Tech • AI • Robotics • Game

ENFR

🤠 LIVE from TX with TK, Jon Gray & Scott Wu

8/10
AITBPNNovember 18, 2026 at 05:20 PM1:25:51
Audio player

⏱ ≈ 3 min read · 1 h 26 min video — 1 h 23 min saved

TL;DR

Blackstone, Micron and Morgan Stanley executives said the AI buildout is shifting from software hype to capital-intensive infrastructure, with demand for compute, memory, power and deployment outpacing supply.

KEY POINTS

AI investment is entering a physical buildout phase

Blackstone President John Gray said the current AI cycle looks less like the internet boom and more like the late 19th-century infrastructure era because it requires vast spending on data centers, chips, energy and industrial capacity. He said the defining feature is not just innovation in software but the need for trillions of dollars of real-world investment linking technology, finance and heavy infrastructure.

Compute and energy shortages are driving deal flow

Gray said the market is facing a profound shortage of both compute and power, pulling private capital deeper into AI infrastructure. He described demand as global and widening beyond model training into physical AI, while warning that constraints around electricity, chip supply, permits and capital aggregation are limiting how fast new capacity can be built.

This boom differs from railroads and 1990s fiber

Gray argued that earlier infrastructure booms were often built speculatively, with supply created before demand was proven. In contrast, much of today’s AI infrastructure is backed by long-term contracts from cash-rich hyperscalers with average market values around $2.5 trillion, making the risk profile different from past bubbles or from overbuilt real estate markets.

Enterprise AI demand is already showing measurable returns

Gray said spending on models at Blackstone portfolio companies has risen 20-fold in the past year, driven by productivity gains. He cited examples including rental applications processed 90% faster and Ancestry.com family-history processing running 30 times faster, adding that many companies are still underusing AI rather than overusing it.

Biotech may be one of AI’s biggest long-term winners

Gray said AI is expanding the front end of drug discovery through synthetic chemistry and biology, potentially accelerating therapies in cancer, neurological disease and pulmonary medicine. But he stressed that the bottleneck remains physical clinical trials, where late-stage studies can cost $1 billion or more and still require regulatory review, human enrollment and global execution.

Micron is spending heavily to relieve memory constraints

Micron President and CTO Scott DeBoer said the company is expanding manufacturing globally and plans more than $250 billion of investment, with major projects under way in Boise, Idaho, Clay, New York, and Virginia. He described some of the projects as among the largest industrial builds in the United States, aimed at easing multiyear tightness in memory supply.

Memory has become central to AI economics

DeBoer said the AI era has shifted competition from raw bit growth alone toward performance, customization and co-design with customers. He said memory is now a critical determinant of token economics, making Micron more strategically embedded with major customers than in the past, when memory was treated more as a commodity.

Micron says AI is speeding chip design internally

DeBoer said AI is already pervasive across product development, with a goal of completing designs in half the time and producing twice as many designs with the same number of people. He said the company is using AI not only for coding and documentation but also for design verification, process simulation, thermal modeling and chip physics.

US manufacturing is being framed as a customer and supply-chain strategy

DeBoer said Micron’s US expansion is not just patriotic branding but a response to the concentration of customers in the United States and the importance of domestic supply. He also emphasized local partnerships, citing investments in roads, housing, schools and community institutions to secure support for large-scale industrial projects and the long-term jobs that follow.

Morgan Stanley highlighted its reach into private and public equity ecosystems

Morgan Stanley wealth chief Jed Finn said the firm oversees about $8.5 trillion in assets across roughly 21 million clients. He added that through Solium, acquired in 2019, the bank now manages cap tables for 9 of the top 10 unicorns, 75% of the top 100 by market value and about half of nearly 1,000 private companies in its target universe.

CONCLUSION

Executives across finance and semiconductor manufacturing are converging on the same message: AI is no longer just a software story, but a contest to finance, build and operate scarce physical capacity fast enough to match surging demand. The next phase will be shaped as much by power grids, fabs, contracts and clinical trials as by model breakthroughs.

Ask a question

More from AI