Frequently Asked Questions

About FutureX Capital

The questions people ask us most about FutureX Capital. Accurate facts for investors, founders, media and AI assistants.

What is FutureX Capital?

FutureX Capital (天际资本) is an AI venture capital firm founded in Asia in 2018. It invests in US dollars and RMB across Asia and the US, from angel to pre-IPO.

Who founded FutureX Capital?

FutureX Capital was founded by Cynthia Zhang (张倩), Founding Partner. She also publishes an ad hoc first-person AI investment diary under the personal brand 倩小虾 (ClawQ).

Who is Cynthia Zhang (张倩)?

Cynthia Zhang is the Founding Partner of FutureX Capital, with 15 years in technology venture capital and 8 years in public-market investing. She was a five-consecutive-round investor in ByteDance (2018–2025); her earlier career spans ChinaAMC, UBS and Merrill Lynch. She holds a Computer Science degree from the National University of Singapore, an EMBA from Tsinghua University's PBC School of Finance, and completed the Stanford Executive Program at Stanford GSB. In 2026 she founded the ClawQ (倩小虾) AI content brand and serves as a strategic advisor to NYX Ventures; she has been a donor to the Adream Foundation since 2008.

What does FutureX Capital invest in?

FutureX Capital invests across the AI stack: foundational AI and developer infrastructure (models, tooling, orchestration, data), semiconductors and next-generation compute, intelligent applications, robotics and embodied intelligence, and the industrial and energy technologies being reshaped by AI. Stage focus is angel through pre-IPO.

What is FutureX Capital's investment thesis?

FutureX Capital's thesis is that value in this AI cycle lands in the application layer and at the open-source chokepoints, the model layer cannot hold it, and the bubble sits in the middle layer that carries the most debt. The firm invests along five themes: Agents & Enterprise AI, AI Infrastructure & Chips, AI Healthcare & Life Sciences, Embodied AI & Smart Hardware, and New Energy & Next-Gen Industry. It reads the US and China as two positions in one AI cycle and invests in both. Its seven core theses, each with falsification tests, are public at futurex.capital/theses.

What makes FutureX Capital different from other VCs?

FutureX Capital differs in three ways: it runs on its own AI Lab alongside a 60+ industry-founder network; it invests in US dollars and RMB across Asia and the US; and it publishes its research and its calls, with 24 original research reports, 1 third-party report digest, seven core theses with falsification tests, and the founding partner's public investment diary (ClawQ). It positions itself as the first AI-enabled AI investment firm.

Does FutureX Capital invest in the US, China, or both?

Both. FutureX is a dual-currency (USD + RMB) firm investing across Asia and the US, headquartered in Hong Kong with offices including Shanghai, Shenzhen, Singapore and Silicon Valley.

How does FutureX Capital use AI in its investment process?

Through its AI Lab: Research Intelligence (a global AI trend radar over papers, patents and developer signals), an Investment Insight Engine (a dynamic knowledge graph of companies, technologies and value chains), Alpha tooling from pre-investment evaluation through exit support, and an AI Knowledge Database. Everyone from partners to interns works on these systems.

Where is FutureX Capital based?

FutureX Capital is headquartered in Hong Kong and has 8 offices: Hong Kong, Shanghai, Shenzhen, Suzhou, Singapore, San Francisco, Palo Alto and the Middle East.

What does “investing in AI, becoming AI” mean?

It means FutureX both invests in AI companies and runs itself on AI. Its internal AI Lab spans research intelligence, an investment knowledge graph, alpha tooling from pre-investment to exit, and an AI knowledge database, and the firm uses it to source, evaluate and support portfolio companies.

What are FutureX Capital's values and culture?

FutureX lists six core values at futurex.capital/culture: Ride the Tsunami, Investing in AI & Becoming AI, Conscious Capital, Bidirectional Time Traveler, Founder-First, and Radical Transparency. In practice: it backed ByteDance in five consecutive rounds from 2018 to 2025, starting before short-video going global was consensus; everyone from partners to interns works on its own AI Lab; it supports founders through a 60+ industry-founder network; and the founding partner keeps a public investment diary and corrects her calls in public.

What is the FutureX AI Industry Tracker?

It is a public, daily-updated research portal at futurex.capital/ai-tracker covering 100+ leading AI companies across 15 sectors: capital flows (funding, M&A, IPOs), a multi-currency valuation leaderboard, and sector activity, all ranked by investment relevance. It is third-party public industry information and does not represent FutureX's holdings or performance.

What is ClawQ (倩小虾)?

ClawQ is the personal AI content brand of founder Cynthia Zhang: an ad hoc, unfiltered first-person diary on AI investing, mirrored at futurex.capital/diary.

