Trillion-dollar AI feast: How private equity and credit can get a piece of the pie

Recently, a topic has become the focus of Wall Street discussions: what kind of wealth bonanza will the artificial intelligence frenzy bring. Although most of the speculation surrounding artificial intelligence has so far taken place in the stock market, as indicated by the stock price of chip manufacturer NVIDIA, this excitement is spreading to the serious fields of debt financing and private equity. According to Bloomberg analysis, at least $1 trillion in spending is needed for data centers, power supplies, and communication networks, which will provide the impetus to fulfill the promise of artificial intelligence transforming all areas from medicine to customer service. Some believe the total cost could be twice that figure. Even Wall Street skeptics like Jim Covello, stock research director at Goldman Sachs, who are doubtful of the ultimate profit potential of artificial intelligence, acknowledge that it is worth continuing to invest in companies that provide pipeline services. The tech giants of Silicon Valley and Seattle have amassed wealth and hope to gain some capabilities for cloud storage, providing them with some comfort. All major banks are racing to keep up with this wave of business boom. According to an insider, JPMorgan Chase has established a dedicated infrastructure team to lead its workforce, and Deutsche Bank and other banks are doing the same. A rival banker admitted that his company is handling too many data center transactions to have enough staff to cope with the workload. Debt financing is no different. Morgan Stanley stated that banks do not have enough balance sheet to meet credit demand, so it proposes to work with private capital: there is room for everyone at this banquet. For investment bankers, opportunities are arriving as many seek their next meal ticket. Providing debt financing to companies has long been a major engine of Wall Street profits, but this business has recently gone through a difficult period. Despite the strong momentum in the public stock markets in recent years driven by artificial intelligence frenzy, the return on investment-grade credits has always been low. Leveraged finance teams providing funds for high-risk private equity acquisitions have suffered losses due to a decline in mergers. Dominik Thumfart, head of infrastructure and energy sources for Deutsche Bank in Europe, the Middle East, and Africa, said, "The market outlook is very optimistic. In the coming years, this market will remain a major growth area in financing. The investment curve is very positive." The German bank has provided $170 billion in financing for data centers over the past three years. It's not just the funds for massive new data centers that are at work. The enterprise financing of the two most inactive areas has come to life due to the tech industry's desire for processing power: utilities and telecommunications have suddenly become the hottest credit markets. For private equity giants like Apollo Global Management and KKR, digital infrastructure is an opportunity to turn over a challenging page, when high interest rates disrupted the economic benefits of deals reached during the era of cheap funding. Companies including Blackstone, Brookfield Infrastructure Partners, and Stonepeak Partners have been acquiring data centers and laying the groundwork for the construction of these data centers. Larry Fink, CEO of BlackRock, said that after partnering with Microsoft to fund the development of infrastructure for buildings and energy, his company will raise up to $120 billion in debt related to data centers. Bankers and private lending institutions will be eager to take a share from such transactions. However, even within the artificial intelligence frenzy, there are some cautionary points. Some point out that while private equity firms have a strong foundation in real estate, they have not yet undertaken massive and costly construction work like these data centers these data centers are called "artificial intelligence factories" in the words of NVIDIA founder and CEO Jensen Huang. Innovation continuously disrupts technology, adding risk to long-term capital projects. And followers of artificial intelligence have not yet envisioned a "killer app" to match the success of e-commerce and GPS location newcomers in the Web 2.0 era. Even if they could, the smartest minds in the tech industry are working hard to improve the efficiency of software and hardware to reduce the demand for scale and power. Benjamin Fernandez, head of Esoterics ABS business at Barclays Bank, said, "There are many reasons to remain optimistic." The business mainly involves bonds supporting "non-traditional" assets such as data centers. "But if for some reason, this investment is not successful, and people cannot find a way to monetize artificial intelligence investments, then there may be risks." Active tech giants Tech companies such as Amazon, Microsoft, Google under Alphabet Inc., Meta platform, and Apple are ambitiously looking to manage, process, and manipulate Beijing Vastdata Technology, and with huge cash reserves, they are referred to as "mega-enterprises". Their spending indeed matches these exaggerated terms. Craig Scroggie, CEO of Australian data center group NEXTDC Ltd, said in October that mega-enterprises had invested $52.9 billion in artificial intelligence infrastructure in just three months. Real estate brokerage firm Jones Lang LaSalle (JLL) said that construction projects for U.S. data centers had grown more than sevenfold in two years, further proving the "unrelenting demand" for computing power. Host data centers rent rack space to tech companies. The company estimated in a report in August that the rent for these facilities had risen by 37% in the past 12 months. All this unrestrained spending has accelerated the issuance of investment-grade debt and high-risk leveraged loans, particularly in the United States, which is convenient for private lending institutions and fee-starved investment bankers. Hedge funds are also seeking to profit from the artificial intelligence boom with new debt structures. This has also opened up the asset-backed securities market for Newland Pharm's [Chinese pharmaceutical company] investment opportunities.According to data compiled by Bloomberg, the debt sales supported by data centers this year have surged to nearly a record-breaking $71 billion. When including fiber optic networks and other equipment, the sales figure will be even higher. Matt Beisner, who is in charge of business at Guggenheim Securities in this sector, said that the number of buyers for his data center ABS products has roughly doubled in the past four years.In the past few years, the debt supported by data centers has increased sharply. Nasir Hussain, founder of real estate investment bank Brookland, said: "Due to the strength of these mega-enterprises, it is one of the most popular asset