Goldman Sachs: US hyperscalers need $1.42 trillion revenue by 2030 for AI ROI
Goldman Sachs projects that six major US hyperscale cloud providers (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX) will spend $1.73 trillion on AI computing in 2026-2027. To achieve a 15% return on invested capital, these firms need to generate approximately $1.42 trillion in cumulative revenue from 2028 to 2030. The analysis highlights a $276 billion annual depreciation burden and notes $1.69 trillion in cloud backlog orders as support. Risks include $3 trillion in off-balance-sheet commitments and potential GPU price declines.
IllustrationEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Common ground
- Both agree that AI infrastructure spending is historically unprecedented and carries significant risk of overinvestment.
- Both agree that the off-balance-sheet GPU commitments (around $3 trillion) are a major concern, creating hidden leverage and cash flow risks.
- Both agree that a 20-40% write-down on AI infrastructure by 2028 is likely, but that major tech companies have strong enough balance sheets to absorb it without a systemic crisis.
- Both agree that the regulatory vacuum around AI spending—no disclosure, no stress tests—is a problem that will likely lead to congressional overreaction after the first major miss.
Points of contention
- Neutral Agent sees the spending as rational for individual companies with no better growth options, while Western Agent calls it groupthink and a desperate gamble to justify stock prices.
- Neutral Agent argues internal productivity gains from AI (like automating support) are recurring and can significantly improve margins, while Western Agent says those savings are too small to justify trillion-dollar investments and hurt aggregate demand.
- Western Agent frames the situation as a regulatory failure and potential systemic risk, while Neutral Agent insists it's a coordination failure among rational actors, not a crisis comparable to banking collapses.
Blind spots
- Neither side fully models the scenario where AI doesn't generate massive new revenue but does deliver huge cost savings—which Neutral Agent raises but both fail to explore in depth.
- Both overlook the political economy angle: how national competitiveness arguments are used to justify the lack of oversight, and how that might change after a downturn.
- The debate ignores the possibility that faster-than-expected hardware obsolescence (e.g., GPU lifespans dropping from 5 to 3 years) could drastically worsen returns, beyond what either side accounts for.
WorldAttention’s read
This debate reveals a deep split between seeing the AI infrastructure buildout as a rational, if risky, bet by companies with no better options, versus seeing it as a regulatory failure and collective delusion. Both sides agree the spending is unprecedented, the off-balance-sheet commitments are dangerous, and a 20-40% write-down is likely by 2028—but they disagree on whether that's a manageable correction or a sign of deeper rot. The biggest blind spot is the potential for internal cost savings to partially justify the spending, and the lack of any serious modeling of how faster hardware obsolescence could crush returns. Ultimately, the outcome hinges on whether AI revenue grows fast enough to meet heroic targets, or whether cost-cutting alone can bridge the gap—and neither side has a clear answer.
Reporting timeline
Goldman Sachs Research: puts 2026 global AI investment at $1.019T, with $581B spent in the US. About $200B above the hyperscaler figure markets track. Goldman also expects US AI investment to climb from 1.8% of GDP this
Goldman Sachs Research has released a forecast projecting global artificial intelligence investment will reach $1.019 trillion by 2026, with $581 billion of that total spent in the United States. This figure is approximately $200 billion above the hyperscaler figure that markets currently track. The investment bank also expects US AI investment to rise from 1.8% of GDP this year to 2.8% in 2028, while the global share increases from 0.9% to 1.4% over the same period. To isolate AI spending, Goldman economists treat hyperscalers' 2022 capital expenditure as a pre-boom baseline and count nearly everything above that level in 2026 as AI investment. They also exclude financial leases where possible to avoid double-counting hardware that another company may have already recorded as its own capex. The methodology is necessary because hyperscalers such as Microsoft, Amazon, and Google were already spending heavily on data centers before the AI boom for cloud storage, search, video, and advertising, and their capex reports do not split ordinary spending from AI-specific spending.
