Top Analysts Reports on AI

Last updated: 02 Apr 2026

AI Spend vs Economic Output global trend

Data last updated: —.

Reasoning

Important note: There is no single official yearly global series for “AI’s real economic output value.” For the second line, this chart uses the most defensible observable proxy: global AI market output/value. The proxy is anchored to UNCTAD’s estimate of $189B in 2023 and $4.8T in 2033, then converted into an annual path; that keeps the chart tied to an intergovernmental source rather than low-quality market blogs.

For the investment/spend line, the series uses Stanford HAI’s yearly global corporate AI investment figures for 2015–2024. Those numbers are useful because they are annual, global, and broad enough to include private investment, M&A, minority stakes, and public offerings. Stanford reports $252.3B in 2024, after a spike to $360.7B in 2021 and a pullback in 2022–2023.

For 2025–2028 investment/spend, the projection is aligned to IDC’s global AI spending outlook, which says the market is nearly $235B in 2024 and will rise to over $631B/$632B by 2028, including AI-enabled applications, infrastructure, and related IT/business services. The series then extends 2029–2035 with a tapering growth curve rather than a flat CAGR, since very high early-stage growth often moderates as markets mature.

For the output line, the chart does not use a speculative “AI GDP” number for each year. Instead, it uses global AI market output/value as a practical proxy. Anchor points come from UNCTAD: $189B in 2023 and $4.8T in 2033. Those are converted into an annual series, then growth slows slightly after 2033 to avoid an unrealistic straight-line extrapolation.

Macro studies broadly support the shape of the output line even though they use different definitions. PwC says AI could lift global economic output by up to 15 percentage points by 2035; Goldman Sachs says generative AI could raise global GDP by about 7% over a decade; McKinsey estimates generative AI could add $2.6T–$4.4T annually; and the IMF suggests AI could raise global output by about 0.5% annually from 2025 to 2030. These are not the same metric as “market output,” but they support a material steepening through the 2030s.

A second sanity check is capital intensity. IDC notes hardware is a major AI spending bucket after software, and Reuters (citing S&P Global) reports Big Tech AI infrastructure spending rose from about $80B in 2019 to $383B in 2025, with $635B planned in 2026. OECD also reported AI firms captured 61% of global VC investment in 2025. This supports the idea that the spend line remains strong even if it is more volatile than the output proxy.

Sources used in chart

    AI chips + DC sales & bookings global trend

    Data last updated: —.

    Reasoning

    Key caveat: Exact 10-year public series for “GPU-specific revenue” and “AI-specific datacenter revenue” do not exist consistently across these firms, so this chart uses the closest public proxies.

    • Chip curve: NVIDIA Data Center, AMD Data Center, Intel DCAI/DCG, Broadcom AI semiconductor revenue, Marvell AI revenue.
    • Datacenter curve: total revenue for the most AI-exposed operators (Equinix, Digital Realty, GDS, VNET, CoreWeave), because operators rarely break out AI-only revenue.

    In the model, the chip/hardware curve rises from $17.58B in 2016 to $170.23B in 2025, then to $1.28T by 2030; the datacenter-operator curve rises from $6.43B to $23.52B to $62.70B. The big driver is disclosed AI demand and guidance: AMD cited a $100B annual data-center chip revenue target within five years; Broadcom said AI-chip revenue should exceed $100B in 2027; Marvell said it exceeded $1.5B AI revenue in FY2025 and expects to beat $2.5B in FY2026; and operators provided revenue/backlog/capacity signals (e.g., CoreWeave backlog and contracted capacity timelines).

    As a macro check, Reuters/S&P Global reported Big Tech AI infrastructure spending rose from about $80B in 2019 to $383B in 2025 and about $635B in 2026.

    Sources used in chart

    Overview

    A curated set of perspectives on AI’s economic impact—split into bullish (value creation) and bearish (skepticism / measurement gaps) viewpoints.

    Bull Reports

    1. Stanford HAI — AI Index 2025: Economy

      Executive summary: The AI Index highlights record AI investment in 2024, rapid growth in generative AI funding, and a sharp jump in reported organizational AI usage. It also notes that most firms reporting financial impact are still early—often citing modest (single-digit) cost savings/revenue gains—while research continues to show productivity improvements that can narrow skill gaps.

    2. Penn Wharton Budget Model — Generative AI and Productivity Growth

      Executive summary: PWBM estimates generative AI increases the level of GDP by ~1.5% by 2035 (nearly 3% by 2055 and 3.7% by 2075), with the boost to annual productivity growth peaking in the early 2030s and fading thereafter. It also estimates roughly 40% of labor income is exposed to automation potential, but only a smaller share of GDP is likely to be profitably affected over time—so the macro impact is material but not “transformative” under current evidence.

    3. Harvard Business Review (Webinar) — AI’s Economic Reality in 2026

      Executive summary: Based on a survey of 1,000+ global executives, the presenters argue that many organizations report significant AI value, but the drivers of that value are often misunderstood. The webinar frames a set of “seven factors” that reliably drive outcomes and introduces a six-stage AI Economic Maturity Model to move from pilots toward measurable returns.

