Outlook & Investment Thesis

Bull case

The case that Vast looks great in 5 years:

1. GPU compute remains scarce relative to demand. If demand for AI compute keeps growing faster than supply, marketplaces with deep liquidity capture more value. Vast's supply-aggregation moat compounds.
2. Marketplace efficiency advantages accumulate. As more supply and demand flow through Vast, the matchmaking quality, pricing signals, and trust data get better. Network effects strengthen.
3. The customer base expands. AI is being adopted by progressively less-technical builders. As more indie / small-business AI builders enter the market, Vast's segment grows.
4. Hyperscaler pricing remains structurally high. The 60-80% gap persists because hyperscalers can't lower GPU pricing without compressing their AI margins broadly. Vast's price advantage holds.
5. Compute futures stabilize the supply economics. Compute futures markets (now live through CME + Silicon Data and ICE + Ornn) could provide more predictable pricing for providers, encouraging more long-tail supply onto Vast.
6. Capital-efficient model wins. If enterprise neoclouds get squeezed on margin (over-built capacity, GPU price compression, debt servicing), Vast's lean structure looks comparatively strong. The marketplace doesn't have a balance sheet problem.

Under this scenario, Vast is a profitable, sustainable, modestly growing business that captures meaningful share of the long-tail AI compute spend without needing to chase enterprise. The strategic value is "the indie / research GPU cloud of record."

Bear case

The case that Vast looks weak in 5 years:

1. GPU prices normalize. The Blackwell generation and beyond eventually meets supply, prices fall, and the 60-80% discount Vast offers compresses. The platform's central value proposition weakens.
2. Hyperscalers respond with marketplace-style products. AWS or Azure launches a managed marketplace with curated supply and enterprise-friendly compliance. The category boundary blurs.
3. Managed inference eats the inference workload. Together / Fireworks / similar continue to win for inference. Vast's GPU-hour model becomes less relevant for that workload class.
4. Big multi-node training continues to dominate marquee AI workloads. Frontier model training requires fabrics Vast can't deliver. As the value of frontier work grows, Vast looks marginalized.
5. Competition compresses take rate. RunPod, TensorDock, and others put pressure on the take rate. Vast either holds rate and loses supply share, or cuts rate and loses revenue.
6. Customer scale-up bleed continues. Vast's biggest customers keep leaving as they grow. The funnel works but the leak at the top means lifetime value stays low.

Under this scenario, Vast is a quiet utility that persists but doesn't grow much. Defensible in its niche, but the niche stops being a strategically important part of the market.

Wildcards

Outcomes that could meaningfully shift Vast's trajectory either direction:

Compute futures markets

The CME and ICE compute futures launches in 2026 are too new to fully evaluate. The implications for Vast cut both ways. Positively: standardized compute pricing could legitimize the spot-market dynamics Vast already runs, making the marketplace look more institutional. Negatively: if compute futures encourage capital-rich players to build forward positions, capacity could grow ahead of demand and depress marketplace prices. See the dedicated guide on physical futures marketplaces for the full picture.

AGI-adjacent capability bursts

If frontier AI capabilities accelerate sharply, the relevant cluster becomes only the very biggest training runs. Vast becomes a tertiary infrastructure layer behind hyperscalers and frontier-lab providers.

AGI-adjacent capability stalls

If frontier scaling laws flatten, the relative importance of long-tail compute (inference, customization, smaller-scale training) grows. Vast benefits.

NVIDIA strategic moves

NVIDIA could change how it allocates supply, supports the secondary market, or partners with marketplaces. Any of these changes would ripple through Vast's supply.

Acquisition

Vast could be acquired. Possible acquirers include hyperscalers wanting marketplace assets, enterprise neoclouds wanting a long-tail channel, or financial buyers wanting a profitable platform business. Vast has been content to stay independent, but the option exists.

AMD / non-NVIDIA chips break out

If AMD's MI300X / MI325X line or other non-NVIDIA accelerators capture share, Vast benefits if it can grow non-NVIDIA supply. The early signals on AMD on Vast are positive but small.

Scenarios for 2027-2030

Scenario A: Vast as the steady marketplace

Vast continues its current trajectory. Profitable, modestly growing, dominant in its segment. GMV grows in the 10-20% range annually. The company stays independent. Market views Vast as the "Airbnb of GPUs" — a known utility, valued at marketplace multiples.

Scenario B: Vast gets acquired

An acquirer (likely a hyperscaler or large enterprise neocloud) buys Vast to capture the marketplace asset. The acquisition partially preserves Vast's brand and segment focus but integrates platform-level economics. The marketplace's defensibility persists; the cultural posture may shift.

Scenario C: Vast pivots up-market

Vast launches a meaningful enterprise tier — curated supply, account managers, SLAs. The company's headcount grows. The cultural break with the indie-first posture is real but the addressable market expands. Successful execution is non-trivial; the segment ceiling could lift materially or the brand could split.

Scenario D: Vast becomes marginalized

A combination of bear-case factors (price compression, managed-inference encroachment, competitive marketplaces) shrinks Vast's relative position. The company remains profitable but the share of compute spend it captures declines.

Probability-weighted, the most likely scenarios are A and B, with C and D as tail outcomes. Vast's lean structure and clear strategic posture suggest more continuity than disruption.

Exit pathways

If Vast ever chose to monetize differently, the options are:

• Continued bootstrap-style growth. Keep operating as a profitable independent business indefinitely. This is consistent with the historical posture.
• Acquisition. Strategic acquirer absorbs Vast for the marketplace asset. Probably the most likely meaningful exit.
• IPO. Possible but unlikely given Vast's reluctance to grow at venture pace.
• Roll-up acquisition by Vast. Vast acquires smaller marketplaces or providers. Reverse direction; less likely given the company's culture.

What to watch as an observer

Indicators that signal Vast's trajectory shifting:

• GMV trends. Hard to observe directly; community signals like "is supply tight or abundant" act as proxies.
• Hiring patterns. A growing engineering team suggests scaling investment. Sales hires suggest enterprise pivot.
• Product launches. Big new product surfaces (managed services, enterprise tiers) would be a strategic signal.
• Fundraising or sale activity. Any public reporting of capital moves would be significant.
• Take-rate changes. Public-facing fee schedule changes signal where Vast is feeling competitive pressure.
• Compute futures adoption. If Vast launches futures-related products (hedging tools for providers, fixed-price contracts for clients), that's a meaningful evolution.

Takeaway

Vast is one of the most strategically interesting companies in the neocloud landscape precisely because it's chosen a structurally different model from the well-funded enterprise neoclouds. Its trajectory tells us something about whether the marketplace model can sustain through hardware cycles, managed-inference competition, and the broader maturation of AI compute markets. It's worth watching not because it's the biggest, but because it's the cleanest test of an alternative thesis.

If you've read all twelve chapters, you have a deep view of Vast specifically. The next step is the competitor guides — for each major competitor in this topic — to put Vast in context against the rest of the landscape.