Read sourceGoldman Sachs Forecasts 50% Surge in Cloud AI Spending to $1.2 Trillion Next Year
Goldman Sachs strategists forecast that the five largest US hyperscale cloud providers will increase their AI infrastructure spending by over 50% next year, reaching $1.2 trillion. The analysis shifts the debate from the scale of spending to the potential returns on this massive investment. Goldman outlines a three-phase AI buildout: Phase 1 (2023-2025) at $633 billion, Phase 2 (2026-2027) at $1.73 trillion, and Phase 3 (2028-2030) at $4.14 trillion. To achieve a 15% return on invested capital (ROIC) on the Phase 2 spending, the six major hyperscalers (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX) would need to generate approximately $1.42 trillion in cumulative revenue from 2028 to 2030. The report highlights a key risk from depreciation, which could be significantly higher if GPU lifespans are shorter than the assumed 5 years. Cloud backlog orders of $1.69 trillion provide some support for the revenue targets, but the analysis also notes risks from off-balance-sheet commitments and the massive financing gap for the overall spending, estimated by Morgan Stanley at $1.5 trillion. Goldman maintains a 'buy' rating on the six firms, viewing near-term return compression as a natural result of a large upfront investment cycle.
Goldman Sachs Analyzes How to Fill the 'Capital Gap' in AI's Second Phase
Goldman Sachs research analyzes the economic viability of massive AI capital expenditures by six major hyperscale cloud companies (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX). The report outlines a three-phase AI investment cycle, with Phase 2 (2026-2027) totaling approximately $1.73 trillion. To achieve a 15% return on invested capital (ROIC) on this spending, the companies would need to generate cumulative revenue of about $1.42 trillion from 2028 to 2030. Key assumptions include a 5-year depreciation for computing assets and a 60-85% utilization rate. The analysis highlights a significant depreciation burden of $276 billion annually, which could increase if GPU lifespans are shorter. Cloud backlog orders of $1.69 trillion provide some support for revenue targets, but risks include $3 trillion in off-balance-sheet commitments, falling GPU pricing, and the even larger $4.14 trillion Phase 3 spending. Morgan Stanley's research corroborates the revenue potential but also identifies a $1.5 trillion financing gap for data center spending. Goldman maintains 'buy' ratings on the six firms, viewing near-term return compression as a natural part of the investment cycle.
Read sourceShow 17 older updatesHide older updates
Goldman Sachs: Six Tech Giants Need $1.42 Trillion Revenue by 2028-2030 for 15% AI Investment Return
A Goldman Sachs research report on US technology sector returns analyzes the investment returns of expanding AI capital expenditure by major cloud providers. The bank examined six large US tech companies—Alphabet, Amazon, Microsoft, Meta, Oracle, and SpaceX—and conducted a stress test on their computing investment returns for 2026-2027. Assuming an average upfront cost of approximately $420 billion per gigawatt (GW) of computing power, with 70% of capital expenditure on computing hardware and 30% on data center construction, and incorporating conservative depreciation rules and ongoing operating costs, Goldman Sachs estimates that the six companies must generate a combined revenue of approximately $1.42 trillion between 2028 and 2030. This equates to roughly $11.6 billion in annual revenue per GW of computing power to achieve a 15% annualized return on invested capital (ROIC) for this round of AI computing investment. The bank notes that the commercial value of different tokens varies, and returns on AI capital expenditure will not be uniform. However, considering market growth potential over the next three to five years, Goldman Sachs expects that capital invested in the AI sector over the next 18 months will generally maintain good investment returns.
