When Citizens analyst upgraded Okta to Outperform with a $170 target last week, the market nodded in approval. The rationale was clear: enterprise AI agent deployment will drive an explosion in non-human identity (NHI) management, and Okta, as the independent identity and access management (IAM) king, is best positioned to ride that wave. But something felt off. This isn't the first time we've heard a centralized identity vendor promise to own the next paradigm. In 2017, I spent six weeks auditing whitepapers for ICOs that claimed to be building the future of identity—most of them were just marketing fluff with speculative tokenomics. That experience taught me to look beyond the narrative. Here, the narrative is dangerously incomplete.
Context: The AI Agent Identity Boom AI agents—autonomous software entities that perform tasks on behalf of humans—are multiplying. From customer service bots to trading algorithms, each agent needs a verifiable identity to authenticate, authorize, and audit its actions. Traditional IAM systems like Okta, Azure AD, and ForgeRock are built for human users, not machines. They manage identities via email, password, and SSO—concepts that don't map well to a swarm of agents that may spawn, die, and replicate across chains. Yet the market is betting that Okta's existing infrastructure, especially its Auth0 developer platform, can adapt. The idea is that agents will use OAuth flows and API keys managed by Okta, just like human users. But this is a category error.
Core: The Technical and Ethical Flaws of Centralized Identity for Agents Let's start with the technical. Okta's architecture is multi-tenant SaaS, built for up to, say, 100,000 human users per enterprise. AI agents could push that to millions or billions of distinct identities within a single organization. The cost scales linearly with usage—Okta charges per MAU or per API call. For a human, one MAU is trivial. For an agent that makes thousands of calls per minute, the bill becomes astronomical. More importantly, centralized identity introduces a single point of failure. The 2022 Okta breach—where a third-party support vendor was compromised—exposed customer data. Imagine that same attack vector applied to AI agent credentials: an attacker could take over an entire fleet of trading agents. The damage would be catastrophic and irreversible on a centralized ledger.
But the deeper issue is philosophical. Okta is a commercial SaaS product. The company controls the protocol, the pricing, and the data. Enterprise lock-in is the business model. AI agents, by their nature, are meant to be autonomous and interoperable across ecosystems—different blockchains, different cloud providers, different DAOs. A centralized identity provider creates a bottleneck: every interaction must pass through Okta's servers. That reintroduces the very trust dependency that decentralized systems aim to eliminate. Based on my work in the 2021 'Block & Brush' initiative, where we built a DAO-governed art marketplace, I saw firsthand how creators and developers struggle when a single entity controls the identity layer. The community lost trust when the platform changed royalty terms. For AI agents, such a dependency would be a recipe for systemic risk.
Furthermore, the 'AI agent identity' narrative is a Trojan horse for Okta's real strategy: upselling existing customers with a new line item. The unit economics shift from human seats to machine identities. But there's no evidence that enterprises will pay a premium for a centralized IAM system to manage their agents. In fact, the opposite is happening. Decentralized identity (DID) standards—like those from the W3C and the ION network on Bitcoin—are gaining traction. They allow agents to own their own identifiers, secured by cryptography, without any central authority. The cost is near-zero, the scalability is indefinite, and the trust model is transparent. Why would a developer building an agent on Ethereum or Solana choose to pay Okta for a centralized identity when they can mint a decentralized identifier for free?
Contrarian: The Pragmatist's Defense of Okta To be fair, decentralized identity is still immature. There are no mainstream UX standards, and interoperability between different DID methods is a mess. Enterprises prefer the comfort of a vendor with a support team, SOC 2 compliance, and a 99.99% uptime SLA. Okta can deliver that tomorrow. A decentralized identity solution might take years to match that reliability. The market is rational to price in this short-term advantage. During the 2022 bear market, I ran a peer-support network that connected 500 developers. Many of them were building decentralized identity solutions, and the feedback was consistent: 'It works, but onboarding is hell.' So yes, for the next 12-18 months, Okta will likely capture some of the AI agent identity spend, especially in regulated industries like finance and healthcare. The $170 target is plausible if you assume a short-term boom.
But here's the blind spot: the fundamental value proposition of blockchain is trust minimization. AI agents, by their very nature, will demand a trust-minimized identity layer. A centralized IAM is a honeypot; it's a single point of trust that can be exploited, censored, or monetized against the user's interest. The architecture of the internet is moving toward edge computing, zero-trust networking, and self-sovereign data. Okta is a relic of the client-server era. The upgrade to Outperform is based on an extrapolation of the past, not a vision of the future. It's like upgrading a horse-and-buggy company because you expect more people to travel on dirt roads. The roads are already being paved with decentralized protocols.
Takeaway: Restoring Faith in Decentralized Promises The market's enthusiasm for Okta reflects a deeper longing for a simple, trusted solution to the AI identity problem. But simple isn't always safe. The blockchain community has spent the last decade building the infrastructure for self-sovereign identity, from the Ethereum Name Service to Ceramic Network to the Veramo SDK. The pieces are coming together. The next big unlock will be when an AI agent autonomously registers its own DID on a blockchain, negotiates access with a smart contract, and executes a transaction—all without asking a central authority for permission. That's the vision. Okta's $170 target is a distraction. As I wrote in my 2017 ethical audit report, 'Auditing ethics before auditing assets.' The ethics of identity are clear: decentralization is the only path to genuine autonomy for AI agents. We are building bridges where code ends and trust begins. And that bridge cannot be owned by a single company.
Transparency is the new currency. The next time an analyst upgrades a centralized identity stock, ask yourself: Who really controls the identity of my AI agent? If the answer is not the agent itself, we have failed the promise of decentralization.