Blackstone chairman and CEO Stephen Schwarzman had his eureka moment while stuck on a bus in China.
Eight years ago, while on his way with other global business leaders to a management conference in Beijing, Schwarzman sat next to Jack Ma, co-founder of Chinese tech giant Alibaba Group Holding. As the bus stalled for more than an hour in traffic, Ma began to tell Schwarzman about the promises and perils of artificial intelligence.
“I had no idea what he was talking about,” recalls Schwarzman. “So he gave me a quick tutorial.”
Schwarzman was dazzled by the new technology’s possibilities, and on his return to New York, he had Blackstone recruit a team of data scientists to begin disseminating AI usage throughout the firm.
Today, Blackstone, the world’s largest alternative-asset manager, relies on AI to assess risks in assets it aims to acquire. It employs AI to forecast demand and instantly provide prices for every business customer using its vast inventory of e-commerce warehouses. Within the company, data scientists sit on investment teams to ensure that AI tools help close deals or turn them down.
AI can even enliven the usually buttoned-down Blackstone presentations to investors and portfolio company executives. For one such event, the Blackstone host asked ChatGPT to help him channel his “inner Tony Robbins.” Following the chatbot’s script, he stirred up the audience with a rousing, “Are you excited to talk about the most disruptive technology of the next few decades?!!”
Hype notwithstanding, AI is transforming Blackstone at a faster clip than any of its rivals.
Blackstone’s business model is well suited for AI. With $1 trillion of assets under management spread over more than 230 portfolio companies and massive holdings in real estate, private equity, and credit funds, Blackstone owns a plethora of the proprietary data that is the lifeblood of AI.
“If you are a large asset manager, you have a lot more data to train your AI models than a small player,” says Vik Sohoni, a senior partner at consultancy McKinsey & Co. who covers financial institutions. “You will also attract a lot more talented people.”
At Blackstone, that infusion of talent includes more than 50 data scientists — far more than its peers employ. Early on, these data scientists concentrated on predictive AI, which scours through data to forecast everything from budgets and sales to customer clicks on a website.
In the last couple years, Blackstone data scientists have combined predictive AI with generative AI, which recently captured the public imagination thanks to ChatGPT. Generative AI creates new images and text from the data it has been trained on to instantly gain insights and recognize patterns across every possible business activity.
“We’ve been incorporating data scientists in our equity businesses and in a bunch of our portfolio companies,” says Jonathan Gray, Blackstone’s president and Schwarzman’s anointed successor. “We believe that gives us a real competitive advantage.”
Nor is Blackstone — an investor and financer for entertainment companies like Reese Witherspoon’s Hello Sunshine that have been paralyzed by striking actors and screenwriters — immune to concerns that AI will reduce incomes and wipe out jobs.
Meanwhile, suspicions abound that AI is being used as a marketing tool, in much the same way that hoopla over environmental, social, and governance strategies helped companies lure investors. “There’s always hype with every new technology,” says Ju-Hon Kwek, another senior partner at McKinsey, who tracks AI in asset management. “Hype attracts capital.”
And it’s true that Blackstone could use an AI boost in its fundraising, which, like the rest of the alternative-asset management industry, has lagged recently. In the second quarter of this year, capital inflows at Blackstone were only a third of the assets raised during the same period in 2022.
“Wherever there is excitement about returns, about where the world is going, it’s easier to fundraise in those areas,” concedes Gray.
But the doubters can’t diminish AI fervor at Blackstone. In quarterly phone calls with the firm’s portfolio company CEOs, Schwarzman insists they embrace AI or risk falling behind rivals. “This is a transformational technology,” he tells them. “You have to be the first mover in your industry.”
And to help these companies out, Blackstone offers them the same blueprint it has embraced.
First, make chief executives feel personally comfortable with AI. Encourage them to use ChatGPT or Google’s Bard to write emails and memos — maybe even in iambic pentameter or in Taylor Swift-like verses just for fun.
Then appoint a tech-minded senior executive to oversee the implementation of AI in the company to improve productivity, create new products and services, and reach out to an ever-larger number of customers.
Blackstone also urges its portfolio companies to make full use of its own AI resources, including its deep well of proprietary data.
“We ask them: What are the biggest challenges that you have been unable to tackle?” says Matt Katz, who heads Blackstone’s data scientists. “If it’s a private aviation company, is it forecasting future flights? For software businesses, how do you approach identifying potential customers and driving cross-sale strategies? For a real estate logistics business, where is there pricing upside?”
