Eren Berke Saglam

co-founder at @stoaexchange (YC S26) · prev @citi @dartmouth · building the foundation of ai capex market

[Eren Berke Saglam @erenberke0]

Weigh your options before you commit, there is a reason why a new neocloud is popping up every day. Most of the time owning the compute can be a better financial decision than renting.

[Berat Celik @beratcelik0]

A 3-year GPU commit costs more than owning the servers.

Most teams we talk to can't get GPUs on demand anymore, so they sign an offtake to reserve capacity. Decent rates only come with commitments of three years or more. Ask for six months or a year and the hourly price jumps.

When we run the numbers, owning beats a three-year commit, including colo (rented rack space) and power. At the end, you still have a server you can sell.

At $500K to $650K for a new 8x B300 server, owning comes to $3 to $3.70 per GPU-hour over three years before resale. A three-year commit costs $4.40 to $5.30 per GPU-hour, with 30% prepaid. Owning pulls ahead between months 10 and 26.

The hardest part is figuring out what you can sell the hardware for. Used GPU transaction prices aren't public. At Stoa Intelligence, we price five @nvidia platforms every week using executed trades and firm dealer quotes.

The A100 80GB came out in November 2020. One bought at launch hit $0 on the books in November 2025 on a five-year depreciation schedule. Our weekly price for a used 8x A100 system stayed between $51,763 and $58,314 from July 10 to October 2, 2026.

Owning is more doable than it sounds. A complete used 8x A100 server costs under $100K and gives you 640GB of GPU memory in one machine. Used 8x H100 systems got 12% cheaper over the same period. Suppliers on Stoa quote new B300 and B200 systems with firm prices and delivery dates, and used H100 and H200 systems available now.

Renting makes more sense if your usage is spiky or you need a few weeks of extra capacity. Buying takes cash up front, and we help with financing options.

The math and price charts are in the note below.

If you're about to sign an offtake, DM me your quote. We'll price out owning the same hardware.

Already own GPUs? Stoa Intelligence can value your fleet every week. DM me to set it up.

[Eren Berke Saglam @erenberke0]

Today we are launching Stoa Intelligence, a product I am particularly excited about.

For months the same question kept coming up in customer conversations: where would a given GPU configuration clear today?

There is plenty of pricing in this market. The problem is it lives in private quotes and bilateral trades, so any one desk only ever sees its own flow. Rental and residual value models fill some of the gap. What they cannot tell you is where a specific configuration is being bid, offered, or traded right now.

That matters when you are deciding whether to hold inventory or rotate a fleet. It matters a lot more when you are valuing a portfolio or lending against the hardware.

Stoa Intelligence takes pricing activity captured on Stoa and turns it into hardware marks, current market levels by configuration, fleet and collateral monitoring, and analysis for valuation and GPU backed financing. A wider view of the market than any single desk can build from its own quotes and trades.

Sign up at https://t.co/zydaN1IntD.

[Eren Berke Saglam @erenberke0]

@beffjezos We can help at @stoaexchange. Have used A100s ready to ship

[Eren Berke Saglam @erenberke0]

@emilares @MarengoYCS26 @gadmarconi @ycombinator congrats guys! possibly a client in the future 👀

[Eren Berke Saglam @erenberke0]

@StockMKTNewz what does selling debt mean? do you mean issuing bonds? secondary sale of any financing portfolio? the vagueness here doesn't help

[Eren Berke Saglam @erenberke0]

@rywicks @pennant_ai @ycombinator Congrats!

[Eren Berke Saglam @erenberke0]

@bosmeny @paulg the man, the myth, the legend

[Lloyd Armbrust @larmbrust]

PG buying domains for random S26 companies is the most PG thing I’ve ever seen

[Eren Berke Saglam @erenberke0]

I think you have to consider the incentives for all the parties:

$NVDA: this means taking some of the balance sheet weight of backstopping/underwriting, and outsourcing this stabilizing role to financial institutions. smart and unlocks more revenue for them as barrier to entry to buy nvidia chips is now smaller.

financial institutions: they have the ability to source the debt, and sell it out in the secondary market. Debt ETFs, structured credit, there are liquidity vehicles that they have access to which nvidia may not. Also smart for them as they get the 25% residual value guarantee from nvidia and get access to this highly demanded AI credit profile

how can we be responsible about it? we need to understand the value of these chips better in the secondary market. As Jensen puts it, the value of compute and the hardware is market driven now, rather than just through a price tag.

[Za @ZaStocks]

Everyone hates the $NVDA financing which probably means it’s ultimately going to work.

No, we’re not smarter than Jensen Huang and the executives at Goldman, Blackrock, Blackstone, Apollo, and Brookfield who are structuring these deals.

[Eren Berke Saglam @erenberke0]

@8teAPi Who wins the financing game will be those that understand the residual value deeply and price beyond the 25% guarantee

[Eren Berke Saglam @erenberke0]

We provide the pricing and trading layer of this hardware at @stoaexchange. Residual value so far has been resilient, but this type of a financing scheme makes transparent, market driven residual value marks more important than ever

[Jensen Huang @JensenHuang]

https://t.co/pppEbPVtfE

[Eren Berke Saglam @erenberke0]

“This is Nvidia's way of saying that Hyperscalers will not be the only ones that get investment grade (IG) financing. Nvidia is making IG financing available to all Neoclouds and AI Platforms.”

The way you put it unfortunately reminds me of something that starts with subprime.. i think thats kind of a misrepresentation

[Eren Berke Saglam @erenberke0]

@_doganay_ege Waiting to get some for data centers! Congrats

[Eren Berke Saglam @erenberke0]

seeing how our clients leverage gpu hardware directly instead of compute rentals to adapt open source workflows amazes me every day. opens up so many possibilities

[Tanay Jaipuria @tanayj]

Per Pinterest earnings call, open models they post-trained on their own data costs under 8% per transaction versus comparable closed models

Their CEO said: "Any CEO that’s not taking advantage of open source models is almost certainly wasting a lot of their shareholders’ money"

[Eren Berke Saglam @erenberke0]

definition of pmf

[Rhys @RhysSullivan]

we literally have no idea how to make software like this anymore

[Eren Berke Saglam @erenberke0]

@zeroxkyle he's a legend when it comes to timing execution. not necessarily fundamentals. trading has always been about first principles

[Eren Berke Saglam @erenberke0]

@paulg as software becomes more commoditized with AI, the focus on hardware seems to have gone up. its encouraging to see more and more hardware startups

[Eren Berke Saglam @erenberke0]

there's much more that comes with being a founder, but I think that's a succinct way of putting it

[Paul Graham @paulg]

17 yo asked what makes a good founder. I said being good at building stuff and being sufficiently fierce. That's about it.