Pivot and fail fast: lessons from 12 years to IPO

Executive overview

Building a company through pivots requires letting go of past momentum and relationships faster than feels natural. SES AI learned this lesson through a 12-year journey from MIT research to NYSE-listed battery company, developing lithium metal cells and using AI to accelerate material discovery. The core insight: speed of execution matters more than emotional attachment to your previous direction.

Emotional attachment slows execution. Release it faster.

Letting go during pivots

When you pivot, you sacrifice relationships, momentum, and past work. It's natural to feel emotionally attached to what you built, but growth demands abandonment. The biggest mistake: holding onto the old too long rather than committing fully to the new direction.

Learning from A123's failure

A123 Systems was America's battery hero but expanded too fast on government funding. They hired 5,000 people in three months, built giant factories, and abandoned focus on quality and cost. Revenue grew with massive negative margins until the company collapsed. Key lesson: growth must be organic, not inorganic. One tab-welding defect shipped to GM cost them the contract.

Capital discipline beats peer spending

When markets are hot, raise as much cash as possible but invest it selectively. Many peers spent aggressively on CapEx during 2022-2023 and now face shutdowns; SES AI maintained financial discipline. Companies that spent $100 on CapEx are now forced into fire-sale liquidation below one cent on the dollar.

Board and investor selection is critical

A bad investor or board member is worse than any technical challenge because it's unworthy of your time and unsolvable by execution. Be extremely selective with investors: a dollar from a big name is worth the same as a dollar from an unknown, but the chemistry and alignment matter enormously.

Clear power structures enable quick decisions

When the CEO lacks clear authority over the board, pivots and difficult decisions slow down. Define power explicitly in shareholder and board agreements. Don't be polite: if investors won't give you CEO power, don't accept the title.

AI accelerating battery validation

New battery chemistry normally takes 10 years to validate in real-world conditions. SES AI uses three AI solutions (for Science, Manufacturing, and Safety) to compress this into one year, taking real-world data and training models to predict nine-year behavior. This acceleration is now forming a new ecosystem combining gigafactory data with large language models and compute partners like NVIDIA.

Beyond batteries: transportation and climate

SES AI is pivoting from battery company to power company for transportation. High-altitude pseudo-satellites (drones flying at 30,000 meters as flying cell towers) use lithium metal for remote Wi-Fi access. The Arctic Computing Center scales beyond batteries to address climate challenges like ocean plastic waste and new antibiotics.

Reframing failure and success

Success and failure are temporary states, not emotions. Assign less emotion to outcomes and treat them pragmatically. This mental model prevents overthinking and keeps focus on execution.