Seventy-four. That's the number Salesforce is using to tell a much larger story. In a company of roughly 75,000 employees, seventy-four people is a rounding error — less than one-tenth of one percent of the global workforce. Yet the announcement landed with the careful weight of a quarterly earnings call: another headcount reduction, the fourth in less than a year, this time branded as "AI-driven transformation." I watched the press release ripple through my timeline, and I couldn't shake a familiar feeling. We've seen this film before in crypto — the sharp turn where a small technical adjustment gets dressed in the language of revolution. The headline said efficiency. The subtext said survival. And the real story was the gap between those two things.
Before we go any further, let's be honest about what we know and what we don't. Salesforce is a cloud-based CRM giant, a company that has acquired Slack, Tableau, MuleSoft, and ExactTarget, and whose product suite spans sales, service, marketing, and platform clouds. It's a mature enterprise SaaS company, not a startup chasing hypergrowth. The fourth round of layoffs in fewer than twelve months is a fact. The number of affected employees, seventy-four, is a fact. The official narrative, that AI is fundamentally reshaping the enterprise and forcing a shift from "chasing headcount" to "chasing efficiency," is a framing — a piece of narrative architecture that deserves the same scrutiny we would give to a whitepaper claiming to be "decentralized because we said so."
In the crypto world, words like "decentralized," "trustless," and "community-owned" are often used as placeholders for real structural guarantees. We've seen protocols claim decentralization while their governance tokens were concentrated in a few wallets. We've seen stablecoins claim stability while their reserves were, shall we say, creatively collateralized. And we've watched Tether control over seventy percent of the stablecoin market while its reserves have never received a fully clean, fully independent audit — a problem the entire industry pretends not to see. Why do we ignore it? Because the narrative is more comfortable than the audit. The same thing is happening now in the legacy tech world. Salesforce is telling us that AI is making jobs obsolete, that "efficiency" is the new corporate religion. But I've audited enough systems to know that you should never trust a claim you haven't poked at from every angle. You should always ask: who benefits from the story, and what data would prove it?
Let me start with the data I know. In my years as a data scientist and later as a protocol project manager, I've learned that materiality is a function of context, not of absolute numbers. A seventy-four-person layoff at a company with a seventy-five-thousand-person headcount is immaterial to the income statement. It won't move quarterly margins. It won't trigger WARN Act notifications if the cuts are spread across multiple states. It won't even register as a line item in the annual report. So why announce it at all? The answer is, materiality in the narrative economy is different. Four rounds of layoffs in a year — even small ones — create a cadence, a rhythm that tells investors, employees, and customers a very specific story: the company is disciplined, the company is serious about margins, the company is willing to make painful choices to protect profitability. That's the signal. And in that signal, we can find the same pattern I've seen in DeFi protocols adjusting their interest rate curves.
When a company says "AI drove these layoffs," it's a claim about causality that needs evidence, not a press release. In DeFi, we have a similar problem. Aave and Compound's interest rate models are completely arbitrary in many cases — they have nothing to do with real market supply and demand. If you dig into the formulas, you'll find inflection points and slope parameters that were chosen by a developer on a Tuesday afternoon, not derived from a deep analysis of liquidity shocks. Yet we treat those curves as if they were natural laws. Similarly, the "AI efficiency" narrative has become a kind of natural law in corporate communications. The logic goes: AI is getting better, so human labor becomes less necessary, and therefore layoffs are technological progress. But where's the evidence that the specific functions being cut are actually being replaced by AI? The press release doesn't tell you. The fourth-round seventy-four employees — were they in sales? In customer success? In engineering? We don't know. Without that detail, the AI attribution is a black box, or more accurately, an opaque oracle.
Let me bring in some of my own background. In 2016, before the crypto boom and bust cycles that would define the next decade, I was still a data scientist in Buenos Aires, spending my nights in Hyperledger meetups. I was one of a handful of women in a room of male cryptographers, and I learned quickly that the most powerful tool I had was not the ability to write smart contracts, but the ability to translate technical concepts into human values. When I wrote my Spanish-language tutorial on "Trustless Collaboration" and hit ten thousand readers, I felt the first stirrings of what I now call the narrative economy: money flows to stories that people believe. Code alone doesn't drive adoption. Adoption is driven by the story we tell ourselves about what the code will do to our lives. This is why I pay such close attention to phrases like "AI-driven transformation." It's not that I dismiss the possibility that AI is enabling productivity gains. It's that the phrase is doing so much work — covering up cost reduction, covering up strategic retreats, covering up a reaction to macroeconomic pressure — that it has become a kind of narrative Ponzi scheme.
