Hello, 你好, 안녕하세요, नमस्ते, مرحبا, Hola, Bonjour.
Have you figured out how to own your share of the AI industry yet?
你想好如何拥有属于你的那份人工智能产业了吗？
AI 산업에서 당신의 몫을 소유하는 방법을 이미 알아내셨나요?
क्या आपने यह पता लगा लिया है कि AI उद्योग में अपना हिस्सा कैसे हासिल करें?
هل اكتشفت بعد كيف تمتلك حصتك في صناعة الذكاء الاصطناعي؟
¿Ya has descubierto cómo poseer tu parte de la industria de la inteligencia artificial?
Avez-vous déjà trouvé comment posséder votre part de l'industrie de l'IA?
We’ve all seen the incredible appreciation of companies participating in the AI buildout. But we haven’t all been able to invest.
Some of us had trouble getting a brokerage account with access to US equities. Others just didn’t have the time to research hundreds of companies and figure out which ones were relevant.
We decided to knock down those barriers.
Reserve has turned the AI supply chain trade into five decentralized token funds, each for a unique layer of the supply chain:
- $BUILDOUT — AI infrastructure overall: the whole stack
- $POWER — AI power: the electricity it runs on
- $PHOTON — AI photonics: the light layer that moves the data
- $NEOCLOUD — AI capacity: the companies that rent compute as a service
- $ROBOTS — Robotics: AI coming into the physical world
Each is a single, publicly investable token on @BNBCHAIN, built from US-listed companies and held as @OndoFinance Global Markets tokenized stocks.
Because they’re onchain, we can bring them online faster than traditional finance: a US ETF takes around 75 days from registration to live trading, while an onchain fund like these can go live just a few days after the assets are tokenized - Ondo only just released most of these stocks on June 17th! And of course each of the above trades 24/7 and is redeemable onchain for the assets inside it any time.
These products are available in about 145 countries for about 6 billion people, not including the US. Why launch bundles of US stocks and exclude US buyers? Regulatory constraints don’t (yet) allow such products here - we hope to serve a US audience in the future. But the entire world is interested in owning its share of the AI revolution, so we thought it prudent to begin with this global offering as soon as possible.
This new line of DTFs covers every layer of the physical stack that’s being built out to enable the next generations of AI. Why cover physical infrastructure rather than the software companies?
AI is much, much more than a software story.
Before any model can answer a prompt, someone has to fabricate the chips, build the datacenters, generate the power, and lay the cables.
It’s one of the largest industrial buildouts in history, and a big slice of it runs through several dozen public US-traded companies, from mega-cap semiconductor producers to obscure food additives companies.
Let’s break them all down.
BUILDOUT
The context: What we call “AI” is actually a complex physical stack of chips, racks, cooling, power generation, and networking. Some people want to go deep on which layer of the stack is the next bottleneck in an attempt to frontrun the industry and get the highest return. Others just want to buy the whole stack and enjoy broad exposure. Looking at the historical backtest of this category, you can see this wouldn’t have been too bad a strategy since the start of 2026.
The numbers: World Semiconductor Trade Statistics put global semiconductor sales at roughly $796 billion in 2025, climbing to ~$1.5 trillion in 2026 in a memory-led surge as AI demand outruns supply.
All that spending flows through a concentrated set of mostly public companies: the GPU and specialized chip designers, the companies that actually fabricate the chips, the memory makers, and the equipment giants whose machines make advanced silicon possible at all.
The Reserve AI Infrastructure DTF ($BUILDOUT) is built around these companies: the 25 largest US-listed names across the AI hardware stack. One token gives you exposure to a weighted basket spanning the entire AI infrastructure stack: from AI chip designers NVIDIA ($NVDA, AMD ($AMD and Broadcom ($AVGO, to the main factory that produces them all TSMC ($TSM and memory maker Micron ($MU, down to the equipment makers: ASML ($ASML, Applied Materials ($AMAT, Lam Research ($LRCX, whose machines every advanced chip depends on.
You can find more info and trade $BUILDOUT on the Reserve website here: https://app.reserve.org/bsc/index-dtf/buildout/overview
POWER
The context: Where there’s smoke, there’s fire… and where there’s AI, there’s electricity. A single AI data center can draw as much power as a major city, and the grid wasn’t built for hundreds of them. Power is foundational. That’s what makes it one of the most interesting bottlenecks in the AI buildout.
