"It was Jensen Huang who came to us!" Wall Street’s Six Giants Join Forces with NVIDIA to Unlock $500 Billion: Compute Power Becomes, for the First Time, a Collateralizable Asset...

Jon Gray: First of all, I have to say—this is brilliantly executed. Our capital markets are enormous—and that itself is one of America’s most critical advantages. Just think: every year, we provide roughly $700 billion in financing for the auto market, and trillions of dollars for the housing market. I believe AI compute will ultimately evolve into a similarly massive market—and the fundamental driver behind this is supply and demand. Take our portfolio companies as an example: over the past six months, we’ve seen their demand for large language models increase roughly sevenfold. Yet compute supply has not kept pace. Data centers are insufficient, power is insufficient, and chips are insufficient. So what will happen next is that investors begin to realize compute itself is a financeable asset. For instance: When you buy a house, banks certainly assess your creditworthiness—but they also appraise the property’s intrinsic value. When airlines purchase aircraft, lenders evaluate the airline’s credit—but also the aircraft’s own market value. One key constraint on AI infrastructure financing in the past has been that many investors would say, “My risk exposure to any single hyperscale cloud provider is capped at X,” or “My exposure to any foundational model company is limited to Y.” But if people truly come to recognize that—regardless of who ultimately uses these assets—the compute itself holds strong, inherent value, that changes everything. Especially NVIDIA. Its GPUs and CUDA software possess exceptional versatility and broad applicability. I believe capital markets will eventually recognize this opportunity. Of course, if the overall market size expands rapidly, financing spreads may widen. But long-term, once investors fully grasp both the supply-demand gap and the intrinsic value of compute, ever-increasing capital will flow in. Bruce Flatt: Let me add one point. Infrastructure has always been Brookfield’s core business—from the company’s earliest days, we’ve built infrastructure across the board. Initially, it was mainly energy: hydro, solar, wind, natural gas—and later, data centers. Now, through our collaboration with Jensen Huang and NVIDIA, we’re moving to the next layer—not only financing such projects but directly building compute infrastructure. And what we see today is crystal clear: no matter how much power we build, it’s never enough; no matter how much compute we build, it’s still insufficient—simply unable to keep up with the explosive demand Jon just described. So the real question isn’t whether this industry is over-financed. The true issue is that we’re simply not building fast enough. Four: Structuring Capital Inflows Appropriately Becky Quick: But ultimately, the question always circles back to one thing: can this demand sustain itself? Jensen, you likely see further ahead than anyone here. Is it possible that, yes, supply currently lags demand—but that demand could shift over time? Bruce Flatt: Jensen is certainly best positioned to answer that. But let me share what we’re observing within our industrial portfolio companies: we’ve only just begun deploying AI.

AI Data Centers, Energy, and Investment: The Industrial Revolution of the Intelligent Era

Amid the deep entanglement of AI and energy, the investment value of 'green compute' is being rediscovered and repriced by capital markets. Han Xiaoyan observes that as AI large models accelerate their iteration cycles, willingness and demand for training investment surge—driving simultaneous growth in both compute usage and density. Global tech giants are ramping up investments in power infrastructure, heralding the dawn of a high-visibility investment era for compute infrastructure. China is forging a uniquely distinctive practical path. In Ulanqab, Inner Mongolia, the world’s first integrated wind-solar-storage-compute base has entered operation: data centers connect directly to wind and photovoltaic arrays, significantly boosting local renewable energy absorption. In Guizhou, leveraging abundant hydropower resources, the national ‘East Data, West Computing’ hub leverages natural cooling sources alongside liquid-cooling technology—markedly improving energy efficiency. Schneider Electric forecasts that by 2026, liquid cooling will transition from a niche option to the mainstream cooling solution for data centers. When rack-level power density climbs to 240 kW, traditional air cooling becomes untenable—making liquid cooling a hard requirement to meet AI compute demands. Four: Financialization—the ‘Triple Play’ from Real Estate to Capital Markets Meanwhile, AI infrastructure is undergoing unprecedented financialization at breakneck speed. A once-niche data center real estate sector has evolved into one of the largest capital mobilization arenas in modern history—now attracting private equity, sovereign wealth funds, pension capital, and structured credit markets. Its economic model resembles a hybrid of cloud computing, energy infrastructure, and financial speculation. Global top-tier capital firms—including Blackstone, Apollo, Brookfield, and KKR—are actively deploying in AI-related infrastructure. Debt markets are exceptionally active, with developers fiercely competing to raise billions for mega-projects. A new financing ecosystem is emerging—centered around GPU procurement, long-term power purchase agreements (PPAs), and lease contracts with hyperscalers like Microsoft and Oracle. Larry Fink, BlackRock’s founder, declares: “Mobilizing private capital to build AI infrastructure—such as data centers—will unlock trillions of dollars in long-term investment opportunities.” This wave of financialization is shifting AI from a domain exclusive to tech companies to the very center stage of global capital markets. In China, Han Xiaoyan, Head of ESG Investment Research at Harvest Fund, notes that the impetus for low-carbon transformation in data centers is shifting—from policy compliance and corporate social responsibility toward more fundamental drivers: cost reduction, operational efficiency, and long-term value creation. Given that electricity accounts for roughly 70% of total data center operating costs, even marginal improvements in energy efficiency translate directly into substantial profit differentials. Capital is now targeting three precise strategic axes: enhancing energy efficiency on the consumption side, replacing grid power with green electricity on the supply side, and optimizing site selection based on environmental factors. Yet beneath the frenzy, clouds of uncertainty are gathering. This infrastructure boom rests upon a core assumption: that AI’s scaling laws will remain valid—that more chips, more data, and more energy will consistently yield smarter systems and greater economic returns. But this assumption now faces increasingly serious challenges. The emergence of DeepSeek sent shockwaves across the industry precisely because it suggests high-performance AI models might be trained at computational costs far lower than Silicon Valley anticipated.

Brookfield’s Teskey Sees No Slowdown in AI Infrastructure Build

Oct 13, 2026

Brookfield Asset Management CEO Connor Teskey forecasts sustained demand for AI compute and energy infrastructure through the decade, highlighting strategic partnerships with OpenAI, Bloom Energy, and an Nvidia consortium totaling over $500 billion in financing.

bloomberg

Brookfield’s Teskey Sees No Slowdown in AI Infrastructure Build

FAQ