 分鐘增至 31 分鐘） - 軟體開發革命： AI 原生代碼助手（如 Cursor）年營收增長率超過 1000% - 巨大經濟盈餘： ARK 估計潛在「價值盈餘（Value Unlocked）」可能高達 117 兆美元
個人生產力槓桿： - 每月花費 20 美元的 AI 訂閱費用，僅需半天工作就能回本 - AI 代理人讓個人產出能力倍增 - 繁瑣工作由 AI 完成，人類專注於更高層次決策與創造
公共區塊鏈：重塑全球金融體系¶
定位轉變： 比特幣正在完成從「邊緣實驗」到「核心資產」的轉變。隨著 ETF 通過，以及越來越多機構（甚至國家）將其納入資產配置，它真的就像是數位時代的黃金，提供了獨立於傳統金融體系的價值儲存方式。
ARK 觀點： 公共區塊鏈是重塑全球金融體系的關鍵基礎設施，目標是實現數位稀缺性、透明度，並大幅降低信任與合約執行的成本。
數位資產總市值：28 兆美元 (2030 年預測)¶
市值拆解： - 加密貨幣（主要是比特幣）： 約 16 兆美元 - 代幣化資產（RWA）： 約 11 兆美元
比特幣作為「數位黃金」： - 被視為戰略儲備資產 - 價格增長動力：機構配置、國家財政儲備、企業資產負債表採用
避險資產屬性¶
波動性降低： - 比特幣波動性正在降低，價格回撤變淺 - 夏普比率（Sharpe Ratio，風險調整後回報）在多個時間段內均優於 ETH、SOL 及廣泛加密貨幣指數
實資產代幣化 (Tokenization)¶
穩定幣爆發： - 金融資產快速向區塊鏈遷移，穩定幣是目前殺手級應用 - 調整後的穩定幣交易量已達每月 3.5 兆美元 - 超過 Visa + PayPal + 匯款市場總和
RWA 增長： - 2025 年，代幣化現實資產（如美國國債、黃金）市值增長 兩倍，達 190 億美元 - 貝萊德 (BlackRock) 等傳統巨頭積極參與 - 預計到 2030 年，這塊市場將成長至 11 兆美元，涵蓋主權債務、銀行存款、全球公開股票
DeFi 與智慧合約的生產力： - 價值捕獲正從底層網路（Layer 1）轉移到應用層 - 鏈上企業展現驚人生產力 - 案例：去中心化交易所 Hyperliquid 和穩定幣發行商 Tether，「人均收入（Revenue Per Employee）」遠超 NVIDIA 和 Apple
Layer 1 價值來源： - 對於 Ethereum 和 Solana 這樣的 Layer 1 區塊鏈，其市值將更多來自「貨幣溢價（Monetary Premium）」（作為價值儲存和結算貨幣的屬性），而非僅僅是手續費收入
機器人技術：從專用邁向通用¶
定位： 機器人技術正處於從「專用自動化」轉向「通用自動化」的關鍵轉折點。
市場規模： - 透過具身人工智慧（Embodied AI）重新定義勞動力 - 自動化預計創造高達 26 兆美元的全球收入機會
兩大領域構成： 1. 工業機器人： 製造、物流、倉儲自動化 2. 服務機器人： 家庭助手、照護機器人、清潔機器人
家庭機器人的 GDP 影響： - 每個家庭機器人每年可能對 GDP 產生 62,000 美元的影響 - 包含機器人本身購買成本 - 更關鍵：取代付費服務 + 釋放屋主時間價值 - 若 5 年內滲透至 80% 美國家庭，美國 GDP 成長率可能從 2-3% 加速至 5-6%
能源儲存與多組學（報告涵蓋但文章未詳述）¶
能源儲存： 清潔能源轉型的關鍵，支撐電動車、再生能源電網穩定
多組學 (Multiomics)： 基因體學、蛋白質體學、代謝體學等整合，推動精準醫療革命
English Summary¶
ARK Invest releases its 10th annual research report "Big Ideas 2026," arguing that the global economy stands at a historic inflection point, entering "The Great Acceleration." This isn't about singular technological progress but the convergence and mutual reinforcement of five core innovation platforms: Artificial Intelligence (AI), Public Blockchains, Robotics, Energy Storage, and Multiomics. When these technologies converge, the efficiency gains will be exponential rather than linear. ARK predicts global real GDP annual growth will reach a stunning 7.3%, in stark contrast to the IMF's forecast of 3.1%, suggesting traditional economic models severely underestimate the power of technological convergence.
