Where capital is heading: crypto’s slump, Chamath’s LPS bet, and the data-center land grab
Capital is moving in two directions: back into high-volatility crypto for asymmetric upside, or into physical bets that support the AI boom, such as long leases on land, reliable power, and hardened buildings. That tug of war matters more for business leaders and CFOs than today’s market headlines.
Liquidity isn’t the story, selectivity is
Changpeng Zhao (CZ) has argued the crypto bear market reflects investor caution, not a lack of capital. Liquidity is available, but investors are choosy about where they place it. At the same time, some high-profile allocators are shifting capital toward what’s called “LPS”, Land, Power, and Shell, the physical components of AI-ready data centers.
“LPS”, Land, Power, and Shell.
The headline benchmark: TeraWulf’s Anthropic lease
TeraWulf’s investor filing explains why LPS draws attention. The company announced a 20-year lease with Anthropic for a purpose-built campus in Hawesville, Kentucky, planned for roughly 401 megawatts of critical IT load. TeraWulf describes the contract as approximately $19 billion of contracted lease revenue over the initial 20-year term, with staged capacity expected to begin service in the second half of 2027 and to ramp to 401 MW by early 2028 (per TeraWulf’s press release).
The arithmetic is simple but can be misleading. $19 billion over 20 years averages about $950 million a year, but TeraWulf’s release and normal accounting practice make clear revenue recognition will follow phased builds, occupancy milestones and the contract’s specific terms. The company also lists forward-looking risks tied to construction, permitting, power availability and counterparty performance.
Why investors see LPS as different
Three realities make LPS appealing to some allocators right now:
- Scale and time sensitivity. Large AI models and dense inference clusters need multi-hundred-megawatt campuses on tight timelines, and long leases with creditworthy tenants can look like durable cash flows if projects finish on schedule.
- Supply frictions. Industry trackers report rising permitting fights and legislative pressure around data-center buildouts, which can delay new supply and create a scarcity premium for sites that are already approved or near grid interconnection.
- Hardware commercialization risk. Chip startups face long R&D cycles, manufacturing constraints and consolidation risk. Forbes reported Groq, founded in 2016 and an early Social Capital investment, later reached a large licensing and hiring arrangement with Nvidia, showing the consolidation pressure in chips.
Permitting and politics are real value drivers
Data Center Watch reported at least 75 U.S. data-center projects were blocked or delayed in early 2026, representing roughly $130 billion of proposed investment. The group also tracked more than 300 state-level data-center bills introduced in a compressed period, and documented widespread community opposition across many states. Local political resistance, from zoning fights to environmental reviews, can turn a build-ready parcel into a long, uncertain timeline. That friction is precisely what can give LPS assets a premium, if those numbers hold up under follow-on verification.
Risks: don’t mistake a headline contract for guaranteed returns
Big headline deals validate demand, but they also carry concentration and contingency risks. The TeraWulf, Anthropic lease is a useful benchmark because it signals demand and scale, but the $19 billion is a contractual total contingent on project completion, tenant occupancy and the tenant’s credit profile.
Analysts add caution. Jordi Visser of 22V Research expects returns on AI infrastructure to normalize as the market matures. Normalization can happen if permitting reforms speed up, grid capacity expands, more entrants reduce scarcity, or markets more accurately price construction and counterparty risk.
Risk matrix, the four valuation levers to model
- Construction risk, cost overruns and schedule slips change IRR materially. Treat build timelines as probabilistic, not deterministic.
- Permitting & regulatory risk, delays of 12 to 36 months are realistic in contested jurisdictions and should be scenario-tested.
- Power & energy cost risk, AI campuses use a lot of power. Model plus or minus 30 percent swings in input costs and consider the economics of on-site generation or long-term power purchase agreements (PPAs).
- Counterparty/tenant credit risk, concentration on one or two tenants raises downside. Require parent guarantees, investment-grade support, or replacement-tenant clauses where possible.
Where capital appears to be flowing (for now)
Some investors are actively securing the physical inputs for AI campuses. Multiple reports say Chamath Palihapitiya and associates have shifted capital toward infrastructure plays and are procuring power capacity and sites, reflecting a bet on scarcity in LPS rather than on early-stage chips. At the same time, Binance’s CZ frames the picture differently: capital exists in markets, but investors are pickier about risk and reward.
Those two paths can coexist: patient capital buying durable, power-anchored assets and capital chasing crypto upside. The central question for markets and portfolio managers is whether the scarcity premium built into LPS pricing will last long enough to deliver higher returns, or whether supply and policy changes will compress those premiums.
Practical guidance for CFOs and capital allocators
If you’re sizing exposure or underwriting a deal, translate the risk matrix into explicit scenarios and documentation tests:
- Scenario modeling: run IRR and NPV sensitivity to permitting delays of 0 / 12 / 24 / 36 months, and to power-price shocks of ±30%.
- Tenant stress tests: model tenant downgrade or exit scenarios with and without parent guarantees; require minimum investment-grade support or escrowed payments where possible.
- Construction safeguards: require turnkey guarantees, fixed-price EPC schedules or clear completion bonds and milestone-linked draws to protect against overruns.
- Exit planning: assume returns will compress over time; set realistic hold periods and target IRR ranges that reflect normalization risk.
- Policy monitoring: track state legislative activity and local permitting timelines closely. These are now primary value drivers for LPS assets.
Key questions, and short, honest answers
- Is the $19 billion from the TeraWulf, Anthropic deal guaranteed cash?
Per TeraWulf’s press release, the $19 billion is contracted lease revenue over the initial 20‑year term; it is forward‑looking and contingent on phased buildout, occupancy and the tenant’s credit, and TeraWulf identifies construction, permitting and power risks in the same filing.
- Are data centers actually being blocked or delayed at scale?
Industry tracker Data Center Watch reported dozens of blocked or delayed projects and hundreds of state bills in early 2026, which industry participants cite as evidence that permitting frictions are materially impacting supply timelines.
- Why would an investor prefer LPS over betting on AI chips?
Chip startups demand long R&D cycles, manufacturing access and face consolidation risk, Forbes reported Groq’s commercialization struggles and a subsequent large arrangement with Nvidia, while LPS targets site and power scarcity and long‑duration lease cashflows that some investors view as lower‑tail‑risk in the near term.
- Will AI infrastructure returns stay elevated?
Analysts such as Jordi Visser at 22V Research expect returns to normalize as the market matures; elevated returns can persist while permitting friction and grid constraints endure, but those factors can and likely will change over time.
A practical next step
If you manage allocation or underwrite LPS deals, don’t rely on narrative alone. Get the contract, read the filing or press release yourself, and run at least three scenarios, optimistic, base, and conservative, that stress permitting delays, tenant credit and energy-price swings. Ask your deals team for documented guarantees and completion protections before you treat headline contract totals as cash in the bank.
Both narratives can be true at once: patient capital buying durable, power-anchored real assets and capital chasing crypto’s asymmetric upside. Good allocation is less about picking a single winner and more about sizing exposure, pricing visible risks and watching the policy and grid developments that will decide whether LPS stays a scarcity premium or becomes another crowded real-estate sector.