With $1 billion in new capital and a partnership with JPMorganChase, SambaNova is making a strong statement in the AI inference market.
The company has also put its SN50 AI chip on the map, all of which points to a major step forward for enterprise AI infrastructure.
There is a sense that the race for enterprise AI is moving into uncharted territory. For some time the industry has been preoccupied with creating ever larger and more potent models, but the focus is now turning to the infrastructure required to make them work in practice. Be it for business analytics, fraud detection or chatbots, any AI application needs inference that is cost effective, reliable and fast.
SambaNova has seen the opportunity in this shift and raised $1 billion in a Series F round, putting its valuation at $11 billion and underlining what investors think of the future for enterprise AI. It is a timely announcement for the firm. In addition to the funding, JPMorganChase has come on board to use SambaNova’s chips for on-premises inference, and SoftBank will be the first to deploy the new SN50. Such moves put SambaNova in good standing as a competitor in an expanding market.
Confidence in AI Inference
General Atlantic led the Series F round, joined by the likes of BlackRock, Intel Capital, Temasek, GIC and the Qatar Investment Authority. With this money SambaNova intends to take its platform to international markets and speed up enterprise rollouts.
The size of the investment tells you where the smart money is: they see the next big thing in artificial intelligence not in training foundation models, though that is still necessary, but in inference. That is the process by which an AI model analyses information and produces a result for a real-world application. As companies stop experimenting and move to production, the need for an efficient setup is only going to grow.
A Partnership with JPMorganChase
In a move that adds to its momentum, SambaNova has revealed that JPMorganChase has chosen its Reconfigurable Dataflow Units (RDUs) for high-performance, on-premises work. One can appreciate why a financial institution would want that level of control over sensitive data while enjoying low-latency processing; security and compliance are non-negotiable.
For SambaNova it is validation of its technology. It also provides an answer to the GPU-heavy infrastructure that has long dominated the field, giving enterprises something built for inference rather than a general-purpose graphics processor.
SoftBank and the SN50 Chip
Then there is the matter of the SN50. SoftBank is set to be the first partner for the newly launched chip, which fits with SambaNova’s longer term plan for an integrated AI infrastructure platform. The company touts the SN50 for its cloud-scale capabilities, cost efficiency and speed. It is about more than just selling hardware.
The Next Growth Engine
You could say that for years the conversation has been about model training. But with AI being rolled out in healthcare, banking and manufacturing, inference is of equal import. It is the part of the operation that runs continuously, making predictions and handling requests in real time. Enterprises are looking for systems with the flexibility to handle mission-critical tasks without breaking the bank.
This is where SambaNova’s new backing comes in. It shows that investors are open to alternatives in the chip market. There is a demand, particularly from those with highly regulated data, for specialised hardware that offers better on-premises control than the traditional GPU-first approach.
What it boils down to for an organisation is that having a powerful model is not enough anymore. You need the right kind of infrastructure to deploy it at scale and with the proper security. A specialised platform can lower costs and get AI out of the pilot phase and into daily operations.
SambaNova has the resources to go global with its latest raise. Whether it can turn that into widespread customer adoption through its partnerships and hardware is the test. But with the product portfolio and validation it has today, the company is well placed to have a say in how the AI landscape develops.
