NextBigFuture analysis: Hyperscaler AI capex appearing on depreciation schedules as 'sunk cost,' transforming 2026-27 margin depression into 2028+ profit.
The most misunderstood line item in the AI buildout is depreciation. Every hyperscaler is reporting compressed margins as tens of billions of dollars of AI capex hit the income statement through depreciation and amortization while revenue is still ramping. Amazon's accounting illustrates the dynamic: roughly $220 billion of AWS-attributable 2027 capex, with 70% in short-life assets (AI servers on approximately five-year lives, general servers on six), produces about $30.6 billion of annual depreciation that lands in full from day one, while incremental revenue takes about three years to reach 100% utilization. Year one looks breakeven-to-negative even when the underlying investment is excellent.
SpaceXAI—the entity formed by the February 2026 SpaceX–xAI merger, public on Nasdaq as SPCX since its record $85 billion IPO on June 12—is running a different version of this model, and the difference is the whole story.
**Wholesale Leasing Collapses the Revenue Lag**
AWS sells cloud services at retail through a multi-year enterprise adoption curve. SpaceXAI signs wholesale GPU leases that pay full freight almost immediately. Anthropic has committed to $1.25 billion per month (roughly $15 billion per year) for the full capacity of Colossus 1—approximately 230,000 GPUs including H100, H200, and early Blackwell, and 300+ megawatts—running through May 2029 with 90-day mutual termination rights. Google is paying $920 million per month (roughly $11 billion per year) for approximately 110,000 newer-generation GPUs, ramping at reduced fees from June through September 2026 and hitting the full rate on October 1, 2026, through June 2029. Google can terminate if committed GPUs aren't delivered by September 30, 2026; after December 31, 2026, either party can exit on 90 days' notice. Google has described the deal as bridge capacity for Gemini Enterprise demand. Reflection AI committed to $150 million per month (roughly $1.8 billion per year) for GB300 capacity—the third disclosed deal and the clearest evidence that demand extends past the two anchor hyperscalers.
That is $27.8 billion per year of contracted run-rate against an estimated roughly $15 billion of annual depreciation on the assets in service. The buildings are built, the turbines are bought, the chips are racked. The major capex behind these contracts is already spent. Under straight-line accounting, the leases flip the same facilities that produced 2025's $6.4 billion operating loss (on $3.2 billion of revenue) into GAAP-operating-positive territory on signed paper alone, before any new deal.
**The Pricing Curve: Chip Generation Sets the Rate**
The three contracts reveal an observable price curve. Anthropic's Hopper-heavy Colossus 1 implies roughly $38–44 billion per gigawatt per year. Google's newer-mix block implies roughly $50–55 billion per gigawatt per year. One generation of chip improvement is worth approximately 30% more per GPU per month. New 2027 capacity will be B300 (Blackwell Ultra) and early Rubin, which supports a blended new-lease assumption near $46–52 billion per gigawatt per year if supply stays tight. The honest caveat is that per-gigawatt rates may plateau even as chips improve, because newer chips draw proportionally more power. What improves for the tenant is tokens per dollar, which supports the same rate per gigawatt rather than an ever-rising one.
**The Power Ramp: The Gating Physical Constraint**
Capacity currently stands at roughly 1.75 gigawatts on-site, with 6–7 mobile turbines (approximately 220 megawatts) added monthly through a locked supply chain—a 49.9%-owned Solaris joint venture, an APR Energy acquisition, 60–65% of Solar Turbines' Titan production, and twelve 380-megawatt Doosan units on order with the first permanent pair targeted for early 2027. That trajectory reaches roughly 4–4.3 gigawatts by March 2027, powering on the order of 830,000 deployed chips. The monthly cadence should be treated as a projection, not a disclosure—the S-1 gives quantities without dates.
The counterweight is legal, not logistical. The NAACP and environmental groups' lawsuit was amended in June after discovery found nearly 30 additional gas turbines at the Southaven site—57 total—with emissions claims under the Clean Air Act. An adverse ruling on turbine operation is the single most direct threat to the power ramp. However, no environmental rulings have required removal of turbines; usually fines and remediation follow. SpaceXAI holds various environmental waivers and permits from the states and a statement from the DOJ that operations are national security critical.
**Financial Projections and Verdict**
Assuming roughly 4 gigawatts of exit capacity with roughly half leased externally, a $46 billion per gigawatt base rate on new leases, internal inference revenue growing from a roughly $4 billion run-rate (Grok, Cursor), cash opex near 20% of revenue, and roughly $15 billion of depreciation and amortization, three readings matter. First, even the bear case is operating-profit positive, because the signed $27.8 billion run-rate alone covers total costs. Second, fiscal-year revenue is much lower than exit run-rate—Q4 2027 annualized revenue in the base case is $90–105 billion, and conflating the two is the most common error in bullish writeups. Third, company-level free cash flow is a different question entirely: continuing capex of $50–70 billion per year keeps free cash flow negative-to-breakeven except in the bull case. Proforma cashflow positive on already-built assets is true and meaningful for return on invested capital; it is not the same as the company generating cash. That timing arrives not in 2027 but in 2028 or later.
**Durability Beyond 2027**
The two anchor contracts carry 90-day termination rights after their initial periods, and both tenants are spending historic sums to stop needing rented compute. Alphabet raised $84.75 billion in equity in June explicitly for its own AI infrastructure, on top of $180–190 billion of 2026 capex. Today's $50 billion-plus per-gigawatt rates are scarcity pricing. The bet embedded in every number above the signed line is that scarcity persists through the 2027 signing window. Reflection is the encouraging counterpoint—at $1.8 billion per year it represents only 6% of the signed base, but it is the first tenant that isn't a hyperscaler building an exit ramp, and the AI-lab tier is where durable second-cycle demand would come from.
The capex-written-off-over-5-to-6-years-creates-positive-earnings thesis is validated for this asset base. The unresolved question is what the second lease cycle prices at. The turbine litigation is the second-order risk; depreciation lives that prove too generous for fast-obsoleting accelerators are the third. SpaceXAI looks bullish on 2027 operating economics—the contracted base makes the near-term case a matter of delivery, not forecasting—but cautious on rate durability past 2027, when the bridge contracts hit their exit windows in what should be a better-supplied market.