Who Is Building Super Intelligence?

This page answers a narrow question: which organizations have publicly said they are trying to build Super Intelligence, in the research sense of a system that exceeds human cognitive performance across domains? It tracks stated goals, in the organizations’ own words. It does not assess capability, and it does not imply that any of them has succeeded. None has.

The list is organized by how explicitly each organization has adopted the goal: first those that put the word in their name or mission, then frontier labs whose stated target is AGI or “powerful AI” but whose leadership discusses Super Intelligence as the next step.

Inclusion criteria

An organization appears here if it meets at least one of the following, with a public, dated, linkable source:

1. Its legal name, division name, or published mission statement uses “superintelligence” or “super intelligence” to describe what it is building.
2. Its founders or chief executive have described building Super Intelligence (or a stated equivalent such as “superintelligent autonomous systems”) as the organization’s aim, in writing or on the record.
3. It is a frontier model developer whose stated goal is AGI or comparable, and whose leadership has publicly framed Super Intelligence as the subsequent objective.

Marketing use of the word alone does not qualify. Nor does adopting the federal “SI” label for ordinary AI products after Executive Order 14434, which we log separately in the terminology tracker.

Tier A: Super Intelligence in the name or mission

Safe Superintelligence Inc. (SSI)

The clearest case. On June 19, 2024, Ilya Sutskever, Daniel Gross and Daniel Levy announced SSI with the line “Superintelligence is within reach,” describing it as “the world’s first straight-shot SSI lab” with “one goal and one product: a safe superintelligence.” The company’s website still opens with “Building safe superintelligence (SSI) is the most important technical problem of our time.” Its updates page records a $1 billion raise in September 2024, Gross’s departure and Sutskever’s formal appointment as CEO in July 2025, and a July 2026 compute partnership with NVIDIA. By its own description, SSI has “one goal and one product.”

Meta Superintelligence Labs (MSL)

On June 30, 2025, Mark Zuckerberg told Meta employees that “As the pace of AI progress accelerates, developing superintelligence is coming into sight,” and that “We’re going to call our overall organization Meta Superintelligence Labs (MSL),” led by Alexandr Wang as chief AI officer, with Nat Friedman overseeing AI products and applied research. A July 30, 2025 letter set out the goal: “Meta’s vision is to bring personal superintelligence to everyone.”

Lila Sciences

Launched by Flagship Pioneering with $200 million in seed funding in March 2025, Lila uses a domain-specific form of the word. Its website states, “LILA is building Scientific Superintelligence,” framed as “Scientific Superintelligence to solve humankind’s greatest challenges” through AI-directed autonomous laboratories. CEO Geoffrey von Maltzahn describes the ambition as building “the highest scientific intelligence on Earth.” Lila’s target is narrower than general Super Intelligence, which is why we note the qualifier.

Reflection AI

Reflection AI qualifies on its founding statement rather than its current branding. Announcing the company on March 7, 2025, it wrote on LinkedIn: “At Reflection, we’re building superintelligent autonomous systems,” and “We believe that solving autonomous coding will enable superintelligence more broadly.” Founders Misha Laskin and Ioannis Antonoglou previously worked on Google’s Gemini models. The company raised $2 billion at an $8 billion valuation in October 2025. Its current website leads with a different framing, “We’re making intelligence open and accessible to all,” and does not use the word. We keep it in Tier A on the strength of the founding statement and will move it if the company publicly revises its goal.

Tier B: frontier labs pursuing AGI who discuss Super Intelligence

OpenAI

OpenAI’s Charter targets AGI, defined as “highly autonomous systems that outperform humans at most economically valuable work.” Its leadership has since gone further. In January 2025, Sam Altman wrote: “We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word.” In June 2025 he wrote, “OpenAI is a lot of things now, but before anything else, we are a superintelligence research company,” and “Humanity is close to building digital superintelligence.” In October 2026, asked about the federal rename, Altman said OpenAI had “no plans” to rebrand but “we have been talking about super intelligence a lot, and we will continue to talk about it.”

