Is the AI Infrastructure Supercycle Broadening Beyond Hyperscalers?
Yes, and the evidence is in the order book. Nvidia’s Q2 FY27 revenue hit $96.2 billion, up 106% year-over-year, with data center revenue surging 117% to $89 billion. The company guided Q3 revenue to $108 billion and fiscal 2028 growth to 70%, nearly double consensus. Supply commitments more than doubled quarter-over-quarter to $279 billion, with $267 billion due by end of fiscal 2029, locking in component supply ahead of the Vera Rubin platform ramp. Vera Rubin is expected to contribute roughly 20% of data center revenue in the current quarter. Gross margin is guided to a 71-72% floor in fiscal 2028 as memory costs rise, then settling at 72-73%. At Nvidia’s revenue scale, each gross margin point equals roughly $1 billion per quarter.
The demand base is widening. AI labs are projected to provide 25% of Nvidia’s business next year, while neo-cloud capacity is surging to 8+ GW from 3 GW. Amazon plans 2 million additional GPUs for 2027-2028. Yotta Data Services is ordering 50,000 Vera Rubin GPUs plus 45,000 GB-300 chips in a $7.5 billion deal, India’s largest to date, for a 120 MW Delhi data center going live mid-2027. Google is investing $15 billion in a 1 GW AI data center hub in Visakhapatnam, operational by late 2028. Dan Ives estimates every $1 of Nvidia spend generates $8-10 of total ecosystem spend across storage, networking, cloud platforms, and software.
Physical AI is emerging as the next vector. Anthropic’s Model Hardware Standard connects LLMs to manufacturing equipment and scientific instruments in hours rather than weeks, with early partners including Genentech, Carnegie Mellon, Universal Robots, AWS, and Danaher. Nvidia’s reported $13 billion acquisition of Hugging Face would give it control of the largest open-weight model platform, steering millions of developers to Nvidia hardware. The headwind: OpenAI’s $50 billion compute spend this year and $500 billion Stargate commitment face 120+ state moratoriums blocking $156 billion in data center projects, while Texas ERCOT faces 474 GW of interconnection requests.
Has Enterprise Software Finally Proved It Can Monetize AI?
Yes. The SaaSpocalypse narrative is dead. Salesforce surged 22.6% to $252.05 after Q2 revenue rose 11% to $11.35 billion and Agentforce ARR jumped 240% to $1.5 billion, approaching 10% of the revenue guide. The Claudeforce partnership with Anthropic integrates Claude reasoning directly into Salesforce workflows with a mandatory trust layer. cRPO grew 14%, ahead of revenue growth, and Agentic Work Units surged 97% quarter-over-quarter. Management raised fiscal 2027 revenue guidance to 11-12% growth and hiked EPS guidance 28%.
The read-through is broad. ServiceNow rallied 10% on the Salesforce print, with its AI Control Tower for autonomous agents positioned as the platform-layer beneficiary. CrowdStrike jumped 20% after ARR grew 25% to $5.84 billion, with 51% of subscription customers now on six or more modules and FY27 revenue guided to $6 billion. AppLovin delivered 46% year-over-year revenue growth to $1.9 billion with a 78% operating margin, defying typical ad seasonality. Meta’s $18 billion teen safety settlement removes a major overhang; the stock trades at 16x forward earnings versus a 23x sector average, with EPS upside from Reality Labs cost cuts and excess AI capacity monetization.
Is Nuclear Power the Defining Long-Dated Optionality of the AI Era?
Yes, but the timeline is 2030-2040, not 2026-2028. Oklo’s 1.2 GW deal with Meta for an Ohio SMR facility targets first phase completion around 2030, but Oklo holds no NRC license yet and trades at $42.53, down 45% year-to-date. NuScale is down 41% with a 16% gross margin. Bank of America values the AI-driven nuclear opportunity at $10 trillion but sees meaningful SMR adoption only in 2035-2040. Microsoft’s 20-year PPA with Constellation to restart Three Mile Island is the only near-term contracted revenue signal. NTPC is targeting 30% of India’s 100 GW nuclear goal by 2047 as part of a ₹16.86 trillion capex plan through FY37. Valuations are repricing for pre-commercial developers; the optionality is real but distant.
Can India Convert Structural Tailwinds Into Cyclical Recovery?
Structurally yes, cyclically mixed. The RBI holds ₹3.75 trillion in surplus liquidity, with FCNR(B) inflows reaching $72.85 billion. India Ratings raised FY27 credit growth forecast to 15% and deposit growth to 13.6%, but the loan-to-deposit ratio sits at 85%, system RoA has fallen 6 basis points to 1.31%, and credit costs are rising to 74 bps. PSU banks face steeper margin and credit cost pressure than private peers. S&P affirmed BBB stable but flagged a 7.3% general government deficit and growth slowing to 6.6% in FY27.