Now, let's talk about what actually drives this cycle, because I've lived through similar moments in the crypto bear market. The source article's skeleton tells us that Salesforce is likely under pressure to maintain the Rule of 40 — the benchmark where a SaaS company's revenue growth rate plus profit margin should be at least forty percent. When growth slows to single digits or low teens, the only way to keep the Rule alive is to cut costs. Layoffs are the fastest cost-cutting tool. This is not an AI story. This is capitalism doing what capitalism does when the environment tightens. In 2022, after the Terra/Luna collapse, I spent months mediating between two hundred core contributors in a struggling DAO. The treasury was bleeding, and the immediate instinct was to slash contributor roles. Sound familiar? The arguments were the same: we need to be lean, we need to be disciplined, we need to survive. But the way we implemented the cuts mattered. We designed a "Values-First" governance framework that asked every department to rank its projects not just by revenue contribution, but by their alignment with the community's core mission. We didn't just cut headcount; we cut commitments that weren't serving the people who had entrusted us with the protocol. The result? Internal toxicity dropped by forty percent over three months, and we rebuilt trust among contributors who had lost everything. That experience taught me that efficiency without a human anchor is just extraction.
Connect first, transact second. Always. That's a principle I've carried from those DAO mediation rooms into every subsequent project, and it is the lens through which I want to read Salesforce's latest move. When an enterprise cuts seventy-four jobs, the immediate risk isn't the reduced headcount. It's the signal sent to the people who remain. It's the silent fear among customer success managers that they might be next. It's the diminishing psychological safety that causes your best talent to update their LinkedIn profiles. Seventy-four departures can easily become seventy-four notifications racing through the corporate grapevine, each one carrying a different version of "I should start looking." The hard cost of that cultural decay is not captured in quarterly earnings. It's an off-chain variable, a reputational liability that doesn't appear on any balance sheet — the same way that a DeFi protocol's real risk is not the code bug that everyone sees, but the invisible governance capture that nobody bothers to audit.
Let me be more precise about the economics, because I think the analogy to crypto is even tighter than it appears at first glance. In a mature SaaS company like Salesforce, the most valuable revenue stream is expansion revenue — upsells and cross-sells to existing customers. That revenue depends on the health of the customer success team. If the layoffs touch customer success, net revenue retention (NRR) could start to slip. In a frictionless world, AI would replace the customer success manager with an intelligent agent that answers customer queries instantly and proactively suggests new features. But we are not there yet. Enterprise sales cycles involve long procurement processes, multiple stakeholders, and political nuance that no LLM can replicate. If Salesforce cuts customer support roles to fund an AI moonshot, it risks degrading the very relationship that keeps the moat intact. The switching cost for a major CRM implementation is high, but so is the cost of a neglected enterprise client. In crypto, we see the same dynamic when a protocol chooses to cut security spending to increase tokenholder returns. It works for a quarter, and then someone drains the treasury. The optimization that looks good in a dashboard is often the one that breaks the system.
Efficiency that externalizes costs to users is not efficiency; it's rent extraction. I came to this realization during the 2020 DeFi Summer, when I led community education for Aave's beta launch across Latin America. I organized twelve live workshops, taught five thousand retail users how to navigate smart contract risk, and watched support tickets drop by thirty percent simply because people finally understood what they were signing. That experience ground a simple truth into me: most "user error" is actually "system opacity." When an enterprise or a protocol relies on hidden assumptions, the cost of those assumptions is paid by the user, not the operator. Salesforce's "AI-driven" narrative is a form of opacity. The company is asking us to accept that machine learning is replacing job functions without giving us the data that would prove the same output is being produced at lower cost and equal quality. In contrast, imagine if Salesforce published an auditable AI efficiency report — like the on-chain reserves transparency we demand from Tether. We would demand proof that every terminated role was genuinely automated, with measurable performance benchmarks. Tether has never faced that level of independent verification, and Salesforce won't either — unless we start asking the question.
Because, and here is the uncomfortable truth for both our industries, narrative alone does not sustain long-term compounding value. I've been observing this space for over a decade, and I've seen countless projects promise a "paradigm shift" while their underlying product remained a shell. The ones that survived are the ones that let the narrative catch up to the code, not the ones that pretended code could replace the need for narrative. This is why I am watching the current AI layoff wave with a mix of curiosity and dread. The same tech giants that are cutting thousands of jobs are simultaneously promising that agentic AI will transform business processes. Maybe they're right. Maybe AI really will absorb the low-level work and unlock a new era of human creativity. But if that's the case, where are the productivity metrics? Where are the before-and-after charts? Where is the on-chain oracle of corporate performance that allows stakeholders to verify whether the story is real? We have the tools to build such accountability — that's precisely what blockchain promises — but the industry refuses to use them on itself.