At Blackstone itself, one of the more visible impacts of AI is in its investment process.
Martin Brand, Blackstone’s head of North America private equity, is on the phone almost weekly with bankers offering him investment ideas. Before AI, he would turn over the more plausible pitches to his analysts, who took at least two days to come up with an investment model and send him a PowerPoint presentation.
“Now when I get a call from an investment banker, we have AI tools that can build a high-quality first model while I’m still on the phone,” says Brand. “And this is already used throughout our private-equity business.”
AI modeling led Brand to reject investments in two publicly listed companies — one involving call centers and the other in education — because they are the kinds of businesses likely to be disrupted by recent AI developments. In both cases, AI models predicted a slowdown in new customers and lower customer retention rates.
Once Brand decides to further explore a possible deal for an asset pitched to him over the phone or in person, he turns it over to a deal team that includes data scientists.
The starting point in the investment process is the creation of a first-task memo by junior staff. Nowadays, in most cases, writing a first-task memo still requires two or three days to dig through data, build analyses and models, and produce a detailed plan that is intelligible to Blackstone executives and limited partners.
But aided by Blackstone’s own AI platform, the process can be done within hours, delivering a memo written in familiar Blackstone prose and building an actual model of the desired asset filled with data.
“Is it perfect? No, but it gives you a solid baseline to work off,” says John Stecher, Blackstone’s chief technology officer.
Besides using its own AI platform for dealmaking, Blackstone contracts with outside AI services for a variety of tasks.
Microsoft’s Office 365 suite — backed by ChatGPT — writes and summarizes emails, creates PowerPoint presentations, and speeds up other boilerplate chores.
Salesforce, another big tech player, provides an AI product called Einstein GPT. Blackstone uses it in its private wealth management business to contact clients with personalized investment suggestions. Stecher compares Einstein GPT’s features to Google’s ability to tailor ads to customers’ previous purchase preferences.
On the engineering side of Blackstone, there is an AI product called Cody, from Sourcegraph, that surveys Blackstone’s abundant, fragmented codebases and then writes software codes and creates new software. Stecher estimates that Cody cuts in half the time needed to deal with those tasks.
For all the AI to work at its best, Blackstone had to break down the silos separating databases in accounting, credit, trading, real estate, and the rest of its business activities and bring all the data together in one place. That place is Snowflake, a company that acts as Blackstone’s data warehouse and receives updates from the asset manager by the hour.
Stecher waxes confident about Blackstone’s efforts to prevent cybersecurity intrusions. His technology department has overseen the acquisition of cybersecurity startups that are supposed to prevent AI leaks from Blackstone’s enormous pool of proprietary data.
“We spend a disproportionate amount of time fact-checking and validating that our platforms are performing the way we expect them to,” he says. “We go through reams and reams of data to detect any anomalies that correspond to intrusions in our networks.”
Stecher’s conclusion: “We haven’t yet seen anybody doing anything nefarious.”
Lawrence Glazer isn’t reassured. Glazer is co-founder and managing partner of Mayflower Advisors, a Boston-based registered investment advisor of about 40 employees catering mostly to clients with less than $5 million in liquid assets.
Blackstone is his go-to firm for alternative assets. He and members of his staff attend Blackstone conferences and briefings, including recent pep talks on AI. He is also inundated by AI pitches from other firms promising insights and shortcuts to building investment portfolios and retirement plans.
“It’s the cybersecurity issue that really keeps me awake at night,” says Glazer. “No amount of spending is too much to protect our data and our clients’ money.”
Glazer’s apprehension is stoked by the outpouring of media reports on cybersecurity breaches: hacks into companies’ cloud accounts, the manipulation by outsiders of a firm’s AI tools to gain access to protected data, and experiments by researchers that demonstrate how easily hackers can penetrate the cybersecurity of even generative AI leaders like OpenAI and Google.
“Everybody says AI is going to be the greatest thing in the world,” says Glazer. “But it’s also an existential risk.”
These are properties leased by Amazon and other e-commerce retailers as distribution centers for the final stage in the delivery of products to consumers. Blackstone began buying warehouses in 2010 and soon noticed that e-commerce firms were leasing these spaces at a frenzied pace. Over the next 13 years, Blackstone amassed $175 billion worth of warehouses worldwide, making them its single-largest asset class.
Not only was demand for warehouses growing, but so was the variety of business customers. It became necessary for Blackstone to create an in-house firm, Link Logistics, in 2019 to manage a half-billion square feet of warehouses and service 11,000 customers ranging from e-commerce giants to small family businesses.
