Hook: The $100M Question with No Answer
You open Crypto Briefing, expecting the usual Solana fee spike or EigenLayer restaking drama. Instead, you find this: 'Infinity raises $15M from Touring Capital, OpenAI and Anthropic researchers at $100M valuation.' No product. No code. No explanation of what 'AI infrastructure' even means. Just a number and a list of names.
My first instinct as a blockchain pragmatist?
Alpha hidden in the noise. Or maybe just noise dressed in alpha’s clothing.
In a bull market, every fundraising announcement looks like a rocket launch. But I’ve been around since the 2017 ICO mania—I audited 15 whitepapers in my Bangkok Telegram group, flagged eight as garbage before the crash. This smells familiar. The difference? The investors aren’t anonymous crypto funds; they’re researchers from the two most secretive AI labs on the planet. That’s a signal worth decoding.
Context: The Bare Facts and the Missing Blueprints
Let’s strip the narrative. Infinity—no website, no technical whitepaper, no GitHub repo cited—raised $15 million at a $100 million valuation. Touring Capital led. The cap table includes unnamed researchers from OpenAI and Anthropic. The pitch: “AI infrastructure.” That’s it.
Now, why does a crypto publication cover an AI company? Because the line between crypto and AI is blurring faster than a TON transaction. In 2025, every Layer 2 is adding AI-agent support; every DeFi protocol is flirting with autonomous trading bots. Infinity could be building the plumbing for that convergence. Or it could be a glorified cloud management tool. We don’t know.
But the researcher involvement is the real puzzle. OpenAI and Anthropic employees rarely invest personally—they have NDA restrictions, insider information, and reputational risk. That they chose to put their own money into Infinity suggests one of two things:
- The technology is genuinely groundbreaking and fills a gap they see inside their own labs.
- They’re signaling personal belief in the founders, effectively vouching for execution ability.
Both are powerful. Neither tells us what the product does.
Core: Deconstructing the AI Infrastructure Hype—Through a Crypto Lens
Here’s where my background as a Pragmatic Code Auditor kicks in. I’ve spent years dissecting rollup data availability claims. The pattern is identical: a project raises millions on a broad term, everyone assumes it’s the next big thing, and the actual technical details reveal a commodity dressed as a revolution.
Take the Data Availability layer. In 2023, every rollup needed “dedicated DA.” Celestia raised $55 million. Then EigenDA emerged. Then we learned that 99% of rollups don’t generate enough data to need dedicated DA—they can just use Ethereum calldata for pennies. The hype was a delusion.
Infinity’s “AI infrastructure” suffers from the same disease. AI infrastructure is an umbrella term covering: - Compute orchestration (Kubernetes for GPUs) - Model training pipelines - Inference serving - Data versioning and labeling - Monitoring and observability
Each subcategory has dozens of well-funded incumbents. Why would anyone need Infinity? The $15 million is seed capital, not enough to build a new data center or a competitive cloud service. So it must be a software-defined product—likely a control plane or a middleware that optimizes existing resources.
But here’s the contrarian thought: the researchers investing know that the biggest bottleneck in AI today is not compute—it’s coordination. Training a frontier model requires orchestrating thousands of GPUs across multiple clusters, managing fault tolerance, and minimizing idle time. That’s a hard distributed systems problem. And crypto’s biggest failure has been solving distributed consensus without creating value for the token. If Infinity is applying blockchain-like coordination primitives (trustless scheduling, on-chain audit trails) to AI training, it could genuinely be disruptive.
I’ve seen this before. In my 2025 Bangkok hackathon for AI-agent wallets, I watched teams struggle with multi-chain execution and verifiable inference. The teams that succeeded used cryptographic proofs to guarantee that a model ran correctly on untrusted hardware. That’s infrastructure—but it’s also crypto-native.
The problem: Infinity’s announcement gives zero indication of any blockchain element. It might be pure cloud middleware. And without code, the only thing we can audit is the investor signal.
Personal Experience Signal: In 2022, after the Terra collapse, I pivoted to regulatory training. I learned that the smartest investors often bet on teams, not technologies. The OpenAI and Anthropic researchers are betting on Infinity’s team. But in crypto, we’ve seen “rockstar teams” fail because the technology wasn’t actually needed. Remember Terra? Do Kwon was a “genius” until the music stopped.
Code doesn’t lie, but narratives do. Right now, Infinity’s narrative is a blank check.
Contrarian: The Fragmentation Trap
Here’s the blind spot everyone is missing: AI infrastructure is already fragmenting along protocol lines, just like crypto did with Layer 2s and cross-chain bridges. Every major AI lab uses its own internal tooling. OpenAI has their own orchestration. Anthropic has theirs. DeepMind has GSPMD.
A new, generic infrastructure layer has to convince these labs to trust external software with their most valuable assets—their training pipelines and proprietary data. That’s a harder sell than convincing a DeFi protocol to use a new oracle.
I think of Cosmos IBC. Technically elegant—a trust-minimized cross-chain communication protocol. But the application ecosystem is fragmented, and ATOM captures almost no value. Why? Because the applications don’t need the protocol—they can build their own bridges or use centralized alternatives.
Infinity faces the same risk. If it builds a general-purpose AI infrastructure, the big labs will either ignore it or build their own version. The small teams might adopt it, but then the unit economics are terrible because small teams have low willingness to pay.
The contrarian angle? The smartest minds might be wrong this time. Researcher investments are not a guarantee of commercial success. In 2018, I watched a team of ex-Google Brain researchers raise $20 million for a “decentralized AI training network.” It died in two years because the network effect never materialized. The code was beautiful; the demand was imaginary.
But let me be fair: it’s possible Infinity is solving a real pain point that I haven’t considered. Maybe it’s a secure enclave for training models on sensitive data (healthcare, finance) using homomorphic encryption. That would fit the crypto ethos of trustless computation. And the researcher involvement could mean they’ve seen a prototype that works.
Takeaway: Trust Is the New Currency—But We Don’t Know Who to Trust
Infinity’s raise is a roll of the dice on a yet-unseen product. The smart money—the researchers—are betting on the founders. But in a bull market fueled by FOMO, that’s exactly when bad bets get funded.
My advice: watch for three signals over the next 12 months. If Infinity releases a public audit of its infrastructure, a technical whitepaper, or an open-source component, then it’s serious. If it stays silent and raises another round at a higher valuation without showing product, run.
Alpha hidden in the noise? Maybe. But the noise is loud, and the alpha is buried under a mountain of hype. Code doesn’t lie, but this project hasn’t written a single line of public code yet.
Trust is the new currency—and right now, Infinity is asking us to trust a blank check. I’m holding my fire until I see the transaction logs.