Morgan Stanley forecasts $1.58 tn AI boom as Nvidia (NVDA) exposure hits 92%

Morgan Stanley expects spending on artificial intelligence infrastructure to reach $1.58 trillion in 2027, with Nvidia (NASDAQ: NVDA) positioned to capture a substantial share of investment in computing equipment and data centers.

The bank estimates that 91.7% of Nvidia’s projected 2027 revenue is exposed to AI infrastructure demand, the highest proportion among companies covered in its October 15 research report.

• Morgan Stanley expects hyperscaler and AI cloud spending to reach $1.58 trillion in 2027 and $1.8 trillion in 2028. Nvidia has the highest AI infrastructure exposure among companies analyzed, at 91.7% of projected 2027 revenue.
• Nvidia reported record quarterly revenue of $96.22 bn, up 106% year over year. Its trailing P/E has fallen to approximately 27, the lowest level in seven years, while analysts’ average price target stands at $328.72.
• Morgan Stanley identifies financing costs, a projected 32 GW US data center power shortfall and concentrated customer demand as major risks. A $400 Nvidia share price in 2027 would require earnings growth and a higher valuation multiple.

Vertiv follows at 90%, while Broadcom has estimated exposure of 81%.

Despite Nvidia’s concentration in AI infrastructure, Morgan Stanley identifies the semiconductor manufacturer as offering the most attractive risk-adjusted return among companies in the sector.

The report, “Following AI’s $1.6 Trillion Money Trail: Mapping the Revenue Generated by the Capex Boom,” was prepared by Morgan Stanley’s North America Insight team, led by Nicholas Lentini.

Its forecasts indicate substantial growth in AI infrastructure expenditure, alongside increasing dependence on debt financing, electricity availability and a relatively small group of major technology customers.

Morgan Stanley forecasts $1.58 trillion in AI spending by 2027

Morgan Stanley estimates that hyperscalers and specialized AI cloud providers will spend approximately $980 bn in 2026, rising nearly 60% to $1.58 trillion in 2027 and reaching $1.8 trillion in 2028.

The 2027 projection includes approximately $195 bn in data center leases, which Morgan Stanley categorizes as “synthetic capex” because these commitments do not appear as conventional capital expenditures in cash flow statements.

Hyperscalers have disclosed more than $1.1 trillion in lease commitments that have yet to commence.

According to the bank’s estimates, approximately $1 trillion of projected 2027 infrastructure spending will translate into revenue for S&P 500 companies, with nearly $700 bn attributed to Nvidia.

Morgan Stanley calculates that revenue generated from hyperscaler and neocloud customers will represent approximately 5% of total S&P 500 revenue in 2027 while accounting for roughly 25% of the index’s revenue growth.

The bank expects AI infrastructure investment to generate returns on invested capital of 25% to 50%. Its assessment also draws on four previous computing investment cycles: personal computers, the internet, mobile computing and cloud infrastructure.

Across those cycles, Morgan Stanley estimates that Wall Street underestimated the scale of computing investment by an average of 38%.

Nvidia’s revenue depends heavily on two major customers

Morgan Stanley’s forecasts show substantial concentration within Nvidia’s customer base.

In the bank’s 2027 model, SpaceX and Amazon Web Services account for approximately 54% of Nvidia’s hyperscaler revenue and 72% of its projected growth in that customer segment.

These figures illustrate Nvidia’s dependence on investment decisions made by a small number of large infrastructure operators.

Morgan Stanley considers its revenue assumptions for SpaceX and Amazon conservative. The bank’s model projects that Nvidia’s share of spending by other major hyperscalers will decline from 18% in 2026 to 16% in 2027, although Nvidia expects to increase its market share.

The company’s data center business already accounts for most of its sales, making continued investment in AI computing capacity central to its earnings outlook.

Morgan Stanley’s global investment committee acknowledged this dependence in March, identifying the durability of AI capital expenditure as a primary condition supporting the technology-led equity market advance.

Nvidia revenue rises 106% to $96.22 bn

Nvidia’s latest financial results demonstrate the scale of demand for its computing products.

The company reported revenue of $96.22 bn for the second quarter of fiscal 2027, ended July 26, 2026, an increase of 105.8% from a year earlier. Data Center revenue reached $89.02 bn, up 117% year over year.

Net income increased 126% to approximately $59.69 bn, while adjusted earnings reached $2.22 per share, exceeding analysts’ expectations of approximately $2.09.

Nvidia has exceeded consensus earnings-per-share estimates for six consecutive quarters.

Chief Executive Jensen Huang described demand as accelerating during the company’s second-quarter earnings announcement.

The company is expanding production of its Vera Rubin computing platform, which is expected to support the next phase of demand from hyperscalers and specialized AI cloud providers.

The platform is also central to projections of continued revenue growth in fiscal 2028.

Nvidia valuation falls to seven-year low

Nvidia shares closed at $217.09 on October 19, 2026, up 1.38% for the session, giving the company a market capitalization of approximately $5.23 trillion.

Despite substantial share-price appreciation in recent years, its earnings valuation has declined as profits have grown faster than its stock price.

