Divya van Mahajan

The Ai Roi Surprise Wharton Finds 75 Of Enterprises Seeing Positive R

ai-daily-brief-podcast

Study Document: AI ROI Reality Check — Wharton Finds 75% of Enterprises Seeing Positive Returns

Overview

This episode of the AI Daily Brief (dated November 8, 2025) covers two primary topics: (1) a headlines segment reviewing major AI industry and market developments, and (2) a deep-dive into the third annual Wharton GBK Study of Enterprise Generative AI, which finds that three-quarters of enterprises are already seeing positive ROI from AI investments. The host contrasts this underreported Wharton study with the widely cited (and methodologically criticized) MIT study that claimed 95% of AI initiatives were failing. The episode argues that genuine, measurable AI returns are arriving faster than many expected, even as market sentiment remains volatile.

Source video URL: Not available

Prerequisites

  • Basic familiarity with Generative AI (Gen AI) concepts and enterprise software adoption cycles
  • Understanding of ROI (Return on Investment) measurement in corporate contexts
  • Awareness of the current AI investment landscape, including major players (Anthropic, OpenAI, Palantir, Nvidia)
  • Familiarity with AI agent/agentic AI concepts — software that acts autonomously on behalf of users
  • General knowledge of financial markets, including ARR (Annual Recurring Revenue), free cash flow, put options, and short selling
  • Background on the dot-com bubble as a reference framework for evaluating AI bubble arguments

Main Points

1. Anthropic’s Financial Projections Signal Unrelenting Growth

  • Anthropic expects $70 billion in revenue and $17 billion in positive cash flows by 2028, per reporting from The Information.
  • Current trajectory: $9 billion ARR by end of 2025; $20–26 billion revenue target set for 2026.
  • Claude Code is generating $1 billion in annualized revenue — more than double its July pace.
  • Gross profit margins projected at 50% this year, rising to 77% by 2028; positive free cash flow expected in 2027.
  • Strategy validation: simultaneous pursuit of foundation models, the application layer, and enterprise partnerships (e.g., org-wide deployments for Deloitte and Cognizant covering hundreds of thousands of seats).
  • A new fundraising round is anticipated, potentially targeting a valuation of $300–$400 billion, up from $170 billion in September 2025.

2. AI Stock Market Volatility and Bubble Debate

  • AI stocks led a 2% NASDAQ decline mid-week, attributed to a mix of government shutdown concerns, trade uncertainty, and real economy deterioration.
  • Goldman Sachs CEO David Solomon and Morgan Stanley CEO Ted Pick both acknowledged the possibility of 10–15% drawdowns, though neither predicted a structural collapse.
  • Pinterest stock fell 21% after weak guidance and warnings about tariff-related ad spending, even as its CEO cited AI-driven progress — illustrating that investors now demand bottom-line ROI, not just AI adoption narratives.
  • Invesco CEO Andrew Schlossberg suggested the market is closer to a correction than a further 10–20% rise.

3. Michael Burry’s AI Short and Industry Pushback

  • Burry’s hedge fund, Scion Asset Management, disclosed roughly $1 billion in put options on Palantir and Nvidia — approximately 80% of the fund’s value — betting on an AI bubble collapse.
  • The disclosure came two weeks ahead of the regulatory deadline, suggesting an intentional narrative move.
  • Palantir CEO Alex Karp and Nvidia CEO Jensen Huang both publicly rejected the bubble thesis:
    • Huang: “This is the beginning of the build-out… AI is now so productive, so profitable, and used by so many people.”
    • Karp: “The two companies he’s shorting are the ones making all the money.”
  • Counterpoint: Burry has called for market crashes in 2015, 2017, 2019, 2020, 2021, and 2023. The S&P 500 is up 71% since his 2023 “sell” call.
  • Deutsche Bank is separately considering shorting AI stocks as a hedge against its own large AI data center loan exposure — a sign that institutional risk management around AI infrastructure is evolving.

4. AI Infrastructure Financing Shifts Toward Debt Markets

  • AI infrastructure has historically been cash-flowed by hyperscalers (Amazon, Microsoft, Google), but the scale of continued build-out will increasingly require debt financing.
  • BlackRock’s global head of tech, Tony Kim: “Tech companies will have to shed their aversion to leverage.”
  • Deutsche Bank has already extended billions in loans to data center projects and is expanding lending to smaller neoclouds beyond the major hyperscalers.
  • Debt instruments being explored include synthetic risk transfers (SRTs) and shorting baskets of AI-related equities.
  • Snap–Perplexity deal: Perplexity will pay $400 million in cash and equity to integrate its AI search into Snapchat’s chat function, gaining access to nearly 500 million daily active users. Snap stock rose 25% in aftermarket trading.
  • Amazon lawsuit: Amazon is suing Perplexity, alleging its web crawlers failed to identify themselves as AI agents while scraping Amazon’s e-commerce platform.
    • Amazon’s framing: Perplexity’s actions are analogous to trespass; third-party agents must operate transparently.
    • Perplexity’s framing: Amazon is using bully tactics to block users’ right to employ agentic AI on their behalf.
  • The dispute is framed as a preview of a broader battle over agentic shopping, particularly relevant ahead of Black Friday, when multiple AI labs plan to deploy shopping agents.

