How Tessara launched on X

Tessara is opening wider access to research agents for people who invest in the physical AI buildout, covering chips, memory, advanced packaging, networking, power, cooling, and data centers. Company, research, forecasting, monitoring, and discovery agents share maintained company records, disclosures, and supply-chain relationships, so an investigation builds on that file. The workspace covers more than 380 public companies in the US, Taiwan, Korea, Japan, China, and Europe, with the deepest research in memory and semiconductors. Founder Teng Yan has spent the past year building it for investors who have to run the research desk themselves. They can search earnings calls, filings, news, and other sources, open the evidence behind an answer, and carry a question forward as new disclosures land. Tessara also publishes its own forecasts before results and grades them afterward. One call on the public record expected SanDisk fiscal Q4 2026 non-GAAP margin above 81 percent. The company later reported 84.6 percent. The launch video follows a single investigation: who can earn more from an AI cluster without adding another GPU. It sets repricing of older GPUs, recontracting a cluster after the initial financing is repaid, and higher utilization next to each other, then asks which path produces the most incremental profit and how long older capacity can keep earning.