Let's take a step back and think about the competitive landscape, because this is where the crypto parallel is most instructive. Salesforce competes with Microsoft Dynamics 365, HubSpot, ServiceNow, and a dozen nimble startups. Hiding behind an "AI transformation" narrative while delivering shrinking product support creates an opening. Similarly, in the Layer 2 world, every rollup project after the Dencun upgrade claimed that blob space would be cheap forever and that transaction fees would remain at historic lows. I've been saying this since the upgrade: post-Dencun blob data will be saturated within two years. When that happens, rollup gas fees will double again, and every project that based its user acquisition strategy on permanent low fees will be caught with their pants down. The AI narrative in legacy SaaS has the same trajectory. Right now, "AI-driven efficiency" is a cheap talking point. But when the macroeconomic environment stabilizes, and companies have to show real revenue growth, the narratives that were built on nothing will collapse. The Salesforce case is a classic example: a seventy-four-person layoff is the kind of tiny, surgical cut that either signals healthy pruning or the beginning of a slow bleed. We won't know which until we see the next quarter's AI product revenue and the retention metrics for enterprise accounts.
I need to pause here and speak directly to the founders and protocol builders who might be reading this. I know the temptation to take shortcuts. I know that in the bear market, the first instinct is to cut everything that doesn't show a short-term ROI. But let me tell you what I learned from the months after the Terra collapse. When we designed the Values-First governance framework for that struggling DAO, we did not start with the treasuries. We started with the people who were still in the room. We asked the hard questions: Which roles are truly essential to the community's mission? Which functions can be automated without destroying trust? And then we communicated, openly and painfully, every step of the way. The result wasn't a bloodbath; it was a cleanup. We emerged with a leaner, stronger core. Connect first, transact second. Always. If Salesforce is applying that principle internally, if they are carefully ensuring that the seventy-four cuts are in redundant back-office functions while protecting customer-facing roles and AI research, then I would call this a healthy adjustment. But the public framing matters. Calling it "AI transformation" instead of "cost discipline" is like labeling a round of token buybacks a "community incentive program." It's not inaccurate on its face, but it's purposefully misleading in its implications.
Let's also talk about the employee experience, because I can't separate the technical analysis from the human one. I am a fifty-one-year-old woman who has built a career in a male-dominated industry by being more prepared, more technical, and more empathetic than the metrics require. I've been in rooms where people assumed I didn't understand the smart contract syntax, and I've been in rooms where people assumed that because I cared about community safety, I couldn't care about scalability. I've learned to hold both truths at once: systems are made of code, but they are lived by humans. When a company announces a fourth round of layoffs in a year, the people who remain are not thinking about the Rule of 40. They are thinking about whether their project will be discontinued, whether their manager recognizes their contribution, whether they should start interviewing elsewhere. The sixty-thousand-foot view is useful, but it is not the whole story. The whole story includes the customer who was just getting used to their integration specialist and now has to explain their workflow to a chatbot. It includes the salesperson who was hired six months ago and now worries they've joined a sinking ship. It includes the seventy-four individuals, each with a different rent payment, different family member's medical appointment, different dream. The narrative economy often forgets that the tokens of exchange are human lives.
In the blockchain industry, we have somehow convinced ourselves that decentralized protocols can automate trust and eliminate the need for intermediaries. To a certain extent, that is true. Smart contracts really do execute without a human bridge. But the moment you look at the governance, the proposal process, the community support, you find humans making decisions, humans dealing with disputed claims, humans maintaining the front-end dashboard. The same is true for a centralized company like Salesforce. Even as AI takes over more processes, there will always be a shadow layer of human judgment, human accountability, and human care. The danger is when corporate leaders use the AI narrative to abandon that human layer prematurely. They cut seventy-four people, claim the AI agent will handle the load, and then a month later discover that a major enterprise client is frustrated because they can't get a straight answer. The hidden cost of premature automation is deferred, but it always arrives.