classes for stocks and debt." Blackstone Group has been keen on data center acquisitions and is currently entering the data center development field. In September of this year, Blackstone Group reached an agreement with the Canadian Pension Plan Investment Committee to acquire AirTrunk for about 24 billion Australian dollars (15.5 billion US dollars). Previously, Blackstone Group acquired QTS Realty Trust, which owns dozens of data centers in the United States, for 10 billion US dollars. The company also plans to build a factory in a location in northern England that was originally intended to be a battery plant. The cost of the factory could be as high as 10 billion pounds (13 billion US dollars). Although Blackstone Group has not risked investing so much money in construction before, if everything goes well, data center developers can receive generous returns. Real estate research company Green Street estimates that the profit margin of data centers in London is about 65%. Financiers are enthusiastic about supporting these large projects because future users usually have already signed long-term leases in advance, making these projects safer. According to sources, some banks are willing to provide loans up to 70% or 80% of the cost, sometimes even higher if a lease has already been signed. Marie Baetz, senior credit research analyst at Generali Investments, said: "Recently we have seen many new people joining this field. But I think the overall financing and approach are very rigorous." However, a banker in the industry stated that loan institutions are more cautious about data centers specifically used for artificial intelligence rather than general purposes. He noted that such transactions may bring higher debt costs and lower leverage as the technology still needs to prove its value. In addition, a senior partner at a leading private equity firm expressed concern about the emergence of speculative developments, meaning starting construction before finding tenants, as it is difficult to determine the final demand. Some lawyers refer to potential "zombie projects" that may never be completed. Not everyone believes in the strategy of "if you build it, they will come" is foolproof for those betting on breakthroughs in artificial intelligence. Darone Acheimoglu, a professor at MIT, expressed concern that a lot of funds will be wasted. Despite doubts, bankers and private loan institutions' interest in transactions, especially for sites where blue-chip companies have signed leases, gives most data center owners and developers an advantage in pricing debt. Hussain of Brookland said that tech giants can obtain bank financing for sites leased on a long-term basis at a profit rate below two percentage points. He added that co-location providers generally pay 2.5 percentage points or lower. Jordan Sadler, senior vice president of Digital Realty Trust Inc., said: "Recently, we issued 850 million euros (907 million US dollars) of nine-year bonds at a rate below 4% and refinanced and increased our revolving credit facility to 4.5 billion dollars." The company has signed joint venture agreements with companies like Blackstone Group to invest nearly 9 billion US dollars in the development of mega-scale data centers. Power Play The main issue faced by energy-consuming data centers and their owners is ensuring stable access to overloaded power grids - although this also brings another day of rewards for lenders. In the race to expand electricity capacity, they are prepared to support all direct transactions between utility investments, tech companies, and nuclear plant owners. Demand is urgent. According to a report obtained under the Freedom of Information Act, Irish electricity transmission operator EirGrid warned last year that due to the country's failure to sign new connection agreements, "some major data centers decided to withdraw". This was a major problem, as revealed for the first time by the transparency organization Right to Know. A utility company on the other side of the Atlantic China Welding Consumables, Inc. told the Atlanta Federal Reserve Bank that the electricity consumption of data centers has increased by 17% in recent months. According to Dominion Energy Inc., Virginia is the state with the highest density of data centers in the world, and the peak demand for electricity reached a record six times in July. To meet the energy needs of data centers, the capital expenditure of utility divisions is expected to exceed 200 billion dollars next year, about double what it was ten years ago. This will put pressure on the asset balance sheets of utilities, but Moody's recently relaxed its view on some higher risk hybrid bonds in the industry - allowing them to be considered quasi-equity - opening the door for companies to raise funds without being downgraded. Data shows that the sales of these bonds have increased nearly eightfold this year, reaching 15 billion dollars. Only the bond issuances of large banks can compete with them. Sinjin Bowron, portfolio manager at the hedge fund Beach Point Capital Management, said: "A lot of demand comes from the electricity supply for data centers, with the biggest driver being artificial intelligence. In leveraged credit, there is no direct way to leverage the power of artificial intelligence, but electricity generation is one of them." Like data centers, this has attracted the attention of acquisition giants like Carlyle Group, especially in credits and equities related to nuclear power. From the reopening of decommissioned nuclear power plants like Three Mile Island to the installation of small modular reactors, tech companies are exploring some radical and expensive ideas for power sources. The telecom industry is another industry that has been resurrected by artificial intelligence. Data shows that the industry's transaction volume has more than doubled compared to last year. Private capital has become creative in chasing shares in artificial intelligence, wantingClever moves such as accepting Nvidia chips as collateral for loans have been made.However, the data center remains a major focus. For investors funding this large-scale construction, the pressing issue is how to ensure they achieve the same returns as they did when cashing out from previous real estate investments such as warehouses. Due to the relatively new nature of AI factories, there are not many examples of private equity exits from large-scale investments. Companies leading the way in financing are reassured by the wealth of companies like Google, Microsoft, and their counterparts, believing this will enable them to grow in the long term. "These funds will be spent, and I think the sponsors of this technology, mainly mega-enterprises, have very, very deep balance sheets," said Rob Bittencourt, Partner of Apollo and Co-Head of Opportunistic Credit. "I think, compared to other technology cycles we've seen, this trend will be enduring and instill a sense of patience, if you are willing."