Morgan Stanley chief executive Ted Pick expects AI infrastructure spending to remain a roughly trillion-dollar annual opportunity for five to 10 years, but he is not promising a smooth or fixed path. His forecasts describe an early, multi-stage buildout whose pace will be shaped by power, financing, construction capacity, technology and actual demand.
What Ted Pick is forecasting
Pick’s “another decade” thesis is a description of duration, not a precise annual-spending commitment. In his Q2 2026 earnings call, he used a tenfold expansion from roughly $1 trillion of cloud-compute investment as an analogy for a possible $10 trillion AI-compute buildout over 10 years. He stressed that both the total and the timing were uncertain.
As an Amazon Associate I earn from qualifying purchases.
His Q3 2026 comments put the near-term opportunity at about $1 trillion of annual AI capital spending, potentially sustained for five to 10 years. Pick also said that level could add approximately 50–70 basis points to U.S. real GDP. That is a macroeconomic estimate, not a forecast of any particular company’s revenue or stock performance.
Capex estimates have risen sharply
The numbers discussed on Morgan Stanley’s 2026 earnings calls show how quickly expectations have moved. They are management discussion and scenarios, not guaranteed outcomes.
#1 Best Overall
| Period | Earlier figure discussed | Updated or potential figure | Qualification |
|---|---|---|---|
| 2026 | $575 billion | About $850 billion | Pick said the forecast was raised during the Q2 2026 call. |
| 2027 | About $700 billion | $1.3 trillion | Also described in the Q2 2026 discussion; it is an estimate, not a committed budget. |
| 2028 | Not stated | Up to $1.5 trillion in the Q2 discussion | In the Q3 discussion, Pick said plausible figures could instead be $1.0 trillion, $1.2 trillion or $1.4 trillion. |
The different 2028 figures are important. They do not establish a single official target; they show how sensitive the outlook is to assumptions about hyperscaler plans, equipment availability, power and demand. A headline number of $1.5 trillion should therefore be read as an upper scenario from the Q2 conversation, not as a scheduled expenditure.
Why Pick calls the cycle only 10%–15% complete
Pick said Morgan Stanley’s analysis put the market at roughly 10%–15% of the way through the investment cycle. That assessment reflects the number of linked systems that must expand together:
Compute and networking
AI workloads require large clusters of processors, memory, networking equipment and storage. Replacing or expanding those clusters creates recurring demand even after an initial data center opens, because models, applications and performance requirements change.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteRank #2
- Ideal for Gifting
- Ideal for a bookworm
- Compact for travelling
Data-center construction
Servers alone do not create usable capacity. Developers need buildings, cooling systems, electrical infrastructure and interconnection work. A long cycle can result when projects are planned years before all of that capacity is operational.
Power availability
Electricity is a physical constraint. Delays in generation, transmission or grid connections can slow a project even when financing and equipment are ready. Power shortages can make spending look “not fast enough” despite strong demand for computing.
Financing
The buildout requires substantial capital from companies, lenders and investors. Higher funding costs, tighter credit or concerns about a customer’s ability to pay can postpone projects. Pick said the capital will have to be provided for what he called a “once in a multi-generation change,” but that does not eliminate financing risk.
Adoption and utilization
Infrastructure must eventually support paying workloads. If businesses adopt AI more slowly than expected, capacity can sit underused and returns can disappoint. That timing gap is one reason a decade-long investment cycle can still include abrupt pauses.
Why the path will not be a straight line
Pick explicitly expects periods when investment runs ahead of adoption or when one bottleneck prevents spending from keeping pace with another. His Q3 description was a cycle that can feel “too fast” because of overbuilding, or “not fast enough” because of a different constraint.
| Scenario | What it would look like | Likely pressure point |
|---|---|---|
| Overbuilding | Capacity is added faster than customers put it to work. | Lower utilization, weaker returns and delayed follow-on projects. |
| Power or construction bottleneck | Demand exists, but facilities cannot be energized or completed on schedule. | Permitting, grid connections, equipment or skilled labor. |
| Financing stress | Projects are economically attractive but difficult to fund. | Interest rates, credit availability or lender risk limits. |
| Technology shifts | A new architecture changes the amount or type of infrastructure required. | Obsolescence, redesign costs and uncertain replacement timing. |
| Slower adoption | End users take longer to deploy AI or generate revenue from it. | Unclear productivity gains and delayed customer budgets. |
These risks do not automatically invalidate the long-run thesis. They explain why annual spending could rise, flatten or fall temporarily while the broader infrastructure transition continues.
Rank #4
What the GDP estimate does—and does not—say
Pick’s estimate that roughly $1 trillion of annual AI capex could contribute 50–70 basis points to U.S. real GDP over five to 10 years links the spending cycle to the wider economy. The contribution would depend on how much spending occurs in the United States, how quickly projects become productive and how efficiently capital is allocated.
A GDP contribution is not the same as a guaranteed productivity boom. Capital can be spent on projects that later earn weak returns, and a large construction phase can precede the benefits from applications. Investors should treat the estimate as a scenario for economic activity, not as evidence that every AI-related asset will appreciate.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Why Morgan Stanley is emphasizing the opportunity
The investment cycle is directly relevant to an investment bank that advises companies, arranges financing and allocates capital. Morgan Stanley’s 2026 shareholder letter said accelerating AI deployment may fundamentally change how the firm originates and executes business while its trusted-adviser role remains unchanged.
Best Value
- It can be a gift option
- Comes with secure packaging
- Helpful in various ways
Pick’s framing positions the firm across several parts of the buildout: advising companies on strategy and transactions, structuring financings, providing or arranging capital, and helping investors assess public and private opportunities. That positioning explains why he discusses both the size of the spending pool and the risks that could interrupt it.
How to interpret the forecast as a reader or investor
- Separate duration from precision. “Five to 10 years” describes a possible span; it does not specify the same spending level in every year.
- Track revisions. The move from $575 billion to about $850 billion for 2026 and from about $700 billion to $1.3 trillion for 2027 shows that estimates can change materially within months.
- Watch physical constraints. Power, grid connections, construction schedules and equipment supply can matter as much as demand forecasts.
- Test the adoption case. Spending only creates durable returns if customers use the capacity and receive enough value to keep paying for it.
- Expect dispersion. A large aggregate market can still produce winners and losers depending on utilization, financing costs, technology choices and capital discipline.
The central message is therefore measured rather than purely bullish: AI infrastructure may support one of the longest capital cycles in a generation, but its timing, scale and economic payoff will be uneven.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
