There’s nothing new under the sun—the current AI frenzy reminds me of NFTs.

Author: market participant

Translated by: TechFlow

TechFlow Intro: As the latest wave of AI agents—OpenClaw and Claude Code—sweeps across social media, the author keenly detects a frenzy reminiscent of the NFT era in 2021.

This article dissects how social media amplifies technological narratives, how Wall Street indiscriminately dumps software stocks due to the “AI kills software” bias, and why giants like Salesforce and ServiceNow remain unfairly punished by markets—even after delivering astonishing results.

The author argues we are in the “mid-game” of a great revolution—where both extreme optimism and extreme panic represent premature attempts to price in an endpoint that has yet to arrive.

Full Text Below:

This surge around OpenClaw and Claude Code reminds me of the hysteria of the NFT era.

New technologies emerge with real utility—and simultaneously resonate culturally and narratively within the spirit of the times. Like every technology that captures the collective imagination at just the right moment, it is processed through the same “distortion machine”—the very mechanism that once transformed JPEGs of monkeys into a $40 billion asset class.

The pattern is identical: genuine innovation arrives; early adopters discover real value. Then the social layer takes over—suddenly, conversation shifts away from the technology itself and becomes performative “team alignment.”

Declaring “This is the future” becomes a tribal badge. Writing tutorials, think pieces, and inflating present-day value earns social validation. The compounding of opinions outpaces the compounding of technology itself.

(I promise—a market perspective is coming shortly.)

The Cognitive Distortion Machine

X makes things worse. Social media is increasingly treated as a legitimate lens onto reality—yet it bends the image of facts.

The loudest voices aren’t representative—they’re performing “unwavering conviction” for an audience incentivized to reward such behavior. Every mainstream platform runs on engagement—and engagement rewards extremes. “This is interesting and useful” doesn’t go viral; “This changes everything—and your job is gone” does.

A hundred retweets declaring “This changes everything” aren’t a signal—they’re an echo. That echo is mistaken for consensus; consensus is mistaken for truth; and truth is mistaken for an investable thesis.

Girard would have a field day here. When enough people perform “faith” in a given outcome, the performance itself becomes conflated with evidence supporting that outcome. The NFT era proved this definitively: people didn’t want JPEGs—they wanted “to want what everyone else wants” [1].

What Is Real?

The latest model capabilities are astonishing—far more so than NFTs, which possessed little practical utility beyond speculation and cultural signaling.

I use these tools daily. They improve my productivity in concrete, measurable ways. The underlying models are genuinely impressive, and their improvement trajectory is steep. When I compare what these tools could do six months ago versus today, the incremental gain is massive.

And the broader potential is limitless. AI-assisted programming, research, analysis, writing—these aren’t hypothetical use cases. They’re happening now—and creating real value for those who wield them well.

I don’t want to be the person in 1998 dismissing the internet. That’s not the point—I’m extremely long-term bullish on AI. The point is the timeline, and the gulf between potential and present reality.

What Is Not Yet Real

No—Claude won’t catalyze societal upheaval overnight. This doesn’t mean humans no longer need interfaces to manage work. Nor does it mean Anthropic has already won the AI war.

Consider what the most breathless claims actually ask you to believe: that enterprise software—decades of accumulated workflows, integrations, compliance frameworks, and institutional knowledge—will be replaced in quarters, not years? That seat-based pricing will collapse overnight? That companies generating over $10 billion in annual revenue and 80% gross margins will vanish because a chatbot can write a function? [2]

Dan Ives of Wedbush put it bluntly: “Enterprises won’t rip out hundreds of billions of dollars in existing software infrastructure to migrate to Anthropic, OpenAI, and others.” [3] And Jensen Huang—the person with arguably the strongest incentive to promote AI disruption—called the idea of “AI replacing software” “the most illogical thing in the world.” [4]

Those most eagerly proclaiming the “Endgame” (a term popularized by @WillManidis) tend to be those who benefit most from your “certainty”: followers, consulting gigs, subscriptions, speaking engagements. The incentive structure rewards bold predictions about timing—with zero accountability.

The Market’s Mirror

What fascinates me is that the market commits the same error from the other side of the table.

Anthropic launched its Claude Cowork plugin on January 30. Within a week, $285 billion evaporated from software, financial services, and asset management stocks [5].

The Software ETF ($IGV) is down 22% year-to-date—while the S&P 500 is up. Of its 110 holdings, 100 are in the red. Its RSI hit 16—the lowest reading since September 2001 [6].

Hedge funds are aggressively shorting software stocks—and doubling down [7]. The narrative logic? “AI kills SaaS.” Every seat-based software company is a “walking dead.”

This sell-off is indiscriminate. Companies with vastly different AI risk profiles are lumped together as identical trade proxies [8]. When 100 of 110 names in an index are falling, the market isn’t analyzing—it’s riding the narrative high.

Note: A recovery may already be underway as I write this.

Throwing Out the Baby With the Bathwater

Let’s look inside the companies supposedly facing existential doom.