    4. St. Louis Fed — Tracking AI’s Contribution to GDP Growth

      Executive summary: The post measures how AI-related investment categories (information processing equipment, software, R&D, and data centers) contribute to real GDP growth and finds those contributions were elevated in early 2025 but normalized as growth rates slowed. It argues this normalization reflects slower investment growth (not lower levels) and compares the magnitude of recent contributions with the dot-com boom, concluding AI-related investment has been a significant driver of 2025 growth.

    5. McKinsey — The Economic Potential of Generative AI

      Executive summary: This source couldn’t be fetched from the build environment (access restricted), so I can’t produce a reliable two-sentence summary from the article text. If you share an excerpt (or PDF), I’ll replace this placeholder with an accurate executive summary.

    6. McKinsey — The State of AI

      Executive summary: This source couldn’t be fetched from the build environment (access restricted), so I can’t produce a reliable two-sentence summary from the article text. If you share an excerpt (or PDF), I’ll replace this placeholder with an accurate executive summary.

    7. McKinsey — Beyond the Hype (TMT)

      Executive summary: This source couldn’t be fetched from the build environment (access restricted), so I can’t produce a reliable two-sentence summary from the article text. If you share an excerpt (or PDF), I’ll replace this placeholder with an accurate executive summary.

    8. Marketing AI Institute — McKinsey’s AI Economic Impact (Secondary Summary)

      Executive summary: This post summarizes McKinsey-style estimates that AI software and services could reach multi-trillion-dollar annual economic potential by 2040, with generative AI’s enterprise value concentrated in a handful of functions like customer ops, marketing/sales, software engineering, and R&D. It also notes that traditional GDP growth constraints make some economists skeptical of headline figures, emphasizing that the exact number matters less than sustained organizational action and capability-building.

    9. Cognizant — AI and the Future of Work Report

      Executive summary: This source couldn’t be fetched from the build environment (redirected/blocked), so I can’t produce a reliable two-sentence summary from the report text. If you share an excerpt (or PDF), I’ll replace this placeholder with an accurate executive summary.

    10. World Economic Forum — “AI bubble” is overblown; closing the value gap

      Executive summary: The article argues that “AI bubble” concerns are better framed as a value gap: large investments aren’t consistently translating into outcomes, even though AI could already address trillions of dollars’ worth of task value. It proposes focusing on workforce skills, embedding better organizational context into AI systems, and designing solution-focused deployments tied to real business problems to realize returns.

    11. Medium — The Great AI Divide (AI contribution to economy by 2030)

      Executive summary: The piece argues AI could add massive value by 2030 but that the gains are likely to be concentrated in a small set of countries and firms, creating a widening “AI divide.” It proposes a multi-part action framework focused on infrastructure access, workforce reskilling, governance/ethics, innovation ecosystems, and financing mechanisms to broaden who benefits from AI.

    12. IDC — Report link (containerId: prUS52600524)

      Executive summary: This link returned an error from the build environment (not accessible), so I can’t extract the report’s key findings for a reliable summary. If you can provide the report text or a public excerpt, I’ll add a two-sentence executive summary here.

    Bear Reports

    1. Fortune — Goldman: No meaningful productivity impact (link)

      Executive summary: This source couldn’t be fetched from the build environment (redirected/blocked), so I can’t produce a reliable two-sentence summary from the article text. If you share an excerpt (or a non-paywalled source), I’ll replace this placeholder with an accurate executive summary.

    2. Business Insider — AI’s GDP “blind spot” (Goldman Sachs)

      Executive summary: Goldman argues AI’s economic lift is understated in official GDP because the accounting treatment of semiconductors (and some AI build costs) pushes much of the value into intermediate inputs or uncapitalized intangibles. In the cited estimates, AI added about $160B to real economic activity since 2022, but only ~$45B shows up in GDP—leaving roughly a $115B measurement gap.

    3. Tom’s Hardware — “AI boosted the US economy by basically zero” (link)

      Executive summary: This source couldn’t be fetched from the build environment (redirected/blocked), so I can’t produce a reliable two-sentence summary from the article text. If you share an excerpt (or a non-paywalled source), I’ll replace this placeholder with an accurate executive summary.

    4. Gizmodo — “AI added basically zero” to growth (link)

      Executive summary: This source couldn’t be fetched from the build environment (redirected/blocked), so I can’t produce a reliable two-sentence summary from the article text. If you share an excerpt (or a non-paywalled source), I’ll replace this placeholder with an accurate executive summary.

    5. Goldman Sachs — How will AI affect the US labor market?

      Executive summary: Goldman Sachs Research expects widespread AI adoption to unfold over roughly a decade, with a base-case displacement of 6–7% of workers and a modest (~0.6pp) unemployment-rate increase if the transition is gradual. It also argues AI can create jobs via infrastructure buildout (power and data centers) and new specializations, even as knowledge and creative sectors see earlier displacement pressures.

    6. Goldman Sachs — Why AI companies may invest more than $500B in 2026

      Executive summary: Goldman highlights rising consensus estimates for hyperscaler capex in 2026 (cited at ~$527B) and notes that analysts have repeatedly underestimated AI-related spending, even as investors grow more selective about which capex-heavy names they reward. It argues the “AI trade” may broaden beyond infrastructure toward AI platforms and productivity beneficiaries, while also warning that capex growth slowdowns could pressure infrastructure valuations.