Private Credit Direct Lending: Navigating the AI Investment Bubble
AI is not the bubble. Valuations are.
AI will transform how the economy works. That does not mean every company with "AI" in its narrative deserves to survive—let alone trade at a premium. In a market now facing $100 oil, expanding tariffs and growing scrutiny of AI capex, the question has shifted:
Who actually earns a return on this AI buildout—and who is simply funding it?
This is where the real AI investment bubble sits: in treating participation in the theme as a business model.
Why Investors Are Asking if There’s an AI Investment Bubble
The AI trade is running into a three-front test.
The three-front test: $100 oil, tariffs and AI spending fatigue
Three macro forces are colliding at once:
- $100 oil and higher energy costs
AI is energy intensive. Training and running large models consumes vast amounts of power. Higher energy prices feed headline and core inflation, complicate the rate path and increase operating costs across the AI stack. - Expanding tariffs and supply chain friction
Tariffs act as a tax on global supply chains. For AI, that means more expensive hardware, components and infrastructure. It also raises execution risk for capex-heavy projects dependent on complex, cross-border inputs. - Growing anxiety over AI capex and payback
We are in the middle of one of the largest infrastructure buildouts in modern market history. Investors are starting to ask the question that eventually follows every spending boom: When does all this AI investment become revenue, cash flow and durable returns?
From “must own AI” to “show me the cash flow”
Early in any secular technology shift, the market overpays for exposure and underwrites almost no skepticism. Ownership of the “right theme” becomes more important than underwriting of the actual economics.
AI is moving past that phase.
- The market is no longer granting blanket amnesty to any company with an AI story.
- Investors are starting to separate the AI revolution from the AI valuation.
- Capital is quietly rotating from pure narrative to measurable cash flow and balance sheet resilience.
That is what a maturing AI investment bubble looks like: the theme persists, but the easy money for low-quality labels disappears.
AI vs the Dot-Com Bubble: The Technology Wins, Most Labels Don’t
The dot-com era gave investors a painful but valuable template for today.
What the dot-com era got wrong about value creation
In 1999, owning a dot-com domain was treated like having a business model. It wasn’t.
The internet ultimately changed almost every sector of the economy. But:
- Most companies with “.com” in their name disappeared.
- Many investors who were technically “right” on the internet were wrong on how to express that view.
- The label "internet company" became a poor proxy for value capture.
The lesson: the technology can win while most stocks tied to it lose.
Why an AI label is not an AI business model
AI is likely on the same path:
- Adding "AI" to a company’s name, deck or strategy does not create a moat.
- Referring to generic "AI capabilities" tells you little about demand, pricing power or unit economics.
- A surge in AI headcount and capex is not the same as a clear path to earnings.
The market is beginning to remember that:
- Technology adoption and equity returns are not the same thing.
- A powerful theme can coexist with a selective, localized investment bubble in its most hyped labels.
The internet changed everything. A dot-com domain did not guarantee survival.
AI will change everything. An AI label will not guarantee a future.
How to Recognize the Real Risk in the AI Investment Bubble
The biggest risk around AI today is not “missing the revolution.” It is overpaying for the label and ending up the last holder of stories that cannot fund themselves.
The danger of funding capex instead of harvesting returns
At this stage of the cycle, many public investors are effectively:
- Funding the buildout of AI infrastructure through equity at elevated multiples, while
- Bearing the risk that ultimate economics accrue to a different part of the stack.
This is a familiar pattern:
- The early phase rewards those who sell the picks and shovels—or simply sell stock into enthusiasm.
- Later phases often reward those who bought claims on the resulting cash flows at rational prices.
If your AI thesis stops at "own the obvious beneficiaries" without:
- Mapping who controls pricing power,
- Stress-testing balance sheets under higher-for-longer rates, and
- Assessing how much of today’s capex is speculative,
…you may be financing the machine, not sharing in its yield.