Nvidia trades at slightly more than 27 times trailing 12-month earnings, its lowest trailing P/E ratio since June 2019, according to YCharts.

The company’s average trailing P/E over the past decade was approximately 61.3. Its current multiple is therefore less than half its ten-year average, although it remains above the S&P 500’s trailing P/E of approximately 24.9.

Alphabet, Amazon and Meta Platforms also trade at lower trailing earnings multiples than Nvidia. Forward valuations present a different comparison.

Based on projected fiscal 2028 earnings, Nvidia trades at approximately 14 to 15 times earnings, depending on the forecast used.

That multiple is below the forward valuations of the other Magnificent Seven companies and compares with an estimated 21 to 23 times forward earnings for the S&P 500.

The valuation reflects expectations for continued rapid earnings growth, but also depends on the accuracy of analysts’ fiscal 2028 forecasts.

Wall Street forecasts Nvidia earnings growth of 71%

Consensus earnings estimates for Nvidia have increased substantially during the past three months.

Analysts project fiscal 2028 earnings per share of approximately $15.91, compared with $12.71 estimated 90 days earlier.

The latest projection represents approximately 71% growth from expected fiscal 2027 earnings of $9.31 per share.

Over the past 30 days, analysts submitted 46 upward earnings revisions and no downward revisions, according to the estimates cited in the October 19 market analysis.

The average analyst price target stands at $328.72, approximately 51% above Nvidia’s October 19 closing price of $217.09.

Of 62 analysts included in the cited coverage, 59 have positive ratings, two recommend holding the shares and one has a sell recommendation.

A separate valuation scenario published by 24/7 Wall St. examines the possibility of Nvidia reaching $400 per share in 2027.

At $400, Nvidia would trade at approximately 25 times projected fiscal 2028 earnings of $15.91 per share. That valuation would require an increase of approximately 84% from the October 19 closing price, substantially exceeding the current average analyst target.

The scenario depends on continued earnings growth, successful deployment of Vera Rubin products and sustained hyperscaler demand.

It also assumes investors would assign Nvidia a higher multiple than its current fiscal 2028 forward valuation.

AI infrastructure financing raises credit concerns

Morgan Stanley expects 2027 to produce the largest cash flow deficit in its forecast period for major AI infrastructure customers, increasing their dependence on borrowing.

Hyperscalers have issued approximately $255 bn in unsecured bonds during 2026, equivalent to roughly 30% of estimated capital spending.

The bank modeled how additional borrowing could affect interest coverage ratios, which measure earnings available to service interest obligations.

In its hypothetical scenario, the combined interest coverage ratio for Amazon, Alphabet, Meta and Microsoft, all rated AA or higher, declines from approximately 31 times to 12 times.

Morgan Stanley also identifies higher equipment prices as a contributor to infrastructure spending growth.

Amazon’s finance chief attributed most of an increase in the company’s capital expenditure plan, from $200 bn to approximately $220 bn, to rising memory costs rather than an expansion of planned infrastructure capacity.

That distinction affects forecasts of how much additional computing infrastructure higher spending will actually deliver.

US data centers face projected 32 GW power shortage

Electricity availability presents another limitation on AI infrastructure deployment.

Morgan Stanley forecasts a US data center power shortfall of approximately 32 gigawatts through 2028, equivalent to roughly one-third of the 97 gigawatts required.

The bank’s investment projections assume that purchased computing equipment can ultimately be connected to operating data centers.

Power shortages could delay those installations even if hyperscalers continue ordering processors.

Morgan Stanley warns that companies may accumulate graphics processing units before securing sufficient electricity, potentially leaving expensive equipment unused.

Continued advance purchasing could sustain Nvidia’s near-term chip revenue while increasing inventory exposure for its customers.

The bank also identifies potential regulatory restrictions on data center construction and the possible reduction of government tax incentives as risks to infrastructure investment.

It does not expect a federal moratorium on data center development, but considers tax incentives a potential target for legislative changes.

Morgan Stanley warns of supplier margin pressure

Morgan Stanley’s downside scenario focuses on the possibility that infrastructure spending stabilizes before suppliers have adjusted production capacity. In that environment, shipment volumes could remain relatively stable while equipment prices decline, reducing supplier profitability.

The bank cites the NAND memory downturn in late 2022, when shipments fell approximately 5% but prices declined 20% to 30%.

Morgan Stanley expects suppliers of memory, optical components and other relatively small parts to face greater disruption if orders weaken or customers accumulate excess inventory.

Nvidia could be less exposed to the initial adjustment because of its position as the largest supplier of high-value computing equipment. Nevertheless, simultaneous declines in processor prices and shipment volumes would present a more substantial earnings risk.

Nvidia also faces competition from AMD and internally developed AI processors at Amazon, Alphabet and Microsoft.

Morgan Stanley argues that investors may be underestimating the potential scale of spending reductions if the AI infrastructure investment cycle weakens.

The bank’s projections remain positive, but their realization depends on financing conditions, customer investment plans and the ability of infrastructure operators to deploy the computing equipment they purchase.

For Nvidia, those assumptions are central to both Morgan Stanley’s projected 2027 revenue exposure of 91.7% and Wall Street’s expectations for fiscal 2028 earnings growth.