Read sourceGoldman Sachs: Six Tech Giants Need $1.42 Trillion Revenue by 2028-2030 for 15% AI Compute ROI
Goldman Sachs has released a research report analyzing the return on investment for major cloud providers' expanding AI capital expenditures. The bank conducted a stress test on the compute investments of six large US tech companies—Alphabet, Amazon, Microsoft, Meta, Oracle, and SpaceX—for the 2026-2027 period. Assuming an average upfront cost of approximately $42 billion per gigawatt (GW) of compute power, with 70% spent on hardware and 30% on data center construction, and incorporating conservative depreciation and operating costs, Goldman estimates these six firms must generate a combined $1.42 trillion in revenue between 2028 and 2030. This equates to roughly $11.6 billion in annual revenue per GW of compute power to achieve a 15% annualized return on invested capital (ROIC). The bank notes that the commercial value of different tokens varies, so AI capital expenditure returns will not be uniform. However, given market growth potential over the next 3-5 years, Goldman expects overall investment in AI over the next 18 months to maintain good returns.
Goldman Sachs: Big Tech Needs $1.42 Trillion in AI Revenue by 2030 for 15% ROIC
Goldman Sachs published a research report analyzing the economic scale required for major US hyperscale cloud firms to justify their massive AI capital expenditures. The report focuses on the core market debate: whether the huge AI infrastructure investments by Alphabet, Amazon, Microsoft, Meta, Oracle, and SpaceX will generate reasonable returns. Goldman Sachs built a quantitative framework to stress-test the returns on 2026-2027 compute investments. The analysis assumes $42 billion upfront cost per gigawatt of compute, with 70% spent on hardware and 30% on facilities. Under conservative depreciation and operating cost assumptions, the report calculates that the six firms need to generate approximately $1.42 trillion in cumulative revenue between 2028 and 2030 to achieve a 15% annualized ROIC on their second-phase AI investments. Goldman Sachs notes that while short-term returns will be pressured by the massive upfront spending, the long-term outlook remains attractive. The report cites signals from the Communacopia tech conference that the industry is moving from AI experimentation to deployment, with a focus on optimizing token input-output ratios. It also highlights the launch of consumer AI agent products like Meta's Muse as a paradigm shift that will drive long-term compute demand and monetization through subscriptions, advertising, and e-commerce.
Read sourceGoldman Sachs Calculates Revenue Needed to Close AI Capital Gap in Second Phase
Goldman Sachs research analyzes the economic viability of massive AI capital expenditure by six hyperscale cloud firms (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX). The study divides the AI compute buildout into three phases: Phase 1 (2023-2025, ~$633 billion), Phase 2 (2026-2027, ~$1.73 trillion), and Phase 3 (2028-2030, ~$4.14 trillion). To achieve a 15% return on invested capital (ROIC) on Phase 2 spending, the firms would need to generate approximately $1.42 trillion in cumulative revenue from 2028 to 2030, requiring annual revenue of about $465-476 billion. Key risks include depreciation pressure from GPU lifespan assumptions, with Goldman using a 5-year depreciation schedule versus potential 3-year cycles. Cloud backlog orders of $1.69 trillion provide some support for revenue targets. The analysis excludes approximately $3 trillion in off-balance-sheet commitments and does not address Phase 3 returns. Morgan Stanley's separate research estimates a $1.5 trillion financing gap for global data center spending through 2028.
Read sourceGoldman Sachs Analyzes How to Fill the 'Capital Gap' in AI's Second Phase
Goldman Sachs has published research analyzing the financial viability of massive AI capital expenditures by six major hyperscale cloud companies (Alphabet, Microsoft, Amazon, Meta, Oracle, SpaceX). The report calculates that to achieve a 15% return on invested capital (ROIC) on approximately $1.73 trillion in AI computing spending during 2026-2027, these firms would need to generate roughly $1.42 trillion in cumulative revenue from 2028-2030. This translates to annual revenue of about $465-476 billion. The analysis highlights a significant depreciation burden of $276 billion annually, which is 1.7 times the required EBIT. Goldman Sachs points to $1.69 trillion in cloud backlog orders as a key safety net, though it notes risks including off-balance-sheet commitments totaling $3 trillion and potential GPU price declines. The report also references Morgan Stanley's findings of a $1.5 trillion financing gap for data center spending. Goldman maintains 'buy' ratings on all six companies, characterizing near-term return compression as a natural result of large upfront investment cycles.
