From Data Scarcity to Decision Abundance

Why 95% of AI Projects Fail, and How to Break Free from the House of Cards

Two Studies, One Message

In late 2025, two important and seemingly unrelated pieces of research landed almost side by side.

On one side, MIT’s State of AI in Business report revealed a sobering truth: 95% of enterprise AI initiatives fail to deliver measurable ROI. Only a small fraction reach production, and even fewer change decision-making at scale. The headline is stark: AI adoption is wide, but impact is painfully narrow.

On the other side, Gartner’s Market Guide for Data Management Platforms highlighted the ballooning cost and complexity of today’s data stack. Six distinct categories: data integration, data quality, metadata, master data, persistence, and governance. Collectively, they represent $133 billion in spend today and will add another $100 billion by 2029, bringing the total to $233 billion. And all of this is before AI workloads even fully enter the picture.

At first glance, these two studies speak to different problems: AI failure rates versus the size of the data management market. Together, they tell one story. The fragile, fragmented data infrastructure of the past decade is a primary reason AI struggles to deliver value today.

This Is About Decisions, Not Data

Before the details, it’s worth reminding ourselves what all this investment is actually for. Data, analytics, and AI exist for one purpose: to help leaders make better decisions.

The point is not to build bigger databases, more elaborate pipelines, or shinier dashboards. The point is to reduce decision latency (how long it takes to act), increase decision coverage (how much of the business is instrumented), and raise decision confidence (how much you trust a recommendation).

Somewhere along the way, the industry lost sight of this. It became obsessed with its own categories, tools, and architectures, almost like a hall of mirrors. MIT’s findings on AI failure and Gartner’s picture of the bloated stack are both symptoms of this deeper problem.

The House of Cards

Gartner’s map of the data management market is impressive at first glance. Six well-defined categories. Thousands of SKUs. Tens of billions of dollars in annual revenue.

Look closer, and you see a house of cards. Each category requires experts to operate: data scientists, data engineers, analysts, governance officers. None of these systems generates value on its own. Only when stitched together, at high cost and fragility, do they approximate the “platform” leaders believe they’re buying.

The scale of the problem:

• $133B spent in 2024, growing to $233B by 2029.
• More than 1,000 SKUs, each solving a narrow slice of the puzzle.
• Roughly three times the hidden human cost for every $1 in software, spent on headcount just to keep the stack running.

This is not scale. This is the maintenance of a brittle structure that can’t keep pace with the business.

How Scarcity Took Hold

When the stack is this complex and expensive, organizations develop a scarcity mindset. They ration ambition. Instead of opening access to all data, they pick and choose narrow use cases, just enough to justify spend, but never enough to change the enterprise.

This was the legacy of the Big Data era: sprawling infrastructure that forced leaders into trade-offs and left most of the potential value locked away. And now, as MIT shows, the same scarcity mindset is bleeding into the AI era.

Evidence the Model Is Breaking

MIT’s State of AI in Business provides hard evidence:

• 95% of AI projects fail to deliver ROI.
• Most stall in pilot phases, unable to scale beyond narrow use cases.
• The root problem is not algorithms, but integration, context, and learning.
• Shadow AI, where individuals succeed with flexible tools, shows that adaptability and memory matter more than rigid infrastructure.

Gartner reinforces the point: consolidation may trim the SKU count, but it can’t fix a system designed around silos. The house stays brittle, fragile, and expensive.

Together, these findings confirm what many CFOs already feel: headcount and budgets for data and analytics have bloated, while the impact on decisions stays minimal.

Root Cause: Square Peg, Round Hole

AI is being forced into the same mold as big data:

• Narrow use cases to contain cost.
• Layers of gatekeeping to ensure governance.
• Dashboards and reports as the main outputs.

But AI does not thrive in narrow use cases. It thrives on context, on broad access to data, on the freedom to find patterns humans would miss. Constraining AI into the old scarcity model is like hammering a square peg into a round hole.

Press Reset: Decision-First Architecture

The alternative is not better integration or smarter governance. The alternative is rethinking the stack from the ground up, not data-first, but decision-first.

Principles:

• Connect to truth where it already lives, in SaaS tools and enterprise data warehouses, not new lakes and warehouses.
• Model decisions, not tables. Organize around the outcomes leaders care about.
• Self-healing, agentic learning loops: AI that retains memory, improves with feedback, and recommends actions on its own.
• Natural language access: machine learning to analyze, language models to explain and recommend.
• Guardrails, not gatekeeping: governance built into the flow, not bolted on afterward.