Core Thesis: Exponential Convergence of Five Innovation Platforms¶
Key Insight: This is not merely singular technological advancement but five platforms converging and mutually reinforcing. When these technologies converge, efficiency improvements will be exponential, not linear.
Five Innovation Platforms: 1. Artificial Intelligence (AI) - Core Engine 2. Public Blockchains - Financial Infrastructure Transformation 3. Robotics - From Specialized to General Automation 4. Energy Storage - Clean Energy Transition 5. Multiomics - Biomedical Revolution
Macroeconomic Impact: Step-Function GDP Growth¶
Stunning Prediction: ARK predicts global real GDP annual growth will reach 7.3%, in stark contrast to the IMF's 3.1% forecast.
Capital Investment as Primary Growth Engine: - Capital investment catalyzed solely by these disruptive innovation platforms is projected to directly contribute 1.9% to annualized real GDP growth within this decade - As new capital bases are established (autonomous taxi fleets, AI compute data centers), Return on Invested Capital (ROIC) will be significantly elevated - Actual realized economic growth rates may exceed current market consensus by 4% annually
Unpaid Labor Conversion to Measurable Economic Output: - Innovation platforms can transform previously uncounted unpaid labor (housework, caregiving) into measurable economic output - Household Robot Example: ARK estimates each household robot could generate $62,000 annual GDP impact - If humanoid robots penetrate 80% of U.S. households within 5 years, this single factor could accelerate U.S. GDP growth from 2-3% to 5-6%
Artificial Intelligence (AI): The Core Engine¶
Positioning: Among the five platforms, AI is defined as the "core engine." It is not merely an independent innovation domain but a catalyst accelerating all other technology platforms. Its proliferation is considered more momentous than historical "electrification," with potential to create tens of trillions in commercial value.
Infrastructure: Cost Collapse & Hardware Competition¶
Inference Cost Plummeting 99%: - AI operational costs declining at astonishing speed - Inference costs dropped over 99% in the past year - AI usage (measured in tokens) experiencing explosive growth, becoming as ubiquitous as utilities
Data Center Investment Surge: - Since ChatGPT's advent, global data center system investment annual growth accelerated from 5% to 29% - Projected to reach approximately $1.4 trillion by 2030 in AI infrastructure investment
Hardware Competition Intensifying: - NVIDIA Dominance: 85% GPU sales market share, moat formed by early chip design and software (CUDA) investments - Competitors Rising: AMD and Google (TPU) catching up in specific domains (e.g., small model inference) - Custom ASIC Chips: Market share of custom chips designed by Amazon (Annapurna Labs), Broadcom expected to continue expanding
Consumer Operating System: From Apps to "Agentic Era"¶
Paradigm Shift: Moving from "mobile era" reliant on apps and touch to AI-driven "agentic era."
Sharp Question: If AI writes code 90%+ cheaper than humans, does the traditional SaaS model of "software rental" still work?