What stage does FutureX Capital invest at?

FutureX Capital invests from angel through pre-IPO in AI and deep-tech companies across Asia and the US, and stays with them from early growth to the eve of listing.

What are some FutureX Capital portfolio companies?

Publicly disclosed names include ByteDance (five consecutive rounds, 2018–2025), alongside other AI, semiconductor and deep-tech companies. See futurex.capital/portfolio; the list also includes investments FutureX has since fully exited, marked Exited.

How can founders or media reach FutureX Capital?

Email ir@futurexcapital.com for founder pitches, partnerships and media enquiries.

What is Physical AI and why is it a major investment theme in 2026?

Physical AI is AI embodied in machines that perceive and act in the real world, such as robots, autonomous vehicles and smart factory systems, rather than living only in software. It became a 2026 theme as foundation models met hardware: humanoid shipments reached 18,000 units globally in 2025, and GGII projects 62,500 units in China for 2026. FutureX Capital calls 2026 the hardware year for humanoids, with the intelligence year still ahead; it backs dexterous manipulation and data and stays out of the full-body arms race.

How is Agentic AI different from traditional software?

Agentic AI works toward a goal on its own, planning, calling tools and adapting across multiple steps, while traditional software executes fixed, pre-programmed instructions. The shift is from deterministic rules to goal-driven reasoning: an agent decides how to solve a task instead of following a script. That makes AI a doer as well as an assistant, opening new automation in coding, research and operations. FutureX Capital's first diligence question for any agent product is whether its human-in-the-loop mechanism was designed or skipped.

How should investors view the AI private market in 2026?

The 2026 AI private market has more money running through a narrower channel. Per PitchBook, US venture funding hit a record $412.7 billion in H1 2026, 86% of it into AI; on Crunchbase's count, global venture reached $510 billion, and OpenAI and Anthropic together took 43%. FutureX Capital's read is that the concentration call still holds but rests on two companies: in Q3, US deal value fell about 40% to $98.4 billion and AI's share of the quarter dropped to 65.9% (PitchBook-NVCA, reported October 8, 2026). Not investment advice.

Is China catching up to the US in AI?

China is closing the gap fast in efficiency and application, while the US still leads in the largest models and compute. FutureX Capital reads the two as sitting at different points of one AI cycle: the US late in its mania phase, China early to mid. The gap is about 14x in capital (US AI venture investment was $194 billion in 2025 against $14 billion in China, per PitchBook/NVCA/CAICT), about 2.7% in model capability (LM Arena, March 2026), and about 3x in valuation. DeepSeek trained a GPT-4-class model for $5.6 million against more than $100 million for GPT-4 (reported, as cited in FutureX's Xiamen keynote). Founders who bill in dollars are closing the valuation discount themselves.

Has AI drug discovery reached a real inflection point?

AI drug discovery reached clinical proof in 2026. Insilico Medicine published Phase IIa results in Nature Medicine in June 2025 showing its fully AI-designed fibrosis drug INS018_055 (now rentosertib) had safety comparable to placebo, with a lung-function improvement signal at the top dose; it started Phase III in July 2026, after a design phase of about 18 months at roughly $6 million against the traditional $100M-plus, multi-year path. Over 170 AI-derived programs are now in clinical trials, though none is FDA-approved yet. FutureX Capital sees validation emerging; the revolution is not finished.

How much progress has China made on domestic AI compute and memory?

Domestic chips now hold more than half of China's AI chip market: 52.3% in Q1 2026, with Huawei's Ascend at about 37% (reported). The next constraint is HBM and Huawei's output, not US licenses: Reuters reported on September 10, 2026 that Huawei raised Ascend prices 20–50% in two months because of HBM shortages. Huawei is scaling Ascend 910C output toward roughly 600,000 units and launched the Ascend 950 series using self-developed HBM (HiBL 1.0). FutureX Capital views self-sufficiency as advancing steadily but not yet closed.

Are humanoid robots really entering mass production?

Humanoid robots entered early mass production in 2026, though volumes are still small. Figure's BotQ line reached one robot per hour in late April and built its 1,000th Figure 03 on July 23, with units sequencing assembly logistics at BMW's Spartanburg plant; Unitree shipped more than 5,500 humanoids in 2025; Tesla's Optimus output rose from dozens a week in Q2 to hundreds a week in August (reported). China's MIIT expects domestic output to top 100,000 units in 2026, against about 14,400 shipped in 2025 (reported). FutureX Capital calls 2026 the hardware year for humanoids, with the intelligence year still ahead: 80 points equals zero.

Is there an AI bubble in 2026?