The numbers: The demand curve is close to vertical. Goldman Sachs Research estimates US data-center power demand will roughly triple — from about 32 gigawatts in 2025 to somewhere around 95 gigawatts by 2030. To keep up, hyperscalers are signing power deals directly with nuclear operators and racing to bring new generation online — GE Vernova’s gas-turbine backlog alone has swelled to roughly 100 GW.
Meeting that demand takes a full stack: the companies that generate the power (gas turbines, nuclear, fuel cells), the ones that move and condition it (grid and electrical equipment), and the power chips that deliver it cleanly to the racks.
The Reserve AI Power DTF ($POWER) is built around these companies. One token gives you exposure to a weighted basket of US-listed power names behind the buildout: from gas-turbine and grid maker GE Vernova (GEV), to electrical and cooling leaders Eaton (ETN) and Vertiv (VRT), to nuclear producer Constellation (CEG) and power-chip makers like onsemi (ON).
You can find more info and trade $POWER on the Reserve website here: https://app.reserve.org/bsc/index-dtf/power/overview
PHOTON
The context: Imagine if Paul Revere could’ve just dropped “The British are coming!” in the group chat. Alas, information moves at the speed of its delivery mechanism. For decades, data inside servers moved as electricity through copper wire. But today’s AI chips are advancing so quickly that copper can’t keep up: it generates too much heat, carries too little bandwidth, and can’t reliably move data far enough. The solution is light: data centers are increasingly using optical connections to move information between chips, replacing electrons with photons. This technology, called photonics, radically increases the amount of optical hardware packed into every AI system.
The numbers: Goldman Sachs estimates spending on optical components for AI data centers will grow from $15 billion in 2026 to $154 billion by 2028, or roughly a 10x jump in two years. And the biggest player, NVIDIA, is already moving: in March 2026, they committed $2 billion each to Coherent and Lumentum, two of the largest makers of the laser chips this entire buildout runs on.
Photonics is still a young industry, and its supply chain sits in a small handful of public companies: the ones making the fiber, the lasers, the transceivers, the photonics chips, and the wafers those lasers are grown on.
The Reserve AI Photonics DTF ($PHOTON) is built around these companies. One token gives you exposure to a weighted basket of US-listed photonics names spanning the full optical supply chain: from Corning ($GLW, the company that invented low-loss optical fiber, to laser leaders Coherent ($COHR and Lumentum ($LITE, down to the specialty chips and wafers the whole industry depends on.
You can find more info and trade $PHOTON on the Reserve website here: https://app.reserve.org/bsc/index-dtf/photon/overview
NEOCLOUD
The context: If you need somewhere to crash for a night or two, you don’t buy a hotel. Likewise, if you need AI compute, you probably don’t need to build a whole data center. Building one costs billions, so most companies that need serious compute don’t build AI data centers… they simply rent the computing power. A cottage industry has sprung up to meet that demand: “neoclouds” and power-rich data-center operators that raise the capital, secure the power, and rent out GPU compute by the hour.
The numbers: Independent researchers size the neocloud / GPU-cloud market at about $35 billion a year in 2026, growing past $180 billion a year by 2030 — roughly 5x (Mordor Intelligence puts 2026 near $35B; Synergy Research Group is cited reaching ~$180B by 2030). And the backlogs are already enormous: CoreWeave alone reports roughly $99 billion in contracted backlog, and Nebius around $50 billion from the likes of Microsoft and Meta.
The players split into two camps: the pure-play AI cloud operators, and the power-rich hosts — many of them former miners — converting energized sites into GPU and AI-data-center capacity.
The Reserve AI Capacity Neocloud DTF ($NEOCLOUD) is built around these companies. One token gives you exposure to a weighted basket of US-listed names turning capital, silicon and electricity into compute: from neocloud operators Nebius (NBIS) and CoreWeave (CRWV), to operator-host IREN (IREN), to power-rich hosts TeraWulf (WULF), Hut 8 (HUT) and Applied Digital (APLD).