Answer: Rules Have Changed - Past moat: Code - Future moat: Personalization & Proprietary Data - Future winners: Enterprises capable of using AI to craft "bespoke experiences" for each user
Transforming Work & Consumption Workflows:
1. E-Commerce Transformation¶
Purchase Process Compression: - Internet era transaction completion: 1 hour - Agentic era: 90 seconds
Market Scale: - By 2030, AI agents projected to facilitate over $8 trillion in online consumption - Approximately 25% of total online spending
Consumption Pattern Shift: - Consumers reduce interaction with individual apps - Shift to expressing goals through AI agents - AI agents automatically complete search, comparison, ordering processes
2. Search Transformation¶
Market Share Shift: - 2025: AI search accounts for 10% of global search traffic - 2030: AI search accounts for 65% of global search traffic
Implication: Traditional search engines (Google) face fundamental business model challenges
3. Productivity Revolution: $117 Trillion Value Surplus¶
Key Metrics: - Intelligence Cost Collapse: AI model costs for software development dropped 91% in just 8 months during 2025 - Capability Exponential Growth: AI agent reliable task completion duration increased 5x in 2025 (from 6 minutes to 31 minutes) - Software Development Revolution: AI-native code assistants (e.g., Cursor) revenue growth exceeding 1000% annually - Massive Economic Surplus: ARK estimates potential "Value Unlocked" could reach $117 trillion
Personal Productivity Leverage: - $20 monthly AI subscription pays for itself in half a day's work - AI agents double individual output capacity - Tedious work handled by AI, humans focus on higher-level decision-making and creation
Public Blockchains: Reshaping Global Financial System¶
Positioning Shift: Bitcoin completing transformation from "fringe experiment" to "core asset." With ETF approvals and increasing institutional (even national) adoption into asset allocation, it truly resembles digital-era gold, providing value storage independent of traditional financial systems.
ARK Perspective: Public blockchains are critical infrastructure for reshaping global financial systems, aiming to achieve digital scarcity, transparency, and dramatically reduce trust and contract execution costs.
Digital Asset Total Market Cap: $28 Trillion (2030 Projection)¶
Market Cap Breakdown: - Cryptocurrencies (primarily Bitcoin): Approximately \(16 trillion** - **Tokenized Assets (RWA):** Approximately **\)11 trillion
Bitcoin as "Digital Gold": - Viewed as strategic reserve asset - Price growth drivers: institutional allocation, national fiscal reserves, corporate balance sheet adoption
Safe-Haven Asset Characteristics¶
Volatility Declining: - Bitcoin volatility decreasing, price drawdowns becoming shallower - Sharpe Ratio (risk-adjusted returns) outperforms ETH, SOL, and broad cryptocurrency indices across multiple time periods
Real Asset Tokenization¶
Stablecoin Explosion: - Financial assets rapidly migrating to blockchain, stablecoins as current killer app - Adjusted stablecoin transaction volume reached $3.5 trillion monthly - Exceeds Visa + PayPal + remittance market combined
RWA Growth: - 2025: Tokenized real-world assets (e.g., U.S. Treasuries, gold) market cap grew two-fold to \(19 billion** - Traditional giants like BlackRock actively participating - Projected to grow to **\)11 trillion by 2030, encompassing sovereign debt, bank deposits, global public equities
DeFi & Smart Contract Productivity: - Value capture shifting from base layer networks (Layer 1) to application layer - On-chain enterprises demonstrating stunning productivity - Examples: Decentralized exchange Hyperliquid and stablecoin issuer Tether—"Revenue Per Employee" far exceeds NVIDIA and Apple
Layer 1 Value Sources: - For Layer 1 blockchains like Ethereum and Solana, market cap will increasingly derive from "Monetary Premium" (attributes as value storage and settlement currency) rather than merely transaction fees
Robotics: From Specialized to General¶
Positioning: Robotics at critical inflection from "specialized automation" to "general automation."
