Building the New Foundation

A Note Before We Start

Alpha Compute Corp. (Nasdaq: ALP) is a publicly traded company building both halves of the AI economy: the compute that intelligence runs on, and the distribution channels where attention lives. This is based on a simple thesis: whoever owns both compounds, while everyone else hits a ceiling. This is our first newsletter. The plan for it is simple. Every issue we will tell you what we are building, why it matters, and what we actually think about the industry, including the parts the big players would rather not discuss. We will keep it plain.

Here is the idea behind everything that follows. To win in this era you need to own two things. The first is infrastructure, which is the compute all channels run on. The second is distribution, which are the channels where people spend their attention. Most companies rent one and own the other. We are building both, so that each one feeds the other. The rest of newsletter walks through how and shows that it is already working.

PART ONE · INFRASTRUCTURE

Whoever Owns the Compute Owns the Century

While distribution is the demand side of the equation, infrastructure is the supply side, and right now the entire world has decided the supply side is the thing worth building. Compute has become the foundation of every industry. Intelligence, finance, media, and defense all run on GPUs, yet that layer has stayed centralized, custodial, and extractive, controlled by a handful of platforms that own the infrastructure everyone else depends on. Governments have noticed.

Global spending on sovereign AI is projected to approach $100 billion in 2026, and the list of countries building their own compute now reads like the G20. France is leading Europe, where Mistral raised roughly $830 million in institutional debt to buy about 13,800 Nvidia chips and stand up a data center near Paris. Saudi Arabia launched HUMAIN under its sovereign wealth fund and announced a partnership with Nvidia targeting up to 600,000 GPUs over three years, with a stated goal of making Saudi Arabia the world's third-largest AI market, behind only the US and China. Japan, Canada, Germany, the UK, India, and the UAE have each launched national compute or sovereign-model programs of their own. The common thread is a hard lesson learned in nations everywhere: data sovereignty means very little if the compute underneath it belongs to someone else.

Compute has become important enough that the government itself is now getting involved, from two different directions. Senator Bernie Sanders has proposed an American AI Sovereign Wealth Fund Act that would take a one-time 50% claim on the stock of the largest AI companies and give the public ownership stakes, voting shares, and board seats, which is a serious proposal to take a controlling interest in the largest AI companies. At the same time, the federal government is pushing hard to build, and to make it easier for others to build. It now treats data centers as national infrastructure, with a Commerce Department fast-track for projects above a $500 million capital commitment, a Ratepayer Protection Pledge in which major developers agreed to pay for the new power they consume, and proposed legislation that would let AI data centers run their own off-grid power without tripping federal grid rules. One side wants to take a stake in compute and the other wants to subsidize it, but both agree it is now too important to leave alone.

ALPHA / SOVEREIGN

This is the gap Alpha / Sovereign was built to fill. It is our Enterprise and U.S. government contracting arm, a GPU-as-a-service business that helps the country build out the AI infrastructure it needs while keeping that infrastructure from quietly becoming a surveillance apparatus.

How we keep that promise is the core of the company. When a hyperscaler sells you “private” compute, what you really get is a contract. Your data still runs on machines that company owns and operates, and that it can be forced to hand over. We do it differently. We enforce privacy in the hardware. Using trusted execution environments and hardware encryption built into our B200 and B300 clusters, your data stays sealed inside the chip, so even we cannot see what is being processed. The protection is built into the silicon, not promised on paper. We also recently became a sponsor of the Confidential Computing Consortium, the group that sets the global standards for this kind of work. Building capacity like these costs more than any balance sheet wants to carry on its own, which brings us to the next division.

ALPHA / CAPITAL

Every great infrastructure buildout in history, from railroads to oil to telecom, demanded new financial instruments before it could happen, and compute is following the same script. The cliche that compute is the new oil is turning literal, because just as oil produced futures, indices, and even currencies like the petrodollar, compute is about to spawn an asset class of its own. The most credible voice in finance is already saying so. At the Milken conference in May 2026, BlackRock’s Larry Fink told the room that “a new asset class will be buying futures of compute,” warning of real shortages across compute, chips, memory, and power. Days earlier, on May 12th, CME Group announced the first compute futures market, built on daily GPU rental-rate indices, which is to say oil futures for GPU hours. Institutional capital is financializing compute in real time.

GPU fleets cost more than equity can carry, so Alpha / Capital finances them through dedicated vehicles that turn the hardware into yield. The capital is built around the assets it generates, so growth does not dilute the core. In practice that means special-purpose vehicles structured for GPU deployments, insurance and risk products written around compute assets, and structured financing that lets serious operators fund serious infrastructure without getting stuck in the bottlenecks everyone else hits. As compute turns into something you can trade, hedge, and finance, Alpha / Capital is how we take part in that market directly.

WHAT IS RUNNING TODAY

None of this would mean much if the hardware were not already running. ALPHA-01, our first enterprise-scale NVIDIA Blackwell cluster, runs 504 B200 GPUs on 100% hydroelectric power in Canada, and it is live and serving its first enterprise customer. In Q2 we signed our first offtaker, a $32.2 million two-year contract with a leading AI research firm that brings in $16.1 million in annual contracted revenue. Beyond these signed contracts, the opportunities we are actively pursuing across the AI research, enterprise, and sovereign segments represent more than $200 million in potential lifetime contract value, though these remain at pre-contract stages. ALPHA-02, a 576-GPU B300 cluster in Sweden, is our next site coming online, with power and racks being put in place and first GPUs targeted for Q3 2026, and we plan to keep growing the portfolio from there. The numbers in this section were projections the last time we wrote to you. They have customers and signatures behind them now.

That is the supply side, built and running. Now to the demand side, and the attention all of it serves.

