Goldman Sachs' John Waldron Is Rooting for the English Majors as Physical AI, Data Centers and a $500 Billion Bet Redraw Finance

Inside Blackstone is Blackstone's shop window for portfolio and research data. Host Christine Anderson brings a slice of a $39 trillion investor gathering to the studio and invites Amit Dixit, who leads Asia private equity and has spent close to two decades at the firm. The format is tight: three disruptions, each read through an investment lens. That framing matters because it puts the big picture before individual product stories. It also explains why the episode moves quickly from hardware to capital.

The first disruption is physical AI — software that gains a body in robots, vehicles and data centers. Dixit's yardstick is striking: India's total data center capacity is about one gigawatt, less than a single U.S. city such as Phoenix or Chicago. Yet the country has 850 million people under 35 and leads the world on Instagram and WhatsApp usage, a huge demand base. AirTrunk, owned by Blackstone, is one of Asia-Pacific's largest data center platforms, already scaled in Australia. India is just starting. The $30 billion, 5-gigawatt AirTrunk India plan disclosed in June 2026 shows why that threshold is read as opportunity. Like highways before logistics, data centers must come before autonomous vehicles and drones can scale. The build runs in three steps: 1) land and power, 2) high-density cooling and grid connection, 3) continuous power for training and inference, where a trained model answers real queries.

The second data point is humanoid robots. In the first half of 2026 about 97% of global humanoid production happened in Asia, and about 80% of the bill of materials for any robotics application comes from Asia. That means not just assembly but core parts — motors, reducers, batteries, lidar and control boards — are sourced from the same supply chain. Shipment data cited by Gasgoo and SBS backs the picture. The analogy fits: as with smartphones, the supply chain clusters in Asia while design and brand stories travel globally. The result is scale and cost advantage in Asia, with the narrative premium elsewhere.

The third disruption is defense and drones. The war in Ukraine and tensions in the Middle East changed how governments view expensive platforms. Dixit's Japan example is instructive: for $50 million you can buy one aircraft or about a thousand drones at $50,000 each. Most capitals do the same math — Tokyo, Seoul and New Delhi included. The field was underfunded for years; now an "we should have invested earlier" awareness is expanding budgets. The economics are simple: many small, distributed and quickly replaceable systems are more resilient than one costly, targetable system. That opens a new investment corridor across manufacturing, maintenance and software layers for private capital.

Macro Signals Between Washington and the Consumer

In the Economic Weather Report, Winfield Sickles connects micro data to the macro. The Fed hiked 25 basis points last week, as priced, but the signal was the vote and language. The decision was 12-0 unanimous and Kevin Warsh struck a confident tone at the podium. That unanimity cools talk that the Fed has become politicized, at least for now. More important, the hike was framed not as weakness but strength: stronger growth and competition for capital. Sickles ties that to the CapEx cycle he often describes — profits leading capital spending by two to three quarters. The unanimous 16 September 2026 statement and Reuters coverage support that read.

Why is a hike not bad news this time? Sickles argues that for the first time in a long while, strong growth, tighter policy and still-buoyant financial conditions — how supportive credit, equity and bond markets are together — appear together. That mix, rare since the 1990s, keeps equity and capital markets open. In plain terms: money is expensive but the engine still turns. It is a reminder that risk appetite feeds not only on the policy rate but also on profits and cash flow. Strong balance sheets let investment continue even when money costs more.

On the consumer, the picture is broadening. Across Blackstone's portfolio, about 160,000 hotel keys — sellable rooms — in lodging and leisure assets showed weakness a year ago among lower-income consumers. Last week's positive retail sales print confirmed that improvement now reaches beyond the affluent to all segments. U.S. retail data for July–August 2026, up 0.5–0.6%, corroborates the broadening from the outside. The affluent spent first; now the base is joining. When portfolio data and official statistics point the same way, the signal gains weight.

For the macro, Sickles sketches a flywheel — a wheel that, once spinning, keeps itself going. The chain has three steps: 1) higher revenue and profit, 2) CapEx that follows profit with a two- to three-quarter lag, 3) hiring and consumption. Much of that hiring is blue-collar and on-site. The example is striking: at QTS, Blackstone's data center company, on-site headcount goes from 13,000 at the start of 2025 to 40,000 at the end of 2026 and a planned 70,000 next year. Those well-paid field jobs feed straight into household spending and restart the loop. Like a flywheel, the first push comes from profit, then momentum sustains itself.

Capital and Talent Signals From the Goldman Tower

In the main interview, Anderson hosts Goldman Sachs President and COO John Waldron. Waldron has been at Goldman for 26 years, working with CEO David Solomon for more than 30. The pair split about a thousand client meetings a year, putting them among the most CEO-facing teams on earth. Waldron's first diagnosis: clients feel off-balance, optimistic and uneasy at once. Geopolitical crises have multiplied; while he puts a low probability on a slide into a third world war, he says the contours of how it could happen are now visible enough to be unsettling. Add generative AI — models that create text, images and code — as both huge promise and huge risk, and companies are rethinking where and how much to invest and how to manage the workforce.

Waldron says the three questions he hears most are clear: how are you deploying AI, what about inflation, and how do you navigate geopolitics? Beneath them sits a fourth about bubbles and capital spending, but the real curiosity is: does it work for you, does the investment pay back, and how do you manage token spend — the cost per piece of text the model processes? Goldman's model answers with two legs: about 75% Global Banking & Markets and 25% Asset & Wealth Management. Waldron notes U.S. household wealth above $80 trillion, roughly ten times household income, per Fed flow-of-funds. That stockpile is both opportunity and social and political strain. In the optimistic case, AI lifts productivity — more output from the same inputs — and wealth piles faster; Waldron expects broad productivity gains rather than a wave of job destruction. The question is whether gains spread widely.

On private credit — non-bank direct lending — Waldron is blunt. He calls it a great asset class, not a new invention; both Blackstone and Goldman have done it for years, and lending itself is ancient. What separates good from bad is diversification across borrowers, solid underwriting that tests repayment capacity and collateral, and strong documentation. The critical warning: these instruments are private and not liquid by nature; they belong in the private bucket, not the liquid bucket. Investors who may need cash soon should not park it there. He frames sophisticated investors, transparent advice and education as a shared responsibility of firms like his. Outside, the Financial Stability Board's 2026 vulnerabilities report and Goldman's own cracks-in-private-credit notes complete the caution: when standards slip in a fast-growing market, risk piles up, so discipline matters.

When will productivity show up in the numbers? Waldron says Goldman entered 2026 as a learning year — what works, what costs too much, which model for which job — and expects a real unlock in 2027. His personal example is preparation: gathering the information needed for a meeting now takes less time, giving capacity back in the day. On Blackstone's side, monthly token spend is up ninefold; the economic impact is still early innings. Waldron stresses speed: models improving themselves and the tempo of new model launches is unlike anything seen before. At the same time he elevates cyber risk to the top of the list; it is on everyone's list, but should be at the top, not the middle or bottom. Like a seat belt, its value shows not in the crash but in daily discipline.

On capital, the concrete bet is the roughly $500 billion consortium around NVIDIA. Participants include Apollo, BlackRock, Blackstone, Brookfield and Goldman Sachs. Waldron's thesis is supply and demand: from his own firm and client conversations, demand for compute — the processing power to train and run models — outstrips supply, and $500 billion alone does not tip the balance into oversupply. NVIDIA's clever move is to look for a new way to finance compute, as with auto loans or mortgage securitization, where pools of loans were packaged into securities. The logic is identical: not every firm can buy an expensive asset outright; financing broadens access. Blackstone and Goldman can buy compute today, but many enterprises cannot. Waldron's caveat is to build this responsibly, without excess leverage — borrowing to amplify — or weak protections. Reuters and CNBC on 10 August 2026 sketch that scaffolding.

China, Talent and Leadership: Why Curiosity Is Back

On China, Waldron speaks of "two Chinas." One is the domestic economy: pain that, he says, Americans would not tolerate if it were at home. The other is the innovation economy: speed and density that feels like Silicon Valley or Cambridge, Massachusetts. Both run at once. China's focus, in his view, is less on the race to artificial general intelligence — systems that solve general problems like a human — and more on deployment, efficiency and productivity: delivering high quality at low price. The question he leaves open: if they become the low-cost winner in innovation, what does that mean globally? He does not claim an answer, but the question matters. On talent, his thesis is sharper: curiosity, analytical problem solving and mathematical foundations come back into fashion, opening space for English, history and philosophy majors. Waldron himself failed Goldman's financial-modeling test at first, learning to balance a balance sheet only at Bear Stearns from a second-year mentor. Today that story supports a tilt toward intellectual journey and well-roundedness over memorized formulas. As toil — repetitive low-value work — automates, his distinction helps: the best learning was not building the model but reasoning with a senior over a 27-page meeting set about what the client asked and what to do in the room.

The most human moment is personal balance and leadership. Waldron says a child's "you're here but you're not really here" hit like a dagger; constant travel, managing weekdays as they come and trying to be truly home on weekends. It mirrors a tension many with heavy responsibility know. For leadership he cites Dwight Eisenhower: leadership is the art of getting someone to do what you want because they want to do it. Much of Waldron's day, he says, is that EQ practice — reading intent and emotion to build cooperation. Joe Baratta closes the circle with 25 years of Waldron watching — annoyingly good at almost every sport. As global head of private equity, Baratta reads the IPO window as open for category-defining, diversified, market-leading businesses; against an S&P concentrated in one theme, large examples like Medline or Jersey Mike's and soon Copeland meet demand for exposure to something else. Public markets remain a vital liquidity source for private equity. On talent, his tone is close to Waldron's but drier: "Does it change much? Not really." He recalls being an analyst at Morgan Stanley 35 years ago, photocopying and punching holes from midnight to 3 a.m., versus today where analysts contribute to the investment debate itself. Tools change; good questions and context endure.