I've been collecting examples of this dynamic inside crypto for years. Take the stablecoin market. USDT is the dominant stablecoin, with over seventy percent of the market share. Tether's reserves have never had a truly independent audit. The entire industry pretends this problem doesn't exist because the alternative would be to stop using USDT, and that would be inconvenient. So we keep pretending. In the same way, the enterprise software industry pretends that the AI transition is smooth and inevitable. The data I would want is public, auditable evidence that AI tools are truly replacing the specific functions performed by the terminated employees. That evidence simply doesn't exist. A press release that says "AI changed our operating model" is not evidence. It's a hypothesis.
This brings me to the contrarian angle, because I don't want to make this a simplistic "big tech is evil" piece. There is a version of this story where Salesforce is doing exactly the right thing. The tech industry overhired during the pandemic-fueled growth spurt. Everyone knows it. Every CFO in America is looking at their cost structure and realizing that they have two people doing the work of one and a half. In that context, a sequence of small, focused layoffs is better than one massive bloodbath. It gives the organization time to adapt, to reallocate roles, to test where AI can genuinely be useful. It allows the company to prune dead branches before they rot. The contrarian part of me — the part that has audited bloated DAO treasuries and watched vaporware protocols burn through millions of dollars without any discernible product — understands that discipline is essential. If Salesforce is using this moment to cut low-performing business units and double down on Agentforce, Einstein, and actual AI products that customers are willing to pay for, then the seventy-four-person layoff is a sign of health, not decay. In crypto, we should be equally willing to shut down zombie chains that have no usage and no reason to exist. Efficiency is not the enemy. The enemy is narrative dishonesty.
Let me be more specific about what a genuinely honest AI transformation announcement would look like. It would state the exact number of roles eliminated, the departments, the functions, and the specific AI replacement that now handles each function. It would include measurable outcomes: customer support tickets resolved per day, revenue per sales representative, deployment times for new integrations. It would commit to publishing a quarterly transparency report on AI efficiency, perhaps even on-chain, so that any stakeholder could verify the claims. Is that too much to ask? In the crypto world, we ask much more of our DeFi protocols. We demand proof of reserves, proof of governance, proof of uptime. The fact that a centralized trillion-dollar ecosystem can operate on press releases is a failure of our own standards. We should demand the same rigor from the companies that are using "AI" as a justification for dismissing people. If they ask us to trust the algorithm, we should ask to see the algorithm's report card.
Now, post-Dencun, there is a specific technical lesson I keep returning to when I think about capacity planning. We learned that blob space is a scarce resource, and no amount of "efficiency" can repeal the arithmetic of supply and demand. Within two years, the blob space will be saturated, and rollup gas fees will double. I believe that is as close to a deterministic outcome as we can get in blockchain. The same arithmetic applies to the labor market. There is only so much skilled human capacity in the world. If the tech industry convinces itself that AI is an infinite substitute for human labor, it will be shocked when the remaining humans have too much compliance, too much integration complexity, and too much interpersonal nuance to fit into an automation pipeline. The companies that survive this era will be the ones that treat AI as a tool to augment human capability, not as a cover story for cost-cutting. And the way to tell which is which is to look at the data. If the AI model outputs are audited, if the product revenue is growing, if customer satisfaction is maintained, then the layoffs were probably a necessary reallocation. If the only evidence is a press release, then we are in the realm of narrative theater.
Connect first, transact second. Always. It's a phrase that reminds me that every technical system exists within a social context. Salesforce may be a centralized enterprise, but its actions shape the economic reality of thousands of people. The same is true for the decentralized protocols I work with. Whether you call it a company or a network, the human connection is the foundation of lasting value. If Salesforce wants to lead, it should not hide behind "AI transformation." It should stand behind the decisions, show the receipts, and make the narrative consistent with the observable truth. That is the only way to earn trust in an economy that is already skeptical of smooth stories.
Let me end with a forward-looking judgment. I believe that the era we are entering will be defined less by the capability of AI and more by the accountability of the narratives surrounding AI. The companies that flourish will be those that adopt the transparency principles that the crypto community claims to hold dear — not the bloated, over-hyped version of crypto, but the original cypherpunk ethos: verify, don't trust. For Salesforce, that means publishing AI product metrics alongside layoff announcements, being honest about the distinction between efficiency and cost reduction, and preserving the human touch that enterprise clients actually pay for. For the crypto industry, it means holding the mirror to itself, insisting on independent audits for Tether, and admitting that decentralization is a spectrum, not a binary. If we do that, then the seventy-four cut jobs may be remembered not as another depressing chapter in the AI dystopia, but as a moment when we collectively asked for the evidence behind the story. And the industry was wise enough to give it to us.
