The Wave of Speculative Mania Around AI Reproduces the NFT Era Scenario

When Anthropic released Claude Code last spring and a wave of AI agents took over the market, it became clear: we are echoing the NFT era of 2021. The new technology is truly impressive, but it’s already surrounded by the familiar spectacle of marketing, influence, and speculation. The question isn’t whether AI can transform industries — it’s how collective imagination has turned real innovations into a theater of elite membership.

How social media turned innovation into a cultural myth

Every significant technology goes through the same lifecycle. The first phase — early adopters find real use cases. The second phase — social media seizes the narrative and turns it into a status symbol. Conversations shift from the technology itself to language of belonging: “this is the future” becomes a ticket into an exclusive club.

Platforms like X accelerated this process exponentially. The loudest voices are not representative — they project unwavering confidence to an audience that rewards boldness. Any platform thrives on engagement, and engagement fuels extremes. “It’s useful and interesting” won’t spread widely, but “it changed everything, you’ll lose your job” will.

A hundred retweets with “This is a revolution” are not signals — they are echoes. Echoes are mistaken for consensus, consensus for truth, truth for investment doctrine. When enough people start “demonstrating belief” in a certain outcome, the demonstration itself becomes indistinguishable from proof. The NFT era showed this vividly: people didn’t need JPEG monkeys, they needed to “want what everyone else wants.” This market psychology hasn’t disappeared — it has only changed its face.

The double mistake of financial markets and social media

Here’s a strange symmetry. On one end, social media populists declare: AI has already won, the future is here, all functions are being rewritten from now on. On the other end, Wall Street panics: AI has already killed software, subscription revenues are dead, the business model is obsolete.

Both sides made the same mistake — jumping to the end of the game when many moves remain.

On January 30, 2025, Anthropic launched Claude Cowork. Within a week, the software sector lost $285 billion. The ETF for software ($IGV) plummeted 22% early in the year while the S&P 500 was rising. Out of 110 index components, 100 were in the red. RSI hit 16 — the lowest since September 2001. Hedge funds wildly increased short positions.

Rational analysis gave way to group panic. But logic demands a choice: either investments in AI are collapsing, or AI is powerful enough to wipe out the entire software sector. Both cannot be true at once.

Reality behind the curtain

The capabilities of recent AI models are genuinely impressive. I use these tools daily — they have measurably increased my efficiency. Over the past six months, progress has been enormous. Real, not hypothetical, applications are happening now in programming, research, analytics. This creates real value for those who know how to use them.

But here’s the critical point: there is a huge gap between our current state and the technical vision of the future. This gap will be filled chaotically, gradually, and specifically for each company.

No, Claude will not cause an immediate social collapse. That doesn’t mean people no longer need work management interfaces. It doesn’t mean Anthropic has won the AI war.

Imagine the logic you’re being asked to believe: enterprise software — years of accumulated integrations, compliance, and institutional knowledge — will be replaced in quarters? The pay-per-use model will vanish overnight? Companies with annual revenues over $10 billion and 80% margins will disappear because a chatbot can write a function fragment?

Dan Ives from Wedbush directly said: “Corporations will not abandon hundreds of billions invested in software to switch to Anthropic and OpenAI.” Jensen Huang, more than anyone with reason to speak about AI’s disruptive power, called the idea “the most illogical thing in the world.”

Those most vocal about the “endgame” often benefit the most from your “indisputable belief”: followers, consulting, conferences. The incentive system rewards bold predictions without accountability for timing.

When numbers tell a different story

Look at companies the market considers doomed:

Salesforce. Agentforce revenue grew 330% year-over-year. Annual revenue hit $500 million. Free cash flow is $1.24 billion. Forward P/E is 15. The company just announced a target revenue of $60 billion by 2030. This isn’t a victim of AI — it’s building an AI delivery layer for business.

ServiceNow. Subscriptions grew 21%, operating margin rose to 31%. A $5 billion buyback was approved. Their AI suite (Now Assist) has a yearly contract value of $600 million, aiming to surpass $1 billion by year’s end. Yet, their stock has fallen 50% from its peak.

Should valuations be adjusted downward due to risks? Possibly. But that should have happened years ago. When all index components fall in sync, the market stops analyzing and plunges into storytelling.

Differentiated risks, not an apocalyptic scenario

Yes, some companies will be displaced. Point solutions offering standardized workflows are fragile. If your product is just an interface over someone else’s data, you’re vulnerable. LegalZoom fell 20% because when AI can automatically review contracts, paying a traditional provider for the same function becomes hard to justify.

But companies with deep integration, proprietary data, and platform architecture are a different story.

Salesforce is embedded in the tech stack of every Fortune 500. ServiceNow is the enterprise IT department’s accounting system. Datadog, built on consumption, means more AI computations = more monitoring = more revenue. Their growth accelerated to 20% annually.

Selling infrastructure because “AI kills software” is like selling construction equipment because buildings are rising. Logically nonsensical.

Lessons from history and cycles of uncertainty

The SaaS crisis of 2022 is instructive. The segment fell 50%. Median revenue multiple dropped from 25 to 7. Meanwhile, financial reports looked fine. The recovery was significant — Nasdaq up 43% in 2023.

Panic around DeepSeek in January 2025 was structurally identical to now. The release of one product triggered a reassessment of the entire sector’s viability. NVIDIA fell, then fully recovered.

Many draw parallels to the early internet bubble burst, when tech stocks tumbled. But there’s a nuance: Amazon fell 94%, then became one of the most important companies in the world.

Jim Reid from Deutsche Bank told the truth: “At this stage, it’s almost impossible to identify long-term winners and losers.” And it’s precisely this uncertainty — the acknowledgment that we don’t know the end — that makes a mass sell-off a mistake.

The paradox of the end in the middle of the game

Here’s the essence. The real trajectory will be more volatile and less certain than either hype or panic suggests. Some software companies will integrate AI and become stronger. Few will be displaced. Most will adapt — slowly, unevenly, not very Twitter-friendly.

The gap between current reality and the technical future will be filled with chaos of specific progress. This is not material for tweets or panic. It’s the “middle stage” of a great transformation.

Those who succeed will be those who withstand uncertainty, not those who rush to embrace an early final narrative. Great leaders always find a way.