You can find more info and trade $NEOCLOUD on the Reserve website here: https://app.reserve.org/bsc/index-dtf/neocloud/overview
ROBOTS
The context: You blinked, and science fiction became science fact: robots are real now. The same breakthroughs that taught software to reason are teaching machines to perceive, decide, and act, and even to take on physical labor. Robotics is where the innovations of AI transcend the chat box and become an embodied reality.
The numbers: In the bookends of a speculative forecast, Morgan Stanley estimates the humanoid-robot market could reach something like $5 trillion by 2050, with around a billion robots in service. On the other, more conservative end, Goldman Sachs puts the number at about $38 billion by 2035. Either way, capital is already following: industry trackers reported over $34 billion flowing into robotics in 2025.
And it spans far more than humanoids: we’re talking surgical robots, factory and warehouse automation, self-driving systems, and the lidar, machine-vision, and edge chips that help machines navigate the real world.
The Reserve Robotics DTF ($ROBOTS) is built around these companies. One token gives you exposure to a weighted basket of US-listed robotics and automation names: from robotic-surgery leader Intuitive Surgical ($ISRG, to factory-automation Rockwell ($ROK and warehouse robotics Symbotic ($SYM, down to autonomy and sensor names like Mobileye ($MBLY and Ouster ($OUST.
You can find more info and trade $ROBOTS on the Reserve website here: https://app.reserve.org/bsc/index-dtf/robots/overview
How do DTFs work under the hood?
All five DTFs live on @BNBCHAIN. You can buy and sell them onchain at app.reserve.org and @BitgetWallet, or on DEXes like @PancakeSwap X and @CoWSwap — 24/7, with no minimum or maximum purchase size. A 0.3% minting fee and a 0.6% TVL fee apply. Eligibility follows Ondo’s geographic rules: it’s not available in the US or sanctioned jurisdictions, some countries require accreditation or equivalent, and elsewhere you can connect a wallet and buy. You’re responsible for following your local laws.
They’re powered by Reserve, the onchain platform for building decentralized token funds of all kinds: think the blockchain version of ETFs (though a DTF is not an ETF, and is not regulated like one). Reserve got started in 2018 with backing from Sam Altman and Peter Thiel, and has spent millions on code audits. As with any DeFi app, you use it at your own risk.
What’s the role of RSR?
Reserve, the protocol powering all five DTFs, is governed by the Reserve Rights token ($RSR, and RSR is wired directly into how the DTFs work.
- Lock to govern. Users can lock RSR to receive vlRSR (vote-locked RSR) and help govern the DTFs: which companies qualify for a basket, how it’s weighted,etc. On BNB Chain, every DTF feeds a single vlRSR staking vault, on a 7-day governance cycle (with a 7-day unlocking delay).
- Governors share in the fees. Each DTF charges a 0.3% mint fee and a 0.6% TVL fee. After a standard platform cut, the rest flows to vlRSR governors: it’s converted into RSR and added to your staked position automatically, with no manual claiming.
- The monthly burn. The protocol’s cut of fees are also used to buy back and burn $RSR, fully transparent via the RSR burn dashboard on @Dune.
The $RSR contract address on BNB Chain is 0x23f72a3Db61D6CB8aBE5d9AF1Ac4B6c99327bFee. Users who wish to bridge are recommended to use Portal Bridge.
The AI buildout is happening. Own your share of it.
Follow @reserveprotocol for what’s next.
Nothing in this article is investment advice. Past performance may not be indicative of future results. BUILDOUT, POWER, PHOTON, NEOCLOUD, and ROBOTS are concentrated, single-theme baskets of experimental tokenized assets, and none tracks any third-party index — each can be highly volatile, illiquid, and may lose value. The market-size figures cited (Goldman Sachs, WSTS, ABI Research, and Morgan Stanley) are third-party estimates of market size, are inherently uncertain, and are not forecasts of any fund’s performance. ABC Labs designed each fund’s methodology and selected its initial constituents accordingly. Decentralized governance is incentivized to follow that methodology to enact quarterly rebalancing. Reserve DTFs are not ETFs, are not FDIC/SIPC-insured, and neither ABC Labs nor the Reserve project are regulated financial intermediaries; it’s up to you to decide whether to buy any asset on the platform. Constituents and weights shown are illustrative as of the latest quarterly rebalance and may change. Full terms, eligibility & risks: reserve.org/terms_and_conditions.
