Rose Celine Investments
Equity Investor
AI infrastructure providers' heavy capital needs and aging equipment may leave little cash for shareholders despite revenue growth. Depreciation is among their most underappreciated costs: shorter equipment lives raise it, while replacement and financing erode returns. Leverage and dilution magnify economic risks. Some may prosper, but until pricing power, depreciation, financing and long-term returns on capital are clearer, their investment appeal remains hard to share. x.com x.comÂ
Nebius, IREN and CoreWeave profit from AI compute scarcity, which is not a moat, and their margins should compress once supply catches up. Within about five years compute will likely be a commodity across hyperscalers and neoclouds, with customers shopping on price and performance. Neoclouds mostly rent Nvidia GPUs rather than owning differentiated chips, so if Google or Amazon undercuts them with TPU or Trainium compute, they must cut prices or lose customers, as in DRAM, fiber and telecom. x.com x.com x.com x.comÂ
AI's next winners may capture the most value with the least incremental shareholder capital. Equipment suppliers get paid while others largely fund AI factories, though they face competition, cyclicality, concentration and valuation risk. Cash-rich builders can fund much of their own capex, but still need adequate ROIC. Capital-dependent builders need outside financing; leverage could yield enormous returns if ROIC stays well above its cost, but works against them if returns fall below it. x.comÂ
Hardware suppliers and VC keep financing cash-hungry labs and neoclouds, making compute rental prices at CoreWeave, Nebius and IREN look artificially supported. Lab revenue growth is finally plateauing at 800% year over year after roughly tenfold equity gains across the stack over the last decade. The buildout is also crowding out the economy's appetite for debt. x.com x.com x.com x.com x.com x.comÂ
Jan Zuazo Pfingsten
Thematic Investor, Physical Scarcity & AI Infrastructure, Primary-Source Research
AI infrastructure suppliers still have healthy demand despite tense financing. Q4's spending pace contradicts the reading that flat capex means a slowdown, while Cloud growth is 103%. Alphabet's negative free cash flow and share issuance are what punish its stock; a supplier selloff tomorrow would be an opportunity, not an exit signal. x.comÂ
Rising debt costs squeeze Nebius's capital-intensive AI compute rental model. Elice's Korean IPO collapse highlights the risk for companies that raise money to buy chips and rent them out when money stops being cheap. Prefer operators with signed cash-rich tech customers or power they own outright; these companies are not all alike. x.comÂ
Alphabet's surging AI capex is outpacing its payoff. Its first double earnings miss in four years came with free cash flow of -$9.1 billion versus -$2.7 billion expected, long-term debt up 479% in a year to $125.1 billion, and guidance for significantly higher 2027 capex above $300 billion. x.com x.comÂ
Nebius' software efficiency and a possible end to heavy AI financing support a re-rating. Per @DrTomsLens, its engineers got about 43% more video output from the same GPUs, cutting compute cost per second by 30%; newer hardware alone is not enough without capable software teams. Oracle expects to fund growth from free cash flow in FY2029, suggesting capital-heavy AI providers may eventually outgrow debt and dilution. Google Cloud reported 103% revenue growth amid strong AI infrastructure demand. … Show more
Alphabet's AI capex payback is being repriced, making its post-earnings decline more than a one-day mood swing. Guidance and cloud and AI margin commentary are the near-term tests. If growth continues but the multiple compresses, the risk is duration in mega-cap AI, not a collapse in technology growth. x.comÂ
AI suppliers' demand may not justify today's investment. Roughly $3 trillion in revenue absent last year is needed to support the spending, yet GDP growth is weak and many companies may be seeing only incremental gains. If the estimate that Anthropic uses two-thirds of its compute for training is right, it could triple inference supply. The most likely outcome for many companies, roughly half the S&P 500, is slowing growth, compressed margins and multiples; when the bubble ends is uncertain. … Show more
Lenders' growing data-center exposure could make funding costlier and harder for powered-land developers despite a bullish outlook for the AI build-out. Financing concerns can also depress equity issuance prices and increase cash-reserve needs. Customer prepayments, construction contributions and payment backing could help; if lending remains tough, dollars per megawatt may rise, perhaps faster. The most attractive investments pair customer funding with enough equity value and company liquidity. … Show more
Some AI data-center developers may have to absorb higher financing costs because existing leases have fixed economics and tenants need not pay more. That could reduce equity returns even as compute demand grows. Scarce deliverable power, strong counterparties and favorable financing could instead make the best developers more valuable as marginal projects stall. Vertically integrated infrastructure landowners such as Cipher look attractive; today's prices may be the worst of the selloff. x.com … Show more
