WillItSurvive
AI companies and labs · United States

Baseten

Baseten hosts and deploys AI models across multiple clouds for companies running inference at scale.

Losing its protection
Assessed 20 July 2026 · Evidence as of 20 July 2026
The structural protection is eroding.
On the recordSince 20 July 2026 · Checkable from 2027 (est.) · Every call on the record Checkpoint: Expected Series G or IPO decision following Cerebras' Nasdaq debut
The call

Baseten is racing three well-funded peers and two sets of giants to stay useful before inference hosting gets commoditized or absorbed.

Baseten helps companies run AI models on different cloud providers (like AWS, Google Cloud, and Azure) instead of being locked into one. The company built real engineering to make this work, but the engineering is not unique: four well-funded competitors (Fireworks, Together AI, Modal, Replicate) can build the same thing, and Baseten's open-source software means customers can leave anytime. The real threat is that the big cloud companies are getting better at this themselves, and the AI companies making the models (like OpenAI) are also starting to offer cheap model hosting directly. If prices fall in a race between rivals, or if OpenAI and the cloud giants take over this business, Baseten's earnings could shrink fast.

Why
01
AI pressure: OpenAI and Anthropic internalizing hosted inference and fine-tuning for their own models. If frontier labs host inference themselves cheaply, Baseten's core pitch of running other people's models shrinks. Could AI do the core job? Only partly.
02
Bigger players: AWS SageMaker, Google Vertex AI, Microsoft Azure AI. Hyperscalers are closing the multi-cloud gap that Baseten currently fills, threatening its main reason to exist. Could a giant copy it? Probably.
03
What protects it. Neutral multi-cloud routing across 18 clouds that AWS, GCP, and Azure cannot offer without favoring their own infrastructure.
What would change this call
Baseten pursues Series G, IPO filing, or acquisition within 12 months. Would show whether investors believe it can outgrow Fireworks/Together or needs an exit before a price war compresses margins.
Independent third-party ARR estimate for Baseten is published and compared to its claimed 20x YoY growth. Would confirm or contradict Baseten's self-reported revenue figures, which are currently unreconciled with outside estimates.
OpenAI, Anthropic, or a hyperscaler launches a discounted hosted open-model inference product. Would show frontier labs or clouds internalizing Baseten's core 'cheaper alternative' niche.
Key numbers
  • $13B valuation in Series F (2026-06-22)
  • Claimed 20x YoY revenue vs ~$15.8M ARR estimate (2025-11-27)
  • Rival Fireworks AI: ~$800M ARR, ~$15B valuation (2026-06-10)
Named risks
  • Revenue figures unreconciled: company claims 20x YoY growth while third-party estimate put ARR at only ~$15.8M in June 2025. Severity: high
  • Margin compression from four-way price war with Fireworks, Together, and Modal in a commoditizing inference market. Severity: high
  • Frontier labs (OpenAI, Anthropic) and hyperscalers internalizing hosted inference could erode Baseten's mid-layer value proposition. Severity: medium
  • Rapid four-rounds-in-18-months fundraising cadence signals heavy cash burn tied to GPU procurement costs. Severity: medium
  • Employee reports cite contested leadership and unclear compensation transparency amid hypergrowth as of July 2026. Severity: low
Rivals, as read in July 2026
RivalDirectionEvidence
Fireworks AIcontestingFireworks AI estimated at ~$800M ARR and ~$15B valuation as of June 2026, roughly matching Baseten's scale
Together AIcontestingCombined valuation of Baseten, Together AI, Fireworks AI and Modal exceeds $30B as of June 2026 per industry report
ModalcontestingNamed as a similarly funded peer executing the same multi cloud inference playbook, as of 2026
OpenAIlosing toInvestor Michael Tolo of Blackbird framed Baseten as competing on price with OpenAI and Anthropic for enterprise inference, June 2026
AWS SageMakergaining onHyperscalers closing the multi cloud gap with native managed inference offerings, cited as ongoing threat in 2026 analysis
CerebrascontestingCerebras went public May 14, 2026 on Nasdaq (CBRS), raising $5.5B at ~$66B day-one market cap, opening capital markets for rivals
Questions people ask
Is Baseten profitable in 2026?
No public profitability data exists. Baseten raised a $1.5 billion Series F on June 22, 2026 at a $13 billion valuation, and its asset-light GPU resale model runs on thin margins, suggesting the company is prioritizing growth over profitability at this stage.
How much is Baseten worth in 2026?
Baseten was valued at $13 billion (and $11 billion in a second tranche) after its Series F round of $1.5 billion, announced June 22, 2026, up from a $2.15 billion valuation just nine months earlier in September 2025.
Who are Baseten's biggest competitors?
Baseten's closest rivals are Fireworks AI, Together AI, Modal, and Replicate, all similarly funded inference infrastructure startups. Fireworks AI is estimated at roughly $800 million ARR and a $15 billion valuation as of June 2026, per an industry report, making it a comparably scaled peer.
Could AI labs like OpenAI replace Baseten?
Possibly over time. Frontier labs including OpenAI and Anthropic increasingly offer hosted inference and fine tuning directly, and hyperscalers like AWS SageMaker are closing the multi cloud gap, which threatens Baseten's mid layer positioning as of 2026, though no direct displacement has occurred yet.
Is Baseten a safe company to join in 2026?
Baseten is hiring aggressively, growing headcount roughly from 50 employees in August 2024 to about 285 by June 2026 per Crustdata, a plus-173% year over year increase, but employee review sources cite contested leadership and unclear compensation transparency as of July 2026.
How fast is Baseten growing?
Baseten's own Series F announcement on June 22, 2026 claimed revenue grew approximately 20x year over year with inference volume up 40x, and the platform now processes over 1 billion inference calls daily across 87 clusters and 18 clouds, according to the company's blog post.
Timeline
  • 2026-06-22Series F closes: $1.5B across two tranches at $13B and $11B valuations, led by Altimeter Capital
  • 2026-05-06Baseten launches Frontier Gateway product
  • 2026-03-19Baseten Delivery Network launches, cutting cold starts 2 to 3x on large models
  • 2026-01-23Series E closes: $300M at $5B valuation, led by IVP and CapitalG, NVIDIA participates
  • 2025-09-05Series D closes: $150M at $2.15B valuation, led by BOND
  • 2025-05-01Baseten launches Model APIs and Baseten Training products
  • 2025-02-01Series C closes: $75M, led by Spark Capital (per Crustdata)
  • 2024-03-01Series B closes: $40M, led by Spark Capital (per Crustdata)
  • 2022-04-01Seed ($8M) and Series A ($12M) close, led by South Park Commons and Greylock
Revision history
20 July 2026
First assessed. Losing its protection. The text above is fixed as of this date and is not edited afterwards.
In the same band
Sources