Question
When will Ineffable Intelligence publish its first public technical artifact: a research paper, preprint, benchmark result, or public demonstration?
The median estimate for Ineffable Intelligence's first public technical artifact is November 15, 2027, anchored heavily by founder David Silver's recent public statements but allowing for the slower pace of public research outputs modeled in related forecasts for new AI labs. As of November 14, 2026, the company remains quiet: it has no papers on alphaXiv Ineffable Intelligence Ineffable Intelligence and its website lists no technical artifacts beyond a January 2026 introductory note Ineffable Intelligence. However, the founders come from deeply academic cultures, and the threshold for this milestone is low—a single preprint, workshop paper, benchmark number, or public conference demonstration will suffice.
The strongest evidence for a late 2027 release comes from Silver’s October 13, 2026 Sequoia podcast interview. He forecast "strong evidence within a year" and explicitly stated that Ineffable intends to "publicly share" the science it develops in simulation David Silver and Lasse Espeholt of Ineffable Intelligence David Silver and Lasse Espeholt of Ineffable Intelligence David Silver and Lasse Espeholt of Ineffable Intelligence David Silver and Lasse Espeholt of Ineffable Intelligence. Current operational signals align with this multi-month horizon. Ineffable is still heavily engaged in foundational infrastructure building; it was the first customer for Google Cloud’s new A5X instances (which only reached general availability in November) Google Cloud sets November availability for NVIDIA Rubin ... Google Cloud and NVIDIA Announce General Availability of NVIDIA Vera ..., and the company was still hiring for a "Simulated Universes Engineer" in late September Member of Technical Staff, Simulated Universes. This suggests that large-scale reinforcement learning experiments are only just beginning, making headline results unlikely before mid-to-late 2027. Allowing a modest interval between evidence of scientific progress and a public technical artifact puts the median in mid-November rather than directly at the anniversary of Silver’s interview.
While major breakthroughs will take time, an earlier resolution in the first half of 2027 remains distinctly possible. The minimal resolution criteria could be met by a minor preprint bearing an Ineffable affiliation, an incremental workshop paper aligning with the ICLR or ICML conference cycles, or a limited infrastructure benchmark. For instance, a co-marketed infrastructure talk at Nvidia GTC or Google Cloud Next in early 2027 could easily cross the threshold for a public demonstration or benchmark result NVIDIA, Ineffable Intelligence Team Up to Build the Future ... NVIDIA, Ineffable Intelligence Team Up to Build the Future .... These routes support a 10% probability of a qualifying artifact by February 15, 2027, and a 25% probability by May 25, 2027.
The right tail extends through 2028 and 2029 to the start of 2030 due to the risks of prolonged delays or deliberate secrecy. While Silver intends to share results, a prediction of scientific progress "within a year" is not a firm commitment to publish. Ineffable could mimic the extreme secrecy of Safe Superintelligence (which has published nothing in over two years), or Silver could revert to his AlphaGo-era playbook, preferring to withhold incremental papers until a massive, highly polished milestone is achieved. Furthermore, a massive $1.1 billion seed round provides enough runway to shield the lab from the near-term fundraising pressures that typically force early disclosures. These risks put the 75th percentile at August 15, 2028, and the 90th percentile at January 1, 2030.
Consistent with related forecasts modeling slower public research outputs for new AI labs, the median date was shifted from late October to mid-November 2027.
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited