Connor Teskey on Brookfield, Infrastructure, and the Business of Real Assets

Connor Teskey — CEO of Brookfield Asset Management, which manages about a trillion dollars across infrastructure, real estate, private equity and credit — on how the firm engineers long-duration cash flows out of the backbone of the global economy, why the AI data-centre build-out is its fastest-growing theme, and why he spends most of every deal meeting on the downside.

Key ideas

1. Brookfield sells de-risked cash flows, not asset bets. The firm is comfortable taking construction, operating and development risk, but works hard to strip out market risk. On a new solar farm it locks the CAPEX, the off-take (power-purchase agreement), the construction (EPC) contract and the financing all at once — so a swing in rates, power prices or inflation cannot touch the return. The same template now runs on data centres built for hyperscaler off-takes.
2. Underwrite the worst case and the base case takes care of itself. The vast majority of any deal discussion is spent on downside. Buy high-quality assets with strong downside protection, underwrite the worst case rigorously, and the expected case ends up attractive — with asymmetric upside you did not need to pay for. Westinghouse is the case study: a market-leading nuclear supplier bought when the sector was out of favour, safe on the downside, then handed a nuclear revival as free upside.
3. The AI theme is played three ways, and Brookfield picks the middle one. Not investing in the models (ChatGPT, Anthropic — ‘not our area of expertise’), but building the infrastructure — data centres and the power that feeds them — the largest and fastest-growing investment theme at the firm, fusing its digital-infrastructure, power and real-estate expertise. The third way is using AI inside its 500 portfolio companies, where preventative maintenance and health-and-safety are the quiet wins across a trillion dollars of real assets.
4. Overbuild is certain; ruinous overbuild is a choice. There will unequivocally be overbuild in AI infrastructure, as in every asset class in every cycle. The defence is to build only against long-term contracts with the highest-credit-quality counterparties — never on spec — and only in tier-one markets with multiple end users, so a facility can be re-contracted at the end of a 20-year life.
5. Local autonomy, central capital allocation. Boots-on-the-ground teams in 60-plus markets source, execute and operate with real independence; every capital-deployment decision funnels up to one tight committee. That structure buys local edge and the global perspective to steer capital to the best risk-adjusted return anywhere — an opportunity that is the best a region has seen in two years can still lose to a better one elsewhere.

Summary

The business model — the backbone of the global economy

Brookfield raises capital from the world’s largest pools of money and deploys it into critical assets and services that drive the productivity of the communities they sit in. The consistency is the point: the firm has held to high-quality real assets for decades, even as the contents of that category turn over — hydro dams gave way to solar, nuclear and batteries; ports and railroads now sit beside data centres, fibre and telecom towers, and roughly two-thirds of what Brookfield owns today was not an investable asset class 20 years ago. The packaging has multiplied where the discipline has not: four products a decade ago, sixty today, spanning flagship, mezzanine-debt, super-core and retail-wealth strategies across infrastructure, real estate and private equity.

Engineering out market risk

The craft is turning the construction or operation of a project into a long-term, inflation-linked stream of cash flows. Teskey is candid about which risks the firm will and will not hold: execution, operating and development risk, yes; market risk, no. The mechanism is contractual — lock the four drivers of return (CAPEX, off-take, EPC, financing) simultaneously, and outcomes stop depending on macro moves. Financing follows the same logic: asset-level, non-recourse, long-duration, fixed-rate debt, deliberately harder to arrange but ring-fencing each asset so one bad outcome cannot taint a portfolio — and one great bid is never trapped inside a shared facility. Underpinning it all is a conviction that liquidity is chronically undervalued exactly when it is most needed, so the firm keeps excess capital for the unforeseen, upside or down.

The AI data-centre and power build-out

Data centres have gone from funding ‘the rack and the shell’ to funding the chips, servers, power supply, grid connection, substation and redundancy — the whole facility plus its energy supply chain — still wrapped in a long-term, take-or-pay, inflation-linked off-take with a hyperscaler or a sovereign. More centres, bigger centres, and a wider funding scope are compounding the opportunity at once. The counterparty quality is, in Teskey’s reading, as good as it has ever been — the large technology companies are the highest-credit-quality names in the world. The two most common reasons Brookfield walks from a deal remain the same across every vertical: a revenue construct (or the credit behind it) it does not trust, or construction risk out of proportion to the return.

Culture, talent and capital allocation

Brookfield spent its first century as a direct owner-operator, not an asset manager, and that inheritance shows: it likes businesses it would happily own outright, holds for the long term, and books part of every return from operational improvement — health-and-safety standards are the non-negotiable first move on any acquisition. Its balance sheet is still the largest investor in its own products. On people, Teskey describes a meritocracy with no central-casting stereotype: intellectually curious ‘nerdy’ problem-solvers who also work well in teams, young talent handed responsibility early, and a football-team model of specialists supported by generalists. Looking out, the growth thesis rests on institutional allocations to alternatives roughly doubling over a decade, and on a still-larger, near-unpenetrated frontier — the individual investor: retail, high-net-worth, annuity and 401(k) money that has historically had almost no access to the asset class.

Speakers

• Connor Teskey — President and CEO of Brookfield Asset Management; built and led Brookfield’s renewable-power and transition business before rising to run the roughly $1 trillion manager.
• Shane Parrish — founder of Farnam Street; host of The Knowledge Project.

See also

• Shane Parrish — host
• Value Investing — the downside-first, own-and-hold discipline Teskey applies to real assets at scale
• Private Credit — the migration of lending off bank balance sheets that underlies Brookfield’s credit and Oaktree franchise
• Gavin Baker on AI Infrastructure, Power Constraints, and Semiconductor Investing — the AI capex build-out seen from the semiconductor and power-constraint side
• Nicolai Tangen on Managing $2 Trillion, AI Bubbles, and Contrarian Investing — a fellow steward of vast capital on the AI-bubble question, from the same show
• Sebastian Mallaby on OpenAI's Cash Crunch, the AI Bubble Debate, and the China AI Race — the demand-side counterpart to Brookfield’s supply-side data-centre bet
• Dan Loeb on Activist Investing, AI, and Third Point's Credit Playbook — another investor weighing AI infrastructure against durable returns