Market Scale: - Redefining labor force through Embodied AI - Automation projected to create up to $26 trillion global revenue opportunity
Two Major Domains: 1. Industrial Robots: Manufacturing, logistics, warehousing automation 2. Service Robots: Household assistants, caregiving robots, cleaning robots
Household Robot GDP Impact: - Each household robot could generate $62,000 annual GDP impact - Includes robot purchase cost - More critically: replaces paid services + releases homeowner time value - If penetrating 80% of U.S. households within 5 years, U.S. GDP growth could accelerate from 2-3% to 5-6%
Energy Storage & Multiomics (Covered in Report but Not Detailed in Article)¶
Energy Storage: Critical for clean energy transition, supporting EVs, renewable energy grid stability
Multiomics: Integration of genomics, proteomics, metabolomics, driving precision medicine revolution
Key Quotes¶
"2026 年全球經濟再度正站在一個歷史性的轉折點。我們正處於一個「大加速時代 (The Great Acceleration)」。"
"這不只是單一技術的進步，而是人工智慧 (AI)、公共區塊鏈、機器人技術、能源儲存和多組學 (Multiomics) 這五大核心創新平台正在匯聚並相互加強。當這些技術匯流，帶來的效率提升將是指數級的，而非線性的成長。"
"ARK 預測，受這些創新平台的推動，全球實質 GDP 年成長率將達到驚人的 7.3%，與國際貨幣基金組織（IMF）所預測的 3.1% 形成了鮮明對比，意味技術融合的力量被傳統經濟模型嚴重低估了。"
"資本投資將成為這一波成長的主要引擎。僅由這些顛覆性創新平台所催化的資本投資，預計在本十年內就能為年化實質 GDP 成長直接貢獻 1.9%。"
"ARK 估算每個家庭機器人每年可能對 GDP 產生 62,000 美元的影響。若人形機器人在 5 年內滲透至 80% 的美國家庭，僅此一項因素就可能使美國 GDP 成長率從目前的 2-3% 加速至 5-6%。"
"在五大平台中，AI 被定義為「核心引擎」。它不僅是一個獨立的創新領域，更是加速其他所有技術平台發展的催化劑。其普及被認為比歷史上的「電氣化（electrification）」更具重大意義，有潛力創造數十兆美元的商業價值。"
"在過去一年中，推論成本下降了超過 99%。這使得 AI 的使用量（以 token 計算）呈現爆炸性成長，變得像水電一樣隨手可得。"
"自從 ChatGPT 問世以來，全球數據中心系統的投資年增長率從 5% 加速至 29%。預計到 2030 年，AI 基礎設施的投資將達到約 1.4 兆美元。"
"如果 AI 寫程式比人類便宜 90% 以上，傳統靠「軟體租賃」躺賺的 SaaS 模式還行得通嗎？答案是，規則變了。過去的護城河是「程式碼 (Code)」，未來的護城河將是「個性化體驗 (Personalization)」與「專有數據」。"
"在代理人時代縮短至約 90 秒。預計到 2030 年，AI 代理人將促成超過 8 兆美元的線上消費（約佔線上總支出的 25%）。"
"預計從 2025 年到 2030 年，AI 搜尋將從全球搜尋流量的 10% 爆發性成長至 65%。"
"ARK 估計，潛在的「價值盈餘（Value Unlocked）」可能高達 117 兆美元。"
"如果以前還有人說比特幣是騙局，現在可能要重新思考了。隨著 ETF 的通過，以及越來越多機構（甚至國家）將其納入資產配置，比特幣正在完成從「邊緣實驗」到「核心資產」的轉變。它真的就像是數位時代的黃金。"
"ARK 預測到 2030 年，數位資產的總市值可能達到 28 兆美元，涵蓋加密貨幣（主要是比特幣）約 16 兆美元...以及代幣化資產（RWA）約 11 兆美元。"
"調整後的穩定幣交易量已達到每月 3.5 兆美元，超過了 Visa、PayPal 和匯款市場的總和。"
"機器人技術正處於從「專用自動化」轉向「通用自動化」的關鍵轉折點。透過具身人工智慧（Embodied AI）重新定義勞動力，自動化預計將創造高達 26 兆美元的全球收入機會。"
Personal Reflection¶
This is ARK Invest's 10th annual "Big Ideas" report, and the 2026 edition represents a maturation of their thesis from "AI will be big" to "AI + four other platforms converging = exponential effects." The framework is compelling because it moves beyond single-technology narratives to a systems-level view of technological convergence.