From Scarcity to Abundance

Instead of rationing use cases, give AI all the data. Let algorithms assess quality, relevance, and patterns. Then let language models translate those findings into clear recommendations.

The effect is to move from scarcity (choosing what not to look at) to abundance (analyzing everything and surfacing what matters). This is the path to the 5% club, the minority of companies that actually achieve ROI from AI.

Vibe Analytics: The SQOR.ai Paradigm Shift

If you’ve heard of vibe coding, think of SQOR.ai as what MIT called its first cousin: vibe analytics.

“Vibe analytics transforms data queries to conversations, and dashboards to jam sessions.” — MIT Sloan Management Review

Where vibe coding is about new forms of human-AI interaction, vibe analytics is about changing how businesses interrogate and learn from their data. Instead of teams of data scientists orchestrating every query, leaders can simply ask, and get trusted, deterministic answers in seconds.

SQOR.ai is a Vibe Analytics and Decision Intelligence platform: an autonomous intelligence that lets you ask your data in plain English and get trusted, deterministic answers in seconds, all without the infrastructure or the armies of data experts that business intelligence requires today. The platform is self-healing, so it learns and improves over time.

At SQOR.ai, we believe in a three-step ethos for decision intelligence:

• Sign up.
• Authenticate your data sources.
• Just ask questions.

The responses shouldn’t only answer, they should surprise, opening visibility across silos and delivering clarity leaders and teams can act on immediately.

On speed and value creation, MIT documents a telling pattern: more financially relevant, data-driven insights surfaced in a single mediated 90-minute session than the organization typically generates in 90 days.

On data quality, the rule still applies: there is no “garbage in, vibes out” loophole. With SQOR.ai, that doesn’t mean more work pushed onto executives. Data quality becomes a management discipline at the source rather than a downstream clean-up project. Leaders set clear expectations that teams use their SaaS tools correctly so execution data is captured faithfully in the flow of work.

“SQOR.ai is returning control of the company to the fiduciaries. In the Big Data era, fiduciaries lost control to the data science world, which too often wagged the dog. We must move away from a mindset of data scarcity, which holds back where AI excels, and embrace decision abundance, so AI can do its thing.” — Laz Fuentes, CEO of SQOR.ai

MIT also warns that this shift requires new technology foundations. Agentic AI, the class of systems that embeds persistent memory and iterative learning by design, directly addresses the learning gap that defines the divide. And MIT confirms the scale of the problem: 95% of organizations are getting zero return.

The objective is threefold:

• Help increase revenue by enabling sharper, faster decisions.
• Help reduce cost by cutting the demand on data teams.
• Help mitigate risk by making sure no critical data source is left out of view.

This isn’t about replacing data teams. It’s about freeing them. SQOR.ai becomes the data team’s best friend, absorbing the bulk of routine questions and analysis, while giving business units the answers they need in seconds.

This is the era of vibe analytics: clarity, visibility, and decision abundance.

CFO Math

Compare today’s stack, $233B in software plus roughly three times that in headcount plus governance overhead, against a decision-first architecture: direct connectors, automated KPI extraction, and self-healing agentic AI loops. Lower cost, faster answers, broader coverage. This is budget well spent, and it’s the path to the 5% club.

Press Reset and Start Vibing on Analytics

So much of the industry’s energy has gone into maintaining the house of cards: six categories, thousands of SKUs, billions in spend, and the illusion that more tooling would solve the decision gap. The result has been bloat, fragility, and failure. MIT showed it clearly: 95% of AI projects never deliver ROI. Gartner showed why: a stack designed for its own survival rather than for the users it was meant to serve.

The answer is not another SKU, another integration, or another roadmap. The answer is a reset. A shift from scarcity to abundance, from orchestration to immediacy, from dashboards to vibe analytics.

Companies that cling to the old paradigm will stay stuck in the 95%. Those that embrace abundance will join the 5% club, not by accident, but by design.

The house of cards is already swaying. It’s time to press reset, and start vibing on analytics.

Sources: Gartner, “Market Guide for Data Management Platforms”; MIT NANDA, “State of AI in Business 2025”; MIT Sloan Management Review, “Vibe Analytics: Vibe Coding’s New Cousin Unlocks Insights.”