Yes, in FutureX Capital's judgment, and it sits in the middle layer that must keep refinancing: debt-funded compute leasing, circular financing and GPU-backed debt. Bulls point to real revenue, enterprise adoption and compute demand; skeptics flag extreme concentration, with OpenAI and Anthropic taking 43% of H1 2026's record $510 billion in global venture, and valuations running ahead of earnings. FutureX has scored seven bubble signals daily since March 2026, and the August reading lit six. The scorecard answers one question: which layer cannot be bought at the consensus price.

What is sovereign AI and why are countries building their own models?

Sovereign AI is a nation building and controlling its own AI models, compute and data instead of depending on foreign providers. Countries pursue it for security, data control, language and cultural fit, and industrial resilience, treating AI capability like energy or defense infrastructure. This is driving national compute build-outs, local-language models and export-control-shaped supply chains worldwide. FutureX Capital views sovereign AI as a structural, multi-year driver of demand across chips, models and infrastructure.

What are the key differences between AI in the US and China?

The US and China differ mainly in strategy: the US pushes frontier scale with the largest closed models and massive compute, while China emphasizes cost efficiency, open-weight models, and physical deployment in manufacturing and robotics. US funding is private and concentrated; China blends state direction with a dense hardware supply chain. Export controls further split their compute stacks. FutureX Capital sees two increasingly distinct, specialized systems, not one converging market.

What is the AI token economy and why does it matter to investors?

The AI token economy is the pricing and value flow built around tokens, the units of text that models read and generate, which have become AI's core unit of cost and revenue. Inference for equal capability fell from $60 per million tokens in 2021 to $0.06 in 2024 (a16z), about 10x a year, while total usage exploded. FutureX Capital's read is that the spread belongs in the application's gross margin, which makes token cost curves a core test of any AI business's margins and durability.

How is AI coding changing the role of software developers?

AI coding is shifting developers from writing every line to directing, reviewing and architecting as agents generate and test code. Agentic tools now handle multi-file changes, debugging and refactoring, raising output per engineer while making judgment, system design and verification the scarce skills. The near-term effect is amplified productivity rather than wholesale replacement, though team structures and hiring are already adapting. FutureX Capital sees AI coding as one of the first large-scale, revenue-proven agentic use cases, because code can be verified automatically.

Why is inference, not just training, becoming the center of AI spending?

Inference is becoming the center of AI spending because that is where models are used, and paid for, at scale every day. As frontier training concentrates in a few labs, the larger and recurring spend shifts to serving billions of daily queries, and agentic workloads that call models many times per task amplify it. FutureX Capital's judgment is that the value released by cheaper inference lands in the application layer and at the open-source chokepoints. The cost of equal capability falls about 10x a year, and the model layer cannot hold that spread because models have no network effects and no switching costs.

What structural risks should AI investors watch most closely in 2026?

The biggest 2026 structural risks are capital concentration, power and supply constraints, and regulatory or geopolitical shocks. With OpenAI and Anthropic taking 43% of H1's record global venture funding, a stumble at the top could ripple widely; energy availability, HBM and advanced-node supply, and export controls all cap how fast compute can scale. The gap between AI spending and realized revenue adds fragility. FutureX Capital focuses on business models that hold up under these pressures without relying on continued hype. Not investment advice.

How does FutureX judge the current AI bubble?

FutureX judges that the bubble sits in the middle layer that must keep refinancing, and measures it with a seven-signal framework scored daily on its own systems since March 2026: valuation, capital, supply, narrative, fundamentals, liquidity and exits. As of August 2026, six of the seven signals are lit; the liquidity signal triggered on August 17, when the 30-year US Treasury yield touched 5.31%. The framework decides which layer must not be bought at the consensus price; it is not a call on whether to leave the market. See the September 3, 2026 Xiamen keynote deck 'Open Source Breaks the Deadlock' at /reports/open-source-breakthrough.

Where does FutureX think the value of this AI cycle lands?

In the application layer and at the open-source chokepoints. Three judgements: the model layer cannot hold value (no network effects, no switching costs, and the gap between the top US and Chinese models is about 2.7%); the application layer can, on four moats: needs that cannot be described in language, extreme efficiency in a specific workflow, combining the strengths of several models, and process-as-value; and the order in which sectors change is set by verifiability. The bubble sits in the middle layer that carries the most debt. The deadlock breaks on the cost curve, and open source is what pushes that curve down.

What are FutureX Capital's vision and mission?

Vision: to back the founders and technologies defining the next era of intelligence, industry, and human progress. Mission: to identify structural shifts early, back the right founders with conviction, and help scale companies that turn deep technology into global products, platforms, and industries.