Salesforce’s Agentforce revenue grew 330% year-over-year, with annualized revenue exceeding $500 million and generating $12.4 billion in free cash flow. Its forward P/E sits at 15x. It just announced a $60 billion revenue target for fiscal year 2030 [9]. This isn’t a company being disrupted by AI—it’s building the enterprise AI delivery layer.

ServiceNow’s subscription business grew 21%, operating margin expanded to 31%, and it authorized a $5 billion share buyback. Its AI suite, Now Assist, reached $600 million in annual contract value (ACV), targeting $1 billion by year-end [10]. Yet its stock is down 50% from its highs.

Should these names see modest valuation adjustments for risk? Perhaps. But savvy investors priced that in years ago. As many far smarter than me have pointed out: this sell-off asks you to simultaneously believe “AI capex is collapsing” *and* “AI is powerful enough to destroy the entire software industry” [11]. Both cannot be true. Pick one.

Identifying Real Risk

Will some companies truly be displaced? Yes.

Point solutions—tools offering standardized, single-workflow functionality—are vulnerable. If your entire product is just a UI layer built atop non-proprietary data, you’re in trouble. LegalZoom fell 20%—for firms like this, the concern is substantive [12]. When AI plugins can auto-review contracts and classify NDAs, paying traditional vendors for the same capability becomes indefensible.

But companies with deep integrations, proprietary data, and platform-level foundations are a completely different story. Salesforce is embedded in the tech stack of every Fortune 500 company. ServiceNow is the system of record for enterprise IT. Datadog’s consumption-based model means more AI compute directly translates to more monitoring revenue—its non-AI business growth actually accelerated to 20% year-over-year [13].

Selling digital infrastructure because “AI kills software” is as absurd as selling construction equipment stocks because a skyscraper is going up.

We’ve Been Here Before

The 2022 SaaS crash is instructive. The sector fell over 50%. The median forward revenue multiple dropped from 25x to 7x—below pre-pandemic levels [14]. And earnings remained strong throughout. The rebound was dramatic—Nasdaq rose 43% in 2023. Granted, the trigger then was more interest-rate shock than fundamental deterioration.

The DeepSeek panic of January 2025 is even closer. NVIDIA plunged on fears that cheap Chinese AI models would render the entire AI infrastructure build-out meaningless—only to fully recover [15]. That fear mirrored today’s structurally: a single product launch triggered an industry-wide existential reassessment.

Many observers draw direct parallels between this moment and the early stages of the dot-com bust—tech stocks falling while consumer staples, utilities, and healthcare rise [16]. But one thing about the dot-com bust: Amazon fell 94%—then became one of the world’s most important companies. Markets trying to price the “endgame” mid-game created one of history’s greatest buying opportunities.

Jim Reid of Deutsche Bank put it plainly: “Identifying long-term winners and losers at this stage is almost pure guesswork.” [17]

I’d bet he’s right. And that uncertainty—that acknowledgment that we simply don’t know how this ends—is precisely why this indiscriminate sell-off is wrong.

The Endgame Fallacy

The hype merchants on X and the panic sellers on Wall Street commit the same error—on opposite sides of the board.

One group declares AI has already won, the future is here, and all institutions and job functions are being rewritten from this moment forward. The other declares AI has killed software, subscription revenue is dead, $10 billion in free cash flow no longer matters—because the business model is obsolete.

Both jump to the “endgame” while the game still has many moves left. The chasm between where we are and the technological horizon will be filled—not by a single decisive event—but by messy, incremental, company-specific progress. Some software firms will integrate AI and grow stronger; a few will truly be displaced; most will adapt—slowly, unevenly, and un-tweetably.

The actual trajectory will be more volatile—and less certain—than either hype or panic suggests. Those who do well from here will be the ones who tolerate that ambiguity—not those rushing to lock in an overly premature narrative.

Great operators always find a way.

References

[1] Girard’s Mimetic Desire Theory (https://www.iep.utm.edu/girard/)

[2] Fortune: Why SaaS Stocks Are Falling Irrationally—Like During the DeepSeek Panic

[3] CNBC: Impact of AI Tools on SaaS Software Stocks

[4] CNBC: Jensen Huang Calls “AI Replacing Software” “The Most Illogical Thing”

[5] Yahoo Finance: $285B Wiped From U.S. Software Sector After Anthropic Shock

[6] Yahoo Finance: IGV ETF Technical Analysis

[7] Axios: Hedge Funds Aggressively Shorting Software Sector

[8] Benzinga: Misinterpretations Driving the Software Sector Collapse

[9] Salesforce Investor Relations: Record Q3 Earnings Driven by Agentforce

[10] Futurum Group: ServiceNow Q4 Earnings & AI Platform Momentum

[11] Fortune: The AI Paradox and Irrational Analysis

[12] CNBC: Software Stocks Enter Bear Market, ServiceNow Among Sharply Down Names

[13] StockAnalysis: Datadog Operational Metrics

[14] Meritech Capital: Retrospective on the 2022 SaaS Crash

[15] CNBC: NVIDIA Plunges Amid DeepSeek Concerns

[16] Fortune: Deutsche Bank on Software Stock Bubble and Dot-Com Era Parallels

[17] Deutsche Bank Jim Reid Research Report