Signals you’re holding the wrong side of the AI trade
Warning signs that an AI exposure is more bubble than business:
- Narrative density > financial clarity
Investor materials mention AI repeatedly—but provide little detail on unit economics, customer behavior or payback periods. - Perpetual capex with fuzzy milestones
Management emphasizes “staying ahead” in AI, but cannot specify when investment intensity normalizes or how returns will be measured. - Dependence on capital markets
The company relies on frequent equity or cheap debt to sustain its roadmap—dangerous in a world of higher base rates and rising risk premia. - Commodity positioning in a crowded layer of the stack
Dozens of competitors are selling similar AI tools or services, yet investors are pricing in enduring monopoly margins.
When those conditions meet premium valuations, you are no longer just exposed to AI—you are exposed to the AI investment bubble.
What Surviving AI Winners Will Actually Have in Common
The survivors of this phase will need more than a compelling AI deck. They will need the boring, old-fashioned things that have always mattered in credit and equity underwriting.
From narrative to numbers: demand, margins, balance sheets
Over time, the market will reward AI-linked businesses that can demonstrate:
- Genuine demand
Clear, repeatable use cases where customers are willing to pay for AI-driven outcomes, not just pilots and proofs of concept. - Pricing power
Ability to maintain or raise prices because the product is embedded, mission-critical or productivity-enhancing enough to resist discounting. - Defensible advantages
Proprietary data, network effects, distribution, or domain expertise that make displacement costly and unlikely. - Strong balance sheets
Moderate leverage, term structures that can survive higher yields, and no dependence on perpetual, cheap equity to fund capex. - A clear path from capex to cash flow
A disciplined bridge between investment and earnings—grounded in observable adoption curves and realistic margin trajectories.
These are the same qualities that consistently matter in event-driven and private credit direct lending underwriting. The AI label does not change that.
Using AI best vs building AI infrastructure
Another shift is underway:
- The first phase of the AI trade rewarded those building the core hardware and infrastructure.
- The next phase may reward those who use AI best to deepen moats and expand cash flows.
In practice, that may mean:
- Industrials, logistics and business services firms that quietly embed AI to improve throughput and margins.
- Software or data businesses that use AI to increase switching costs and customer lifetime value, not as a marketing tagline.
- Creditworthy borrowers that deploy AI internally to improve operational efficiency, risk management or capital allocation.
The market tends to overprice the visible "AI story" and underprice the operators capturing value in less obvious ways.
Positioning Beyond the Bubble: Where Capital Must Flow Next in AI
If the early trade was “own everything AI,” the more rational next trade is own where capital must flow, not just where narratives are loudest.
Energy, infrastructure and real assets in an AI-driven world
AI’s physical footprint is large and growing:
- Data centers, power generation, transmission, cooling and connectivity all need capital.
- $100 oil and renewed inflation pressures can keep policy rates elevated, increasing the cost of leverage across this ecosystem.
- Capital-intensive projects with long-duration cash flows will be repriced, for better and worse.
For disciplined investors, this creates an opportunity to:
- Target real assets and infrastructure linked to AI demand where scarcity is real, not manufactured.
- Underwrite contracted or resilient cash flows rather than speculative usage assumptions.
- Negotiate terms and security that capture yield while retaining downside protection.
In other words: focus less on the AI label, more on the infrastructure and assets the AI economy cannot function without.
Finding AI value in private markets and overlooked platforms
Many of the most attractive opportunities in the next AI chapter may be:
- Private
Early-stage or growth companies building the application and agent layers on top of existing AI infrastructure. - Overlooked
Mid-market operators integrating AI into processes to quietly expand margins, without rebranding themselves as "AI companies." - Not yet born
New platforms and business models that emerge once the core infrastructure is in place and priced as such.
Private markets can offer:
- More rational entry valuations than crowded public names.
- Greater ability to negotiate covenants, security and information rights.
- Direct alignment with teams building the next layer of the AI economy.
For a private credit direct lending and event-driven lens, this means:
- Lending against real assets and cash flows whose value is reinforced by AI demand.
- Backing operators who use AI to strengthen their economic position, not just to raise capital.
- Structuring exposure so you participate in AI’s upside while remaining senior in the capital structure.
An Operator’s Playbook for Investing Through an AI Investment Bubble
The right posture in this environment is not fear—and not blind enthusiasm. It is operator discipline applied to a secular theme.
