The AI infrastructure investment boom is facing a severe supply-demand mismatch crisis.

According to analysis by Bloomberg columnist Chris Bryant, while the tech industry is pouring unprecedented amounts of money into AI computing infrastructure, data center construction is lagging severely due to multiple obstacles including power shortages, supply chain bottlenecks, and regulatory approval delays. Morgan Stanley analysts warned in a recent report that "there isn't enough power to support everyone's plans, and the number of semiconductors being produced has exceeded the places available to house them." This assessment echoes similar statements previously made by Musk.

The direct market impact of this mismatch cannot be ignored: the power gap facing US data center developers is projected to reach approximately one-third of the electricity needed for chip sales forecast by Morgan Stanley for the same period through 2028. This means the stock market's optimistic expectations for semiconductor shipments face the risk of being disproven, potentially creating a cascading impact on the financial performance of chip manufacturers, data center developers, and their tenants.

Half of Major Projects Have Yet to Break Ground

The actual progress of data center construction is far behind market expectations. According to data from market intelligence firm Currence, half of the large data center projects announced this year have not yet begun construction.

Construction obstacles span multiple levels: equipment and labor shortages, community opposition, construction bans, permit delays, as well as long grid connection cycles and insufficient gas turbine capacity related to energy supply. BloombergNEF data further shows that by capacity, nearly half of the US data center pipeline under construction is led by developers entering the field for the first time, and the lack of industry experience has left observers lacking confidence in whether problems can be resolved quickly.

Meanwhile, rising borrowing costs and lenders becoming more cautious are also intensifying uncertainty around project financing. Current projections from various institutions for the computing capacity that can ultimately be built over the coming years diverge significantly, reflecting widespread doubts about when supply chain and regulatory bottlenecks will be cleared.

Contract Protections Have Loopholes as Developer and Tenant Risks Differ

For data center landlords and lenders, contract terms can to some extent provide cost overrun protection and constrain tenants' exit rights in the event of delays. However, according to Bloomberg colleague Paul Davies, such protection mechanisms vary by contract, and tenant qualifications and guarantor quality are equally critical.

The IPO filing of London-based startup Nscale Ltd. reveals potential risks: its $44.6 billion computing contract with Anthropic "provides only limited remedies" in the event of supply chain delays. If Nscale fails to deliver on time, Dario Amodei's Anthropic has the right to terminate the contract "without liability."

Oracle Corp. (NYSE: ORCL) similarly faces such pressure. The company had to assure investors that its lease for the Jupiter project data center in New Mexico is progressing normally, even though the project has been hampered by natural gas pipeline permitting issues, and Oracle's subsequent force majeure notice did not boost market confidence. It is understood that even if rent can be deferred, Oracle must still bear partial payment obligations to project financiers.

SoftBank's SB Energy: $50 Billion Valuation Faces Scrutiny

The case of SB Energy, under SoftBank Group, epitomizes the current concerns about AI infrastructure investment bubbles. The company disclosed in its IPO prospectus that it has a contracted computing project pipeline of 8.8 gigawatts, mostly for OpenAI, but only a very small portion is currently under construction, and no data centers are yet operational. This situation has led investors to question its $50 billion valuation.

Notably, NVIDIA Corp. (NASDAQ: NVDA) is intervening with a $105 billion guarantee—backing tenant OpenAI's lease obligations to support SB Energy's construction progress. This move itself indicates that NVIDIA is highly alert to the possibility that data center construction delays could affect its processor sales prospects. NVIDIA also explicitly cautioned in its latest earnings report that shortages of land, power, buildings, or capital could all constrain its future financial performance.

Hyperscale Cloud Providers Face Inventory Dilemma

For "hyperscale" cloud computing giants like Amazon and Microsoft, the supply-demand mismatch between chips and data centers is evolving into a tangible financial risk. These companies typically purchase servers and networking equipment "months" before data centers become operational, so they are reluctant to stockpile chips in advance that cannot be powered and face technology obsolescence risks.

Currently, the scale of assets held by AI companies that have not yet been put into use has approached $350 billion. To bridge the capacity gap before new sites come online, BloombergNEF suggests replacing older chips in existing data centers with NVIDIA's latest, more computationally efficient models. However, this would further increase hyperscale cloud providers' already massive capital expenditures and add depreciation pressure.

Another potential path is relocating data center capacity to countries with more abundant renewable energy, which could provide Europe with more opportunities to participate in AI infrastructure construction.

NVIDIA Market Cap Approaches $6 Trillion but Risks Already Appear in Earnings

Although NVIDIA's market capitalization has approached $6 trillion, investors currently appear not overly concerned about Jensen Huang's company's ability to find buyers for its high-margin processors. However, NVIDIA has explicitly listed data center construction bottlenecks as a potential risk factor in its latest earnings report, and this statement itself deserves market attention.

Currently, supply chain bottlenecks are indeed delivering substantial returns to some participants in the short term: Musk's SpaceX is renting out computing capacity at high prices, Caterpillar Inc. (NYSE: CAT) and Bloom Energy Corp. (NYSE: BE) have on-site power generation equipment orders piling up, and top electrical engineers are commanding salaries as high as $750,000. But whether this localized prosperity can continue depends on whether overall infrastructure construction can keep pace with chip shipments.

As Chris Bryant put it, after years of revelry, the AI feast may be turning into a high-stakes game of musical chairs—ultimately, some may be left holding too many chips with nowhere to put them.