Read sourceGoldman Sachs Analyzes How to Fill the 'Capital Gap' in AI's Second Phase
A Goldman Sachs research report, covered by NetEase Finance, analyzes the economic viability of massive AI capital expenditures by six major hyperscale cloud companies (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX). The report outlines a three-phase investment cycle, with Phase 2 (2026-2027) totaling approximately $1.73 trillion. To achieve a 15% return on invested capital (ROIC) on this spending, the companies would need to generate roughly $1.42 trillion in cumulative revenue from 2028 to 2030. Goldman Sachs highlights a significant depreciation burden, with annual depreciation alone reaching $276 billion. The report points to $1.69 trillion in cloud backlog orders as a key safety net for revenue targets but acknowledges risks including a potential $3 trillion in off-balance-sheet commitments, falling GPU pricing, and the unresolved question of returns on a projected $4.14 trillion Phase 3 investment. The analysis contrasts with Morgan Stanley's findings on a potential $1.5 trillion financing gap for global data center spending.
Read sourceGoldman Sachs: Hyperscalers Need $1.42 Trillion AI Revenue by 2030 to Justify $1.73 Trillion Spend
Goldman Sachs research analyzes the financial viability of massive AI infrastructure spending by six major hyperscale cloud companies (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX). The report divides the AI buildout into three phases: Phase 1 (2023-2025, $633 billion), Phase 2 (2026-2027, $1.73 trillion), and Phase 3 (2028-2030, $4.14 trillion). To achieve a 15% return on invested capital (ROIC) on Phase 2 spending, these companies would need to generate approximately $1.42 trillion in cumulative revenue from 2028 to 2030, or about $465-476 billion annually. The analysis highlights that annual depreciation alone would be $276 billion, 1.7 times the required EBIT. Cloud backlog orders of $1.69 trillion provide some support for revenue targets, but the report notes $3 trillion in off-balance-sheet commitments are excluded from calculations. Morgan Stanley's parallel research estimates a $1.5 trillion financing gap for data center spending through 2028. Goldman maintains buy ratings on all six companies, characterizing near-term return compression as a natural result of large upfront investment cycles.
Read sourceCombined capital-spending outlook for hyperscalers, Alphabet, Microsoft, Amazon, Meta and Oracle across 2026-2027 is now about 66% above where it stood when 2026 began. i.e. up by ~$750 B. https://t.co/GQ33AzKT3U
According to a post by analyst rohanpaul_ai on X, the combined capital-spending outlook for major hyperscalers—including Alphabet, Microsoft, Amazon, Meta, and Oracle—for the period spanning 2026 to 2027 is now approximately 66% higher than it was at the start of 2026. This represents an increase of roughly $750 billion. The forecast reflects a significant upward revision in planned investment by these technology giants, likely driven by demand for AI and cloud computing infrastructure.
Read sourceGoldman Sachs Calculates How to Fill the 'Capital Gap' in AI's Second Phase
Goldman Sachs research analyzes the economic viability of massive AI capital expenditure by six hyperscale cloud firms (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX). The study divides the AI compute buildout into three phases: Phase 1 (2023-2025, ~$633B), Phase 2 (2026-2027, ~$1.73T), and Phase 3 (2028-2030, ~$4.14T). To achieve a 15% return on invested capital (ROIC) on Phase 2 spending, these firms need to generate approximately $1.42 trillion in cumulative revenue from 2028 to 2030, or about $465-476 billion annually. Key risks include depreciation pressure (annual depreciation of $276B is 1.7 times required EBIT), potential shorter GPU lifespans, and ~$3 trillion in off-balance-sheet commitments. Cloud backlog orders of $1.69 trillion provide some support, but Morgan Stanley highlights a $1.5 trillion financing gap for data center spending. Goldman maintains 'buy' ratings on all six firms, viewing near-term return compression as a natural result of upfront investment.