Think like a creditor, not a tourist in AI
To navigate the AI investment bubble with institutional rigor:
- Underwrite cash flows, not concepts
Ask how, when and from whom each dollar of AI-related revenue is earned. Require a credible bridge from spend to earnings. - Interrogate capital intensity
High capex is not a problem if it is matched by durable, high-margin demand. It is lethal when funded by fragile balance sheets and equity enthusiasm. - Stress-test against higher-for-longer rates
Assume energy and capital remain more expensive. Focus on businesses and borrowers that survive under those conditions. - Prioritize scarcity you can verify
Real bottlenecks in power, data center capacity, specialized infrastructure or proprietary data—not synthetic scarcity manufactured by marketing. - Seek asymmetric structures
In private credit direct lending and structured deals, target downside protection with upside participation where AI amplifies cash flows.
This is less about timing a top in AI and more about owning the right side of the trade through a full cycle.
Stay informed, stay liquid, move first
Volatility is the headline. The opportunity is beneath it.
In practice:
- Stay informed
Track not just AI headlines but the second-order effects on energy, rates, capital flows and balance sheets. - Stay liquid
Preserve the ability to rotate away from consensus exposures and toward mispriced infrastructure, real assets and private platforms when dislocations appear. - Move first
Position in scarce assets and overlooked users of AI before they become expensive consensus “AI trades.”
The investors who navigate this regime best will believe in the revolution—but be highly selective about who captures its value.
FAQ: Navigating the AI Investment Bubble With Discipline
Is there really an AI investment bubble, or is this just another tech cycle?
The AI buildout is real, and so are the distortions around it. Like the dot-com period, we have a powerful technology wave and a separate phenomenon of investors overpaying for anything associated with that wave. The bubble risk lies in extrapolating early growth, ignoring capital intensity and granting indiscriminate premiums to AI-branded stories.
How should sophisticated investors adjust their AI exposure today?
Shift from owning the story to owning the economics. That means:
- Trimming exposures where AI narratives are rich and cash flow visibility is poor.
- Focusing on operators with pricing power, balance sheet strength and clear AI monetization paths.
- Allocating to infrastructure, energy and real assets that AI demand structurally supports.
- Exploring private credit direct lending and private equity opportunities tied to the next layer of AI applications and platforms.
What indicators suggest an AI stock is more story than substance?
Look for: heavy AI capex with vague payback; dependence on cheap capital markets; thin differentiation in a crowded space; and valuations that assume sustained hyper-growth without corresponding evidence of moats or enduring margins. When management updates focus on theme alignment rather than economic progress, caution is warranted.
Where are the most compelling AI-related opportunities outside crowded public names?
We see potential in:
- Power, data center and connectivity infrastructure with verifiable capacity constraints.
- Mid-market businesses deploying AI to expand EBITDA rather than market multiples.
- Private companies building specific, defensible AI applications in regulated or domain-heavy sectors where incumbency and data matter.
These areas often offer more attractive risk-adjusted returns than competing for the same handful of AI champions.
How does private credit direct lending fit into an AI-driven market regime?
Private credit direct lending can provide exposure to the AI economy while prioritizing downside protection. Lenders can:
- Finance infrastructure and operators with assets and contracts tied to AI-driven demand.
- Structure covenants and security that protect capital in volatile equity markets.
- Capture elevated yields in a higher-rate environment, with potential upside from AI-enabled growth.
This is a way to participate in AI’s economics without relying solely on equity multiple expansion.
Manhattan’s View: Believe in the Revolution, Be Selective About Who Captures Its Value
AI will change the world. That does not mean every AI-branded security deserves your capital.
At Manhattan Private Credit, we approach AI as operators and creditors:
- We study the infrastructure where scarcity is genuine, not manufactured.
- We focus on companies already converting AI into revenue, cash flow and measurable value.
- We search private markets for the applications, agents and platforms capable of becoming the next layer of the AI economy.
The AI trade has been shaken. Expectations are being repriced. Capital spending is being questioned. That is healthy. The next chapter should not reward every company that claims to be part of the revolution—it should reward those that can turn the revolution into value.
Believe in the revolution. Be selective about who captures its value.
Stay informed. Stay liquid. Move first.
Learn more at manhattanprivatecredit.com.