PART TWO · DISTRIBUTION

Whoever Owns the Attention Owns the Market

The biggest players in technology are spending tens of billions to buy distribution, and the logic is blunt: a model without an audience has no moat. In March 2025, Elon Musk merged X into xAI in an all-stock deal that valued xAI at about $80 billion and X at $33 billion, putting an AI lab and a social network of hundreds of millions of users under one roof to unite “the data, models, compute, distribution and talent.” In April 2026, OpenAI made the first media acquisition in its history, buying the daily talk show TBPN for a reported sum in the low hundreds of millions, because a trusted, direct line to its audience was worth more than the revenue. The reasoning is the same in every case: a model is only useful if people can reach it, so the companies that are winning are buying their way to the audience.

This isn’t a new idea. It goes back to 2017, when eight researchers published “Attention Is All You Need” and introduced the transformer, the architecture behind every large language model you use today. The key word in that title is attention. For a century the economy has run on capturing it: newspapers sold it, radio sold it, and the studios and social platforms turned it into one of the most profitable businesses there is. What changed in 2017 is that we built an engine that can read the data behind that attention and do far more with it than sell an ad, turning it from something you sell once into something that compounds.

We reached the same conclusion, and we acted on it.

ALPHA / GAMES

Six months ago we told you we were going after GAMEE. On May 27th we closed it. GAMEE is one of the most widely used gaming platforms in the world, with 120 million registered users, 10 billion gameplay sessions, and 61 million users reachable inside Telegram’s Mini App ecosystem alone. It is a proven digital economy operating at real scale,

—
$3.9 million in 2025, with Q1 2026 at $926,000, up 56% quarter over quarter;

—
5.57 million users and 88.5 million game plays in a single quarter;

—
1.7M monthly active users and 150K daily active users;

We acquired a 60% controlling interest at an implied $18M valuation. Co-founder Bozena Rezab now leads Alpha / Games as EVP, and Animoca Brands, the prior majority owner, signed a two-year standstill, which means they are staying in and betting alongside us. Alpha / Games is our consumer gaming division, the place where we turn our compute into games people actually play. GAMEE is its first asset, and the plan is to build out a portfolio of titles where both people and AI agents play, compete, and connect, all running on infrastructure we own.

The logic here is straightforward. Our confidential-compute platform needs somewhere to run, and GAMEE already has 120 million people showing up. Every AI-powered game, every agentic gameplay loop, and every privacy-preserving ad inside the platform runs on GPU infrastructure we own. So as GAMEE grows, our compute does more work and earns more revenue, and that revenue helps us buy more compute. Phase II of GAMEE’s Gold Fest campaign launches in Q2/Q3 2026 with a $2M prize pool and an agentic layer that lets users send AI agents in to compete, one of the first real consumer deployments of AI agents in gaming. It will run on our infrastructure.

ALPHA / STUDIOS

Alpha / Games is one way to reach people. Stories are another, and that is what the next division we are launching is for. Alpha / Studios is our creative arm for media, IP, and agentic projects. It exists to tell our story and our portfolio companies’ stories across film, video, games, and agentic media, and to build partnerships that reach across industries.

Alpha / Studios gives the distribution side a voice. Alpha / Games brings the audience. Shaping the message is the job of Alpha / Studios, explaining ideas like confidential compute and digital sovereignty to people who are never going to read a whitepaper. The two work together: one brings people in, the other gives them a reason to pay attention.

That rounds out the demand side. We built the supply side first; here is why the two belong together.

WHY WE BUILD BOTH HALVES

One Foundation, Many Companies

Now put the two halves back together, because that is where the strategy actually lives. The compute backbone sits at the center, a fleet of GPUs we own outright. Dedicated vehicles finance that hardware so growth never dilutes the core, and the companies we build and acquire on top send their revenue back to expand it. Alpha / Games runs on our compute, Alpha / Studios runs on our compute, the sovereign and enterprise workloads run on our compute, and every dollar they earn flows back into buying more of it.

This shape tends to win whenever a new technology shows up. The companies that came out ahead in the television era owned both the networks and the studios that filled them. The ones that won broadband owned both the pipe and the content moving through it. Owning both the supply and the demand is what lets a single business compound instead of just grow. That is the structure we are building toward, and the pieces are already connected and running.

WHERE WE’VE BEEN AND WHERE WE’RE GOING

On the Road, On the Air

We have been making the case in person and on screen. This month we were at iConnections Global Alts in New York, where we talked with institutional allocators about why AI infrastructure is the defining asset class of this cycle. In May we were at the Linux Foundation Open Source Summit in Minneapolis, spending time with the developer community our stack is built on. We also ran our first national television campaign this month, with spots on CNBC, Fox Business, and Newsmax, timed to land with the GAMEE close and ALPHA-01 going live.

We would rather show you than tell you, so we are hosting an online virtual market update call for investors and partners in July. We will walk through both theses, the numbers behind ALPHA-01 and the GAMEE close, and what comes next, with registration details to follow. Looking ahead, Alpha / Sovereign launches in Q3, ALPHA-02 goes live in Q3, GAMEE integration continues through year-end, and the opportunity set keeps expanding.

The First Layer

We call ourselves Alpha Compute because this is the first layer of the new stack, the layer the open internet was always supposed to have. For a long time the infrastructure around you was built to watch you and sell your own data back to you. We are trying to finish what the internet’s founders started, building out confidential computing and sovereign intelligence so that ownership and privacy come from the hardware itself, not from a company’s goodwill.

That is the whole idea: own the channel and own the compute. The attention runs on our infrastructure, and the infrastructure pays for itself by serving that attention. It is a simple loop, and it gets stronger each time around. We think it is worth building, and we are well underway.

Until next issue,

The Alpha Compute Team