6. The Wharton GBK Study: AI ROI Moving Into Focus (Main Episode)

Methodology and Context

  • Third annual longitudinal study from Wharton GBK; surveyed approximately 800 enterprise leaders across multiple functions and firm sizes.
  • Contrasted with a criticized MIT study (52 executives, convenience sample, public earnings statements) that generated the widely repeated claim that “95% of AI initiatives are failing.”

Theme 1 — Everyday AI: From Curiosity to Core Workflow

  • 82% of enterprise leaders now use Gen AI weekly; 46% use it daily (up 17 percentage points year-over-year).
  • 77% report being at least somewhat familiar with Gen AI.
  • Top 10 Gen AI use cases in 2025 (in ranked order):
    1. Data analysis and analytics
    2. Document and meeting summarization
    3. Document and proposal editing and writing
    4. Presentation and report creation
    5. Idea generation and brainstorming
    6. Marketing content creation
    7. Customer service and support
    8. Email generation
    9. Internal support and help desk
    10. Sales content creation
  • Half of the top 10 use cases directly boost employee productivity.
  • 58% of enterprises are testing AI agents, primarily for process automation, analytics, and workflow orchestration.

Theme 2 — ROI Measurement Becoming Standard

  • 72% of companies are formally tracking Gen AI ROI.
  • Functions leading in structured ROI tracking: HR (84%) and Finance (80%).
  • 74% of enterprises report either moderately or significantly positive ROI.
  • Smaller firms ($50M–$2B revenue) are seeing higher ROI so far than large enterprises ($2B+ revenue).
  • 89% say AI has enhanced skills; 43% still fear skill decline.
  • 88% anticipate increasing Gen AI budgets in the next 12 months.
  • 4 out of 5 see Gen AI investments paying off within 2–3 years.

Theme 3 — 2026 Outlook: Performance at Scale

  • Study authors describe 2026 as potentially a shift from “accountable acceleration” to “performance at scale” — where ROI metrics, playbooks, and guardrails enable enterprises to rewire core workflows and deploy agentic systems.
  • Key unresolved challenge: talent acquisition and skill development for the AI era.
  • The host introduces an AI ROI benchmarking study (roisurvey.ai) with over 700 contributed use cases, aimed at providing peer-comparison benchmarks across benefit types: time savings, cost savings, new capabilities, enhanced throughput, reduced risk, improved decision-making, and revenue impact.

Key Concepts

  • ARR (Annual Recurring Revenue): A normalized measure of predictable revenue generated from subscriptions or contracts on an annual basis.
  • Free Cash Flow: Cash a company generates after accounting for capital expenditures; a key profitability indicator.
  • Put Option: A financial contract giving the holder the right to sell an asset at a predetermined price, used here as a bearish bet that a stock’s value will decline.
  • Synthetic Risk Transfer (SRT): A financial derivative used by banks to transfer credit risk on loan portfolios to other investors without selling the underlying loans.
  • AI Agents / Agentic AI: AI systems capable of taking autonomous actions — browsing, purchasing, orchestrating workflows — on behalf of users.
  • Gen AI (Generative AI): AI systems that generate content (text, images, code, etc.) in response to prompts, including large language models like Claude and GPT.
  • Neoclouds: Smaller, specialized cloud infrastructure providers distinct from major hyperscalers (AWS, Azure, Google Cloud), increasingly involved in AI workloads.
  • ROI Benchmarking: The practice of comparing an organization’s AI returns against industry peers to assess relative performance.
  • Wharton GBK Study: A longitudinal annual survey of approximately 800 enterprise leaders conducted by Wharton School researchers, tracking Gen AI adoption and ROI trends.
  • Hyperscalers: Major cloud platform providers (Amazon, Microsoft, Google) that build and operate massive-scale infrastructure.
  • SRT (Synthetic Risk Transfer): See above; a mechanism for banks to hedge credit exposure using derivatives.

Summary

The central argument of this episode is that genuine, measurable AI ROI is arriving at scale in the enterprise — a story that is significantly underreported relative to skeptical narratives. The Wharton GBK study, now in its third year and grounded in a rigorous survey of approximately 800 enterprise leaders, finds that 74% of enterprises are seeing positive ROI from Gen AI, 82% use it weekly, and 88% plan to increase budgets over the next year. This stands in sharp contrast to the widely circulated MIT study, which the host argues was methodologically weak yet became the dominant narrative. Framing the broader context, the episode documents how Anthropic’s financial projections ($70B revenue by 2028, a path to 77% gross margins) and OpenAI’s growth signal that AI is generating real economic value — even as markets wobble, Michael Burry places large bearish bets, and debates about a potential AI bubble intensify. Subplots around Perplexity’s Snap distribution deal and Amazon lawsuit preview the coming battles over agentic commerce. The host’s overall message is that 2026 will be defined not by whether AI delivers ROI, but by how enterprises measure, benchmark, and scale that ROI to compete effectively with peers.