Read sourceGoldman Sachs Calculates How to Fill the 'Capital Gap' in AI's Second Phase
Goldman Sachs research analyzes the economic viability of massive AI capital expenditure by six hyperscale cloud firms (Alphabet, Amazon, Microsoft, Meta, Oracle, SpaceX). The study divides the AI buildout into three phases: Phase 1 (2023-2025, ~$633B), Phase 2 (2026-2027, ~$1.73T), and Phase 3 (2028-2030, ~$4.14T). To achieve a 15% return on invested capital (ROIC) on Phase 2 spending, these firms need to generate approximately $1.42 trillion in cumulative revenue from 2028 to 2030, or about $465-476 billion annually. Key risks include a $3 trillion in off-balance-sheet commitments not counted in the ROIC denominator, depreciation pressure from GPU lifespan (Goldman assumes 5 years; shorter life would raise revenue targets), and pricing pressure on GPU and compute. Cloud backlog orders of $1.69 trillion (up 152% year-over-year) provide some support for revenue targets. Morgan Stanley's parallel analysis estimates a $1.5 trillion financing gap for 2025-2028 data center spending, with only $1.4 trillion covered by hyperscaler cash flow. Goldman maintains 'buy' ratings on all six firms, characterizing near-term return compression as a natural result of large upfront investment cycles.
Read sourceGoldman Sachs projects Amazon, Meta, Google, Microsoft, Oracle to boost AI spending to $1.2 trillion by 2027
Goldman Sachs has released a new projection stating that Amazon, Meta, Google, Microsoft, and Oracle will increase their artificial intelligence infrastructure spending by more than 50%, reaching $1.2 trillion in 2027 and $1.4 trillion in 2028. The investment bank also noted that a surprise of $250 billion in either direction next year would move S&P 500 earnings growth by roughly 6 percentage points in the same direction. The forecast highlights the massive scale of capital expenditure planned by these technology giants as they compete in the AI sector.
Read sourceGoldman Sachs Forecasts 50% Surge in US Hyperscaler AI Spending to $1.2 Trillion in 2027
Goldman Sachs strategists, led by Ryan Hammond, project that the five largest US hyperscale cloud providers—Amazon, Alphabet, Microsoft, Oracle, and Meta Platforms—will increase their AI infrastructure capital expenditure by over 50% next year, reaching $1.2 trillion. This forecast exceeds Wall Street's consensus estimate of $1.1 trillion for 2027. The strategists note that to achieve a return on these investments, the companies will need to generate approximately $300 billion in annual AI revenue in the coming years. They also highlight that capital expenditure as a share of GDP is expected to surpass levels seen during any technology investment cycle since the 19th-century railroad boom. While spending is set to rise significantly, the growth rate is predicted to decelerate from nearly 100% this year to 54% in 2027 and further to 12% in 2028, when total spending is projected to reach $1.4 trillion.
Read sourceGoldman Sachs Sees AI Capital Expenditure Rising, Says Revenue Must Catch Up
Goldman Sachs strategists, led by Ryan Hammond, forecast that capital expenditure by US hyperscale cloud companies will rise to approximately $1.2 trillion by 2027, surpassing the market consensus of $1.1 trillion, and reach $1.4 trillion by 2028. However, they note that the pace of growth and upside surprises will slow compared to recent quarters, with spending expected to double to $800 billion this year. The report highlights that these companies' capital expenditure already exceeds their operating cash flow, meaning future growth will increasingly require debt and equity financing. To achieve breakeven on their 2026-2027 investments, the firms need roughly $300 billion in annual AI-related revenue. For robust returns, consumer and enterprise spending on AI applications must reach about $1 trillion annually. While current AI revenue is insufficient for breakeven, it is growing rapidly, with cloud revenue growth accelerating to 48% in Q2 for Amazon, Google, Microsoft, and Oracle, and combined cloud backlog reaching $1.7 trillion. The report notes that investor skepticism is partly reflected in lower AI infrastructure valuations and hyperscaler price-to-earnings ratios at decade lows.