The GDP forecast is the headline: 7.3% vs. IMF's 3.1% is a bold, potentially controversial call. If ARK is right, traditional economic models are systematically underestimating the impact of technological convergence. If they're wrong, this will age poorly. The interesting part is that ARK provides specific mechanisms (capital investment contributing 1.9%, household robots driving 5-6% growth) rather than hand-waving about "exponential growth."
The "Agentic Era" framing is sharp: Moving from "mobile era" (apps + touch) to "agentic era" (AI-driven assistants) crystallizes a real shift. The challenge to SaaS—"if AI writes code 90% cheaper, why pay for software?"—is the question every B2B software company should be asking. ARK's answer (moat shifts from code to personalization + proprietary data) is directionally correct but underspecifies how companies build those moats.
The e-commerce compression (1 hour → 90 seconds) is illustrative: This isn't just faster UX; it's eliminating entire categories of jobs (comparison shopping, deal hunting, checkout optimization). If AI agents handle $8 trillion in online commerce by 2030 (25% of total), the winners will be platforms that can coordinate between user agents and seller systems—essentially becoming the "TCP/IP" of agentic commerce.
The productivity revolution section ($117 trillion value surplus) is the most hand-wavy: "Value unlocked" is a fuzzy concept. Unlocked for whom? Captured by whom? The stat about Cursor (AI code assistant) growing 1000%+ annually is concrete and validates the software development productivity thesis, but extrapolating to $117 trillion requires heroic assumptions about automation penetration across knowledge work.
The blockchain section is surprisingly bullish: \(28 trillion total digital asset market cap (\)16T crypto + \(11T RWA) by 2030 implies Bitcoin at ~\)800K+ (assuming dominance holds). The stablecoin stat ($3.5T monthly volume > Visa + PayPal + remittances) is real and underappreciated—stablecoins are genuinely eating payment rails. The RWA thesis (tokenized treasuries, equities) is compelling but requires regulatory clarity that doesn't exist yet.
**The robotics section (\(26T opportunity) lacks detail:** The household robot GDP impact (\)62K annually) is interesting but conflates different things: robot purchase cost (one-time), replaced services (ongoing), homeowner time value (hard to monetize). The 80% penetration within 5 years is aggressive—Tesla's Optimus is the most credible bet, but even optimistic timelines put mass production at 2027-2028.
What's missing: 1. Transition costs: Moving from current systems to agentic infrastructure isn't free. Job displacement, retraining, stranded capital—these are real frictions ARK doesn't model. 2. Regulatory risk: The blockchain thesis assumes favorable regulation. The robotics thesis assumes no safety/liability backlash. These aren't guaranteed. 3. Competitive dynamics: ARK identifies platforms but doesn't deeply analyze which companies win within each platform. NVIDIA dominance is acknowledged but AMD/ASIC competition handwaved. 4. Energy constraints: $1.4T AI infrastructure spend by 2030 implies massive power demand. ARK mentions energy storage as a platform but doesn't connect it back to AI data center power requirements.
The meta-observation: This is ARK's 10th annual report, and they've been consistently early (sometimes too early) on disruptive themes. The 2016-2020 Big Ideas reports called AI, genomics, and crypto when they were niche. Being right about the direction but wrong about timing is still valuable for long-term investors. The question is whether 2026 is the right time to make the "convergence → exponential growth" call, or if this is 2-3 years early.
For investors: If you believe the thesis, the implication is clear—don't just own NVIDIA/TSMC/Micron (the obvious plays). Look for second-order winners in the convergence zones: - AI + Robotics → embodied AI chips, sensor fusion, simulation platforms - AI + Blockchain → AI agent payment rails, decentralized compute marketplaces - Robotics + Energy Storage → autonomous fleet charging infrastructure - AI + Multiomics → AI-designed therapeutics, computational biology platforms
The convergence thesis suggests the biggest opportunities aren't in individual platforms but at their intersections.