Google DeepMind

Google DeepMind’s stated mission is “to build AI responsibly to benefit humanity,” with AGI as the explicit long-term target. In June 2026 a team of DeepMind researchers including Shane Legg published “From AGI to ASI,” which examines the transition from human-level AGI to “a system that is more intelligent and cognitively capable than large organizations of humans,” via “scaling AGI, AI paradigm shifts, recursive improvement,” and multi-agent collectives. Demis Hassabis has said that AGI is “only a few years away” and that “Ten years from now, I think we’ll realize that we were standing in the foothills of the singularity now,” while recalling DeepMind’s founding pitch: “Step one: Solve intelligence. Step two: Use it to solve everything else.”

Anthropic

Anthropic generally avoids the word, preferring “powerful AI.” In “The Adolescence of Technology” (January 2026), CEO Dario Amodei defines it: “By ‘powerful AI,’ I have in mind an AI model—likely similar to today’s LLMs in form” with a list of capabilities he summarizes as a “country of geniuses in a datacenter.” He writes that “it cannot possibly be more than a few years before AI is better than humans at essentially everything,” and uses “superintelligent” once, describing “a superintelligent genius in their pocket.” Anthropic is included because its stated target is functionally the research concept, and because Amodei signed the White House Accord on Super Intelligence on September 29, 2026.

xAI (now SpaceXSI)

xAI’s mission is “to advance our collective understanding of the universe.” The company was folded into SpaceX and debuted as SpaceXAI in July 2026; on October 4, 2026 Elon Musk said it would be renamed SpaceXSI, posting “No more AI. SI, it’s better,” and its about page now carries the SpaceXSI name. Note that this is adoption of the federal label, not a new statement of the research goal; the company’s published mission does not use the word. Musk signed the White House Accord on Super Intelligence for xAI.

Microsoft AI

On November 6, 2025, Mustafa Suleyman announced: “To do this we have formed the MAI Superintelligence Team, led by me as part of Microsoft AI.” The post defines “humanist superintelligence” as “incredibly advanced AI capabilities that always work for, in service of, people and humanity more generally,” and states, “At Microsoft AI, we believe humans matter more than AI.” Microsoft AI sits in Tier B rather than A because the term names a team inside a larger division, not the division or company itself.

Thinking Machines Lab

Mira Murati’s company describes its purpose as building Thinking Machines Lab “to make AI systems more widely understood, customizable and generally capable.” Its public materials do not use “superintelligence” or “AGI.” It is listed because it is a frontier lab founded by former OpenAI leadership and widely grouped with the others, but it is the weakest fit on this page by our criteria. If it adopts the goal publicly, we will update; if it does not, it may be removed in a future revision.

Who is not on this list, and why

This page tracks stated goals, not capability. That cuts both ways.

• Companies that adopted the federal “SI” label for existing products (for example Salesforce’s SIForce) are not here. Relabeling AI as SI under Executive Order 14434 is a terminology choice, logged in the tracker, not a research goal.
• Governments are not builders. Executive Order 14434 renamed what federal agencies call AI; it did not announce a program to build Super Intelligence in the research sense.
• Capable labs that have not stated the goal (for example, several large open-model developers) are omitted regardless of how strong their models are. Appearing here is a function of what an organization says it is trying to do.
• Research groups and nonprofits whose work bears on Super Intelligence but who do not describe themselves as building it are out of scope for this page.

How this page is maintained

Every entry is tied to a dated, linked primary source, preferably the organization’s own words. Quotations are verbatim. We write “Super Intelligence” as two words in our own voice and preserve official spellings in names and quotes. Tier placement reflects explicitness of the stated goal, not our judgment of likelihood or progress.

This page was first published on 2026-10-19 and is revised when an organization changes its stated goal, name, or structure. Corrections and additions, with sources, to corrections@sibuilders.io.