Mythic, an AI inference processor startup developing analog compute-in-memory tech, raises $30M Series B1, bringing the total raised to $86M
Nicole Hemsoth / The Next Platform :
Context & Ripple Effects
Mythic's $30M Series B1 lands just over a year after its SoftBank-led $40M Series B, and the 'B1' suffix itself is the signal: rather than jumping to a Series C, investors are extending the existing round to fund the push of its analog compute-in-memory inference chip into production. That architecture — doing AI math inside memory rather than in a separate digital logic block — trades precision for dramatic gains in power efficiency at the edge.
The funding cadence across the corpus validates the patience this technology demands: Mythic later raised a $70M Series C in 2021, and by late 2025 closed a $125M round led by DCVC under a Nvidia-veteran CEO — meaning this 2019 bridge round sits early in a multi-round, multi-year capitalization path for analog inference silicon.
First-order effects
- Mythic gains an extended runway to take its high-speed, low-power inference chip from development toward market without resetting its valuation or cap table, keeping the SoftBank-backed plan intact.
Second-order effects
- Every incremental dollar into analog compute-in-memory raises the bar for rival edge-inference silicon makers — if Mythic hits its efficiency targets, digital-architecture competitors must match its watts-per-inference economics or cede the battery-constrained device market.
Third-order effects
- If the multi-year funding pattern holds — B1 in 2019, C in 2021, $125M in 2025 — analog inference becomes a structural second axis in AI silicon, pushing the industry toward heterogeneous compute where no single digital-GPU design serves both data center and edge.
The trend: AI inference silicon is diversifying beyond digital GPU architectures toward specialized approaches like analog compute-in-memory, sustained by patient, staged venture capital over long commercialization cycles.