Blackstone soon discovered that the long-reliable Excel spreadsheets weren’t quick enough to make the myriad decisions on pricing and demand forecasting or to pick up near-term and seasonal business trends. So Blackstone placed a team of data scientists in Link who built a centralized, AI-powered algorithm to drive all pricing decisions across its logistic assets in the U.S.
To feed its AI tools, Link counts on a mammoth amount of proprietary data gleaned from existing leases on its thousands of buildings. AI then determines instantly how much warehouse space a customer needs for a million beer cans or a dozen scooters, how far away the warehouse is from a FedEx or UPS hub, what the closest airport or ship harbor is, and how much it’s going to cost the customer.
“AI has allowed us to draw insights from all that data in a way we could only do with that kind of computing power,” says Link CEO Luke Petherbridge. “And if this were baseball, we’re only in the first or second inning.”
Besides using AI to transform its internal operations, Blackstone is raking in AI-related profits from the services provided by its data centers. These warehouse-size buildings contain the computers and other digital infrastructure that enable business clients to process the data essential for AI operations.
“In digital infrastructure, there is a well-publicized arms race happening in AI,” Gray told analysts in July. “And the major tech companies are expected to invest $1 trillion over the next five years in this area, mostly for data centers.”
Rivals are playing catch-up with Blackstone. The firm paid $10 billion in 2021 to acquire QTS Realty Trust, one of the largest data center businesses in the world. Since then, with the advent of generative AI, demand for data center services has led to a tripling of QTS’s capacity.
Customers range from so-called hyperscalers like Microsoft and Netflix all the way down to tech startups. And with 2,000 acres of real estate across 36 locations in the U.S. and Europe, Blackstone can move quickly to build new data centers at sites that suit its customers.
Alts competitors Brookfield Asset Management and KKR & Co. are also heavy investors in data centers. Because data centers are prodigious electricity consumers, Brookfield plans to leverage its leadership in green energy — the preferred source for the tech industry — to collect additional revenue. And because close to half the energy consumed by data centers is used to prevent their computers from overheating, KKR has acquired a firm that provides cooling systems.
A continuing, costly problem for Blackstone and its peers is the need to upgrade older data centers with the more advanced computers and semiconductor chips that are essential for generative AI. (For example, a ChatGPT interaction requires ten times the computing power of a regular Google search.) So in August, Blackstone took a lead role in extending $2.3 billion in debt financing for CoreWeave, a startup that expedites the construction of generative-AI-ready data centers.
Incorporating more AI tools into its portfolio companies is another important source of revenue and profit growth for Blackstone. But not all these businesses are as buoyant as logistics and data centers. Blackstone Growth (BXG), whose mission is to bring potentially fast-growing companies into the parent firm’s portfolio, has a spotty record.
Some of its investments, such as dating app Bumble, have been winners. Others, like Oatly, which produces alternatives to dairy products from oats, have seen valuations plunge.
Hard-pressed to meet its fundraising goals, BXG is hoping to get an AI-aided boost for more of its portfolio companies. One in particular — DNAnexus, in which Blackstone is a leading minority shareholder — is often cited as a prize example.
DNAnexus helps pharma and biotech firms handle massive amounts of genomic data that transcends the capabilities of even the largest drug companies. The firms use DNAnexus to speed up the drug discovery process and help personalize medicine in what’s being hailed as the new frontier of health care.
Utilizing an AI-aided data search by DNAnexus, one customer, Regeneron Pharmaceuticals, uncovered a gene mutation in some overweight people that inhibited heart attacks. This led Regeneron — in collaboration with multinational pharma company Sanofi — to develop a drug called alirocumab that helps reduce cholesterol levels in people who don’t fully respond to low-cholesterol diets and statin treatment.
Blackstone has worked to expand DNAnexus’s data infrastructure by installing a data scientist. The portfolio company also turned to Blackstone to help put together a strong board of directors.
“We specialize in companies during their proverbial adolescence,” says Jon Korngold, head of BXG. “If they want to get to a successful adulthood, they need the scale and resources that Blackstone offers.”
But by then, it will probably prove impossible to put a brake on the AI rollercoaster. McKinsey predicts that generative AI’s impact on productivity will add at least $2.6 trillion annually to the world economy. And Schwarzman, now 76, predicts that before he retires, an AI-driven Blackstone will double its current assets under management to $2 trillion.
“So strap yourself in for this one,” he says.
