Tech Hits Turbulence

Global markets are confronting the first major challenge to an assumption that has shaped the investment landscape in recent months. The Democratic sweep in the U.S. midterms has brought the prospect of greater scrutiny of AI and data-centre development sharply into focus, but it has hit a market already keeping a very close eye on technology valuations and hyperscaler capital spending.  

While Wall Street has somewhat steadied today, the scale of the sell-off raises a more important question than whether markets had become too optimistic about AI: has anything fundamentally changed? Across our investment teams, the answer is more nuanced than the headline moves suggest. 

Slower Spending 

Raj Shant, Head of International Business Development, Jennison

This market volatility may lead to a more measured build-out which could prove healthy. Permitting requirements are becoming more demanding, while constraints around power, chips and cooling systems are placing practical limits on how quickly new capacity can come online. At the same time, communities may expect more from data centre developments given the economics these assets can generate. 

That could bring greater discipline to a market where speculative projects have emerged without anchor tenants or clear demand. Some of the more aggressive forecasts had AI-related capital expenditure reaching $1.5–2 trillion in 2027. If spending instead lands closer to $1 trillion, you could see pressure ease across the supply chain while capital is allocated more selectively.  

Investors should also consider how AI investment is being financed. Big technology companies have moved from funding much of their capital expenditure through organic free cash flow towards greater use of bonds, credit and equity. A slower investment cycle could reduce that financing requirement, potentially supporting equity holders.  

The total addressable market for companies like this just is exploding, and that's not going to change, even if new capex over the next couple of years is slower. 

Central Banks on Watch

Guillermo Felices, Global Investment Strategist 

This is consistent with our AI scenarios where we distinguish between a steady build-out, a constrained build and a more damaging AI unwind. The middle scenario is particularly relevant today: political, permitting, power and grid constraints could slow projects without undermining underlying demand, reducing financing requirements while increasing the scarcity value of existing compute capacity.  

The most important question is whether this stress develops into something broader. So far, the evidence is mixed. The fundamental economic backdrop remains solid as seen from the strong U.S. jobs data and demand for compute remains strong. If political, regulatory and permitting constraints result in a slower AI infrastructure build-out, growth could moderate at the margin.  

That could also make central banks more cautious about their policy tightening campaigns and reduce the extraordinary volume of hyper-scaler debt issuance the credit market has been absorbing. For credit investors, less supply could be supportive for spreads.  

The risk is that markets turn a correction into an economic event as a deeper equity decline could hit U.S. consumers through the wealth effect, tighten financial conditions and push the economy towards the more bearish AI-unwind scenario. 

Differentiating Data Centers

Morgan Laughlin, Head of Global Data Center Investments 

We view the current turmoil principally as a short-to-medium-term market adjustment rather than a fundamental break in the long-term case for digital infrastructure.  

Not all data centres are the same. Much of the concern now weighing on markets centres on huge frontier-AI campuses, where hundreds of megawatts of incremental demand can require substantial new generation and transmission infrastructure. That is different from the low-latency cloud infrastructure where demand is more resilient. 

The election result may have accelerated the market’s reckoning with the scale and pace of AI investment, but that is not the underlying cause. Public concern about data-centre development increasingly cuts across party lines, particularly where large campuses affect power availability, electricity costs and local communities. The likelihood was always that development would encounter greater political and regulatory friction. 

That could create near-term challenges. Lenders may become more cautious, equity raising may be harder and refinancing more expensive as markets initially treat data centres as a single category. Yet the long-term drivers behind cloud infrastructure have not disappeared. Digitalisation continues and businesses and consumers still require low-latency capacity.  

Additional regulatory hurdles could, paradoxically, reinforce the value of projects that already have approvals, power and community support by slowing new supply. For investors, the implication is to become more selective rather than retreat from the sector. Planning, power availability, local opposition, financing structures and the end-use of capacity matter more in this environment. 

6096881_PGIM