When Jim Chanos calls something a “hopes and dreams IPO,” institutional investors tend to write it
down. At Global Alts New York 2026, the founder of Kynikos Associates and the short seller who
made his name calling Enron walked through what he sees as the cleanest setup for a market
dislocation since 2000 — and a lot of it has nothing to do with AI itself. It has to do with what investors
are willing to pay for the infrastructure around it.

The peg for the conversation was SpaceX, which was pricing its IPO at roughly $75 billion in proceeds
on a $2 trillion valuation against $19 billion in revenues and negative free cash flow. Chanos’s framing
was characteristically dry: the Starlink core business, by his math, might support “a couple hundred
billion dollars.” Everything above that — Mars colonies, factories on the moon, data centers in orbit —
is being capitalized today on the assumption it gets built tomorrow.

The “TAM Is Infinite” Problem

Chanos’s deeper point wasn’t about SpaceX specifically. It was about what the market is currently
willing to underwrite. “The TAM for space is infinite,” he said. “You can build whatever story you want to
justify the valuation.”

He drew a direct line to Tesla’s CEO-premium model, which trades at roughly 14x revenues on similar
promise-based narratives. “If it was trading as a car company, it would be at $30 to $40 a share, not
$400.” SpaceX, by contrast, priced at roughly 90x revenues. “A completely different animal.”

His one-line summary of the regime: “In bull markets you put a premium on promises. In bear markets
you put a discount on reality. Right now we’re clearly in the former.”

For allocators thinking about hedged equity, long-short, and short selling strategies as portfolio
constructs going into 2027, that’s the operative sentence. The dispersion between fundamentals and
price is wide enough to matter again.

The Short Selling Case Against Data Centers

Chanos has been bearish on the data center build-out since 2022, and his thesis has only sharpened.
Established operators, he noted, generate mid-to-low single-digit pretax returns on capital. “A really
bad business.”

The newer “neo-cloud” cohort — CoreWeave, Nebulas, and similar — he characterizes more bluntly:
equipment leasing dressed up as growth. “They buy a chip from Nvidia, lease it to a hyperscaler or an
AI company like Anthropic or OpenAI. The bet is on depreciation.”

The structural argument is the one allocators should sit with: “Anybody that’s just a middleman in this
chain — data center guys, equipment leasing companies — should never trade at higher multiples than
the company that controls their supply: Taiwan Semi, Nvidia, AMD

In a maturing capex cycle, the market eventually re-rates middlemen down. Chanos thinks the market
will be forced to distinguish, in his words, “what’s special versus what’s a commodity.”

A Connected Short: Alternative Energy

The same logic, Chanos argued, applies to alternative energy names that have been bid up on the data
center thesis. Geothermal, solar, and nuclear plays are trading at 50–70x earnings and 30–40x
EBITDA on the assumption they will power the next generation of AI infrastructure.

His counter is structural rather than ideological. The US is not, in his view, short on power — it has
natural gas in abundance. The real bottleneck is turbines, permitting, and red tape, and that resolves
on a two-to-three-year timeline. Power costs, meanwhile, are roughly 5–7% of data center revenues.
“It’s not the game changer.” He sees the gap between valuation and reality as one of the cleanest
setups available to a fundamental short book.

The Issuance Signal Investors Aren’t Pricing

The most overlooked data point in Chanos’s session may have been the simplest. “For the first time
since 2021, we’re seeing large amounts of issuance.” Wall Street’s printing press is printing stock
again, and historically that has been a late-cycle marker.

He expects 2026 to break all-time records for equity issuance — surpassing 1999, 2000, and 2021.
“Every time we’ve seen a wave of IPOs, it’s generally not been good for the stock market.”

The parallel he drew to the dot-com era was specific and worth thinking through. The 1998–2000 capex
boom around Y2K and telecom build-out artificially inflated S&P; 500 earnings; when that spending
slowed, earnings fell roughly 40% from mid-2000 to mid-2001. Today’s AI and data center capex,
Chanos argued, is functioning the same way — and the same demand myth is at the core. MCI
WorldCom famously claimed internet traffic was doubling every three months. It was actually doubling
every year. Today’s equivalent, in Chanos’s view, is the idea that demand for compute is structurally
infinite.

What This Means for Allocators

For LPs building exposure to alternative investments outlook 2026 themes — particularly hedged
equity, event-driven, and dedicated short books — Chanos’s session reframed the opportunity set. The
thesis isn’t that AI is fake. It’s that the second- and third-derivative trades around AI infrastructure have
been priced as if every cell in the chain captures the same value as the chip designers and the model
labs. They don’t.

Chanos also acknowledged the hardest part of the business: short selling is psychologically brutal, and
the talent pool is thin. Kynikos is building a training program to help allocators identify and develop
short-side analysts — a signal that even the most public bear in the market is thinking about pipeline.

And on the regulatory front, he noted that the Andrew Left case had factual complications beyond the
question of whether short publishers must hold their position. He expects the appeals courts to view the
legal concept differently than the district court did.

The Takeaway

Chanos’s view in one line: the market is paying premium-software multiples for capital-intensive
middlemen, on the assumption that demand never blinks. If issuance accelerates the way he expects,
the marginal buyer disappears at exactly the moment the marginal seller — corporate insiders, IPO
syndicates, secondaries — shows up in size.

Conversations like this one are why allocators show up to Global Alts New York 2026. The most
active LPs in alternatives use iConnections to find the managers who are positioned for exactly this
kind of dispersion — funds running dedicated short, long-short, and event-driven mandates with the
discipline to underwrite the AI infrastructure trade from both sides.