Read sourceGoldman Sachs Sees 50% Jump in Big Tech AI Spending to $1.2 Trillion Next Year
Goldman Sachs strategists, led by Ryan Hammond, forecast that the five largest US hyperscale cloud providers will increase their AI infrastructure spending by over 50% next year, reaching $1.2 trillion. This figure surpasses Wall Street's consensus estimate of $1.1 trillion for 2027. The analysts note that to break even on these investments, the companies will need to generate approximately $300 billion in annual AI revenue in the coming years. The report highlights that capital expenditure as a share of GDP is set to exceed any previous tech investment cycle since the late 19th-century railroad boom. Amazon, Alphabet, Microsoft, Oracle, and Meta are expected to spend $800 billion on AI capabilities this year. While spending will rise sharply in 2025, the growth rate is projected to slow. Potential constraints include shortages of electricity, labor, and memory chips, as well as data center construction limitations. The strategists expect spending growth to decelerate from nearly 100% this year to 54% by 2027 and 12% by 2028, when total spending is forecast to reach $1.4 trillion.
Read sourceGoldman Sachs Sees 50% Jump in Big Tech AI Spending to $1.2 Trillion Next Year
Goldman Sachs strategists, led by Ryan Hammond, forecast that the five largest US hyperscale cloud providers will increase their AI infrastructure spending by over 50% next year, reaching $1.2 trillion. This figure exceeds the Wall Street consensus estimate of $1.1 trillion for 2027. The report notes that to break even on these investments, the companies will need to generate approximately $300 billion in annual AI revenue in the coming years. The strategists compare the current investment cycle to the railroad boom of the late 19th century, stating that capital expenditure as a share of GDP is expected to surpass any tech investment cycle since then. Amazon, Alphabet, Microsoft, Oracle, and Meta are projected to spend $800 billion on AI capabilities this year. While spending will increase significantly next year, the growth rate is expected to slow from nearly 100% this year to 54% by 2027, and further to 12% in 2028, when total capital expenditure is projected to reach $1.4 trillion.
Read sourceGoldman Sachs: Big Tech Needs $1.42 Trillion in AI Revenue by 2030 for 15% ROI
Goldman Sachs has published a research report analyzing the economic scale required for major US hyperscale cloud firms to justify their massive AI capital expenditures. The report focuses on the core market debate over whether the huge investments in AI computing power by companies like Alphabet, Amazon, Microsoft, Meta, Oracle, and SpaceX will yield reasonable returns. Goldman Sachs built a quantitative framework to stress-test the returns on 2026-2027 compute investments. The analysis estimates that these six firms need to generate a combined cumulative revenue of approximately $1.42 trillion between 2028 and 2030 to achieve a 15% annualized return on invested capital (ROIC) on their second-phase AI investments. The report notes that while short-term returns may be pressured by the large upfront spending, the firm believes attractive ROIC levels will materialize in the medium to long term. It also highlights a shift from AI experimentation to implementation and the emergence of agent-based AI products as key drivers of future demand.
Read sourceGoldman Sachs: Big Tech Needs $1.42 Trillion in AI Revenue by 2028-2030 for 15% ROIC
A Goldman Sachs research report analyzes the economic scale required for major US hyperscale cloud firms to justify their massive AI capital expenditures. The report focuses on the core market debate: whether the huge investments in AI computing power by companies like Alphabet, Amazon, Microsoft, Meta, Oracle, and SpaceX will yield reasonable returns. Goldman Sachs built a quantitative framework to stress-test the returns on computing investments for 2026-2027. The analysis estimates that these six firms need to generate a combined cumulative revenue of approximately $1.42 trillion between 2028 and 2030 to achieve a 15% annualized return on invested capital (ROIC) on their second-phase AI investments. The report notes that while short-term returns will be pressured by the upfront spending, the long-term outlook is positive, driven by increasing AI adoption and a shift from experimentation to implementation. It also highlights the emergence of consumer AI agents as a key future revenue driver.