Standard Intelligence, a San Francisco-based AI research company focused on building aligned artificial general intelligence, has raised $75 million in new funding from Sequoia and Spark Capital. The investors joining the round include Sonya Huang and Mikowai Ashwill from Sequoia and Yasmin Razavi from Spark Capital, all of whom the company describes as deeply supportive of its long-term mission. The round also includes a group of prominent angels and advisors, among them Milan Kovac, Stanley Druckenmiller, and Andrej Karpathy, bringing together an exceptional set of backers for what remains a six-person team.
The company says its early results with its FDM-1 model moved computer use from a data-constrained regime to a compute-constrained one, a meaningful inflection in the trajectory of agentic AI research. The new capital will unlock several orders of magnitude of compute scaling for that work. Standard Intelligence believes that with the FDM model series it has a viable path to scale agentic capabilities through video pretraining, with an expectation of achieving superhuman performance on general computer tasks in the same way that current language models have achieved superhuman performance on coding tasks.
The funding also enables Standard Intelligence to invest in the blue-sky research it says is necessary to achieve its long-term mission of building aligned general learners. The company argues that to realize the transformative potential of AI, models must be capable of generalizing far out of their training distributions, actively exploring and building skills in entirely new environments. This represents what Standard Intelligence describes as a substantial shift from the current paradigm of model training, where models are largely optimized to perform well within the bounds of their training data.
Standard Intelligence says it believes current alignment techniques are insufficient to predictably and safely steer a model with human-level learning capabilities. To address this, the company is conducting research on small versions of the problem in controlled environments, working to develop what it calls a science of alignment for general learners. With the new funding in place, Standard Intelligence says it is actively hiring world-class researchers and engineers to join its team in San Francisco.