Policy reforms are accelerating. The CCEA FDI approval threshold is moving from ₹5,000 crore to ₹15,000 crore. Japan pledged ¥10 trillion for semiconductors and supply chains, with Tokyo Electron’s Dholera fab and Tata Electronics’ Assam OSAT already underway. BIS certification easing for high-tech firms addresses a top compliance complaint from 72% of Japanese manufacturers. Defence production has quadrupled to ₹1.8 lakh crore with a ₹50,000 crore export target by 2028-29. Steel premiumization is structural: value-added mix sits at 55-60% across JSW, Tata, AM/NS, and Jindal, with capacity growth (23%) outpacing demand (20%) and HRC capacity up 29%. NTPC’s ₹16.86 trillion capex through FY37 targeting 244 GW provides multi-year visibility.
How Should Allocators Position Around the US Fiscal and Trade Collision?
Own the collision, don’t ignore it. The US national debt has passed $40 trillion. CBO projects net interest costs exceeding $1 trillion in 2026 and reaching $2.1 trillion by 2036. The Social Security OASI trust fund depletes in Q4 2032, triggering an automatic 22% benefit cut without reform. The 30-year Treasury yield has breached 5.2%, the highest since 2007, driven by fiscal premiums, AI-driven corporate issuance (11% of 2026 IG supply), and sticky inflation. Treasury Secretary Bessent doubled long-bond buybacks to $4 billion per operation while Fed Chair Warsh maintains a hawkish stance at Jackson Hole, a policy divergence that historically pressures the dollar.
The dollar bear positioning is extreme: UDN call volume surged 1,144% with a 158:0 call-to-put ratio. The US-Canada trade war escalated to 50% tariffs on August 22, with auto tariffs rising to 50% on January 1, 2027. Vermont sends 31% of goods exports to Canada; Michigan has 1.2 million jobs at risk. Allocators should overweight gold and mining royalty companies (operating leverage on gold), unhedged foreign equities, and international ETFs as dollar-weakness hedges, while favoring intermediate-duration Treasuries (IEI’s 3-7 year profile returned 3.73% over three years) over extended duration (EDV down 12.67% over five years).
Is Space Commercialization Generating Real Revenue or Just Narrative?
Revenue is real, but valuations price in perfection. SpaceX hit $1.9 trillion market cap at $140.87 with Q2 revenue up 92% year-over-year to $7.8 billion. Starlink generated $4.29 billion, AI revenue reached $2.56 billion, and a Google deal pays $920 million monthly for 110,000 Nvidia GPUs through June 2029. But Q2 capex hit $18.4 billion, the stock trades at 64x price-to-sales, and the economic moat in orbiting data centers remains speculative. Rocket Lab delivered $234 million Q2 revenue (+62%) with a $2.36 billion backlog (+137%), but Neutron launch is delayed to Q4 2026 or 2027 and adjusted EBITDA turns positive only after first Neutron flight.
Firefly Aerospace is the preferred defense/space pick at 12.9x P/S with a $1.4 billion backlog and NASA lunar landing credibility, versus AST SpaceMobile at 149x P/S. GE Aerospace generates $7.3 billion free cash flow on a $210 billion backlog with 19% revenue growth, while Archer trades at 626x P/S pre-revenue. The allocation framework: own GE for cash flow and backlog durability, size Rocket Lab as a high-beta launch/services play, and treat SpaceX as a venture-style position until competitive moats clarify.
Are Crypto ETF Flows Confirming a Durable Institutional Bid?
Yes, but the composition matters. Bitcoin near $80,000 with $1.9 billion weekly spot ETF inflows and IBIT at $60 billion AUM capturing 62% of sector flows confirms institutional adoption. Strategy holds 840,447 BTC at a $75,385 cost basis but paused ATM purchases. Solana surged 23% to $107 on $1.2 billion cumulative ETF inflows, $16 billion in stablecoins, and $56 billion in 30-day DEX volume, though memecoin-driven volume on Pump.fun raises sustainability questions. MARA reported 27% year-over-year revenue contraction with margin compression and insider selling ahead of the rally. The takeaway: pure-play spot ETFs are the cleanest exposure; leveraged corporate balance sheets and operationally strained miners carry idiosyncratic risks that ETF flows don’t solve.
FAQ
Q: Is Nvidia’s 70% FY28 growth guide credible? A: Yes. The $279 billion in supply commitments, Vera Rubin ramping to 20% of data center revenue this quarter, AWS’s 2 million GPU commitment, and neo-cloud capacity tripling to 8+ GW provide multi-quarter visibility. The risk is China exclusion and moratorium-driven delays, not demand.
Q: Does Salesforce’s Agentforce ARR prove enterprise AI monetization at scale? A: At $1.5 billion ARR growing 240%, approaching 10% of revenue guide, with cRPO growing faster than revenue and Agentic Work Units up 97%, the proof is in the order book. The Claudeforce partnership adds a trust-layered distribution channel.
Q: When does nuclear power become a material revenue contributor for AI data centers? A: Not before 2030. Oklo-Meta targets first phase ~2030; Microsoft-Constellation TMI restart is the only near-term PPA. Bank of America sees meaningful SMR adoption 2035-2040. Position as long-dated optionality, not near-term cash flow.
Q: How should India exposure be structured given RBI liquidity surplus but rising credit costs? A: Overweight private banks with diversified funding and lower credit cost trajectories; underweight PSU banks facing steeper NIM compression. Own defence and steel premiumization themes (JSW, Tata, Jindal) for structural growth. Size NTPC for capex visibility but monitor margin pressure from ₹16.86 trillion spend.