SA
SES AI Corp (SES)·Q4 2024 Earnings Summary
Executive Summary
- SES AI recorded first-ever revenue of $2.04M in Q4 2024 with a 63% gross margin, marking the transition from pre-revenue to commercialization; GAAP operating expenses were $30.4M and diluted EPS was -$0.11 .
- Management initiated FY 2025 revenue guidance of $15–$25M and expects total cash usage (operations + capex) of $70–$80M; liquidity runway guided into 2H 2028 supported by $262.5M in cash and short-term investments at year-end 2024 .
- Strategic pivot “All-in on AI” broadened SES from Li‑Metal EV cells to AI-enhanced Li‑ion 2170 cells (humanoid robotics, drones) and BESS; 2025 revenue expected to be driven largely by BESS (AISPEX MOU up to 100 MWh and ~$45M pipeline) .
- Prior quarters showed disciplined spend and growing data/AI capabilities: liquidity $294.7M in Q2 and $274M in Q3; R&D scaled for Molecular Universe and Avatar safety models, setting up Q4 revenue and 2025 guidance .
What Went Well and What Went Wrong
What Went Well
- First commercial revenue with strong gross margin: $2.04M revenue and 63% gross margin in Q4, driven by OEM AI-for-Science contracts and initial deliveries for SoftBank HAPS and Data Blanket drones .
- Strategic expansion beyond EV into humanoid robotics/drones via AI-enhanced 2170 cylindrical Li‑ion cell; NVIDIA-enabled Molecular Universe accelerated electrolyte discovery from “>8,000 years” to months .
- BESS entry with AISPEX MOU (up to 100 MWh and ~$45M pipeline), plus conservative and first-time FY25 guidance ($15–$25M revenue) indicating improved revenue visibility .
- Quote: “All in on AI is not only critical, but is the future.” – Qichao Hu (CEO) .
What Went Wrong
- GAAP net loss widened in Q4 to -$34.5M, including a -$8.6M fair value change in Sponsor Earn-Out liabilities; operating expenses rose to $30.4M with R&D $20.9M .
- Consensus estimates comparison was unavailable due to S&P Global request limits, limiting beat/miss analysis for EPS/revenue [GetEstimates error].
- BESS MOU is non-binding and market is fragmented; management acknowledged initial focus on medium-sized, behind-the-meter segments to gain data and credibility before larger deployments .
Financial Results
KPIs and Operating Context:
- First revenue sources: AI-enhanced Li‑Metal/Li‑ion work for auto OEMs, initial deliveries to SoftBank (HAPS) and Data Blanket (drones) .
- FY 2024 total cash usage (operations + capex): $78.3M vs prior guidance $80–$95M (below low end) .
- Year-end 2024 liquidity: $262.5M; no debt outstanding per prelim release .
Guidance Changes
Earnings Call Themes & Trends
Management Commentary
- “We are no longer a pre-revenue company and look forward to continuing on our commercialization path by growing revenue in 2025 and beyond.” – Qichao Hu (CEO) .
- “This revenue comes with a 63% gross margin, demonstrating the strong value proposition of our technology and All-in on AI strategy.” – Jing Nealis (CFO) .
- “Commercialization of AI for Safety has also opened a market for Battery Energy Storage Systems that we believe is 10 times larger than automotive.” – Qichao Hu (CEO) .
- “Our discovery of an AI-enhanced molecule enabled us to unveil a new 2170 cylindrical cell … opening humanoid robotics and Li‑ion EV applications.” – Qichao Hu (CEO) .
- “With our capex-light business model, we now expect to maintain adequate liquidity into the second half of 2028.” – Jing Nealis (CFO) .
Q&A Highlights
- Business model mix and margins: Q4 gross margin 63% was a mix of AI services and cell sales; future margins will depend on volume and application mix; aim for high-margin CM and software layering over time .
- Revenue mix 2025: Management expects the bulk of FY25 revenue from BESS, with remainder from EV and drones/robotics; providing full-stack (cells, pack, BMS) solutions for BESS .
- Subscription model: Plan to offer annual subscriptions to Molecular Universe maps/models and Avatar; hardware used as a “hook” to sell software .
- Manufacturing geography: Currently CM mainly in Asia; drone customers prefer U.S.-based manufacturing; SES plans phased shift to U.S. .
- IP and competitive moat: Electrolyte molecules can be copied; SES focuses on speed and proprietary AI maps/models that competitors cannot easily replicate .
Estimates Context
- Wall Street consensus estimates (S&P Global) were unavailable at the time of this analysis due to request limits. As a result, beat/miss vs consensus for Q4 2024 EPS and revenue cannot be determined at this time [GetEstimates error].
- Implication: Sell-side models likely need to incorporate SES’s first revenue ($2.04M with 63% GM in Q4) and initial FY25 revenue range ($15–$25M), plus the BESS-heavy mix and capex-light approach highlighted in guidance .
Key Takeaways for Investors
- SES crossed the commercialization threshold with first revenue and strong gross margin; a meaningful narrative shift for the stock from R&D to revenue-generating operations .
- The pivot to “All-in on AI” broadens TAM and accelerates monetization via software and materials across EV, robotics/drones, and BESS; 2170 Li‑ion product opens near-term non-EV opportunities .
- 2025 revenue guidance ($15–$25M) and BESS focus (AISPEX pipeline) suggest early revenue scaling driven by Avatar-enabled safety/manufacturing; watch for conversion of MOU to firm orders and deployment milestones .
- Cash discipline remains credible: FY24 cash usage came in below guidance ($78.3M); liquidity of ~$262.5M supports runway into 2H 2028 under a capex-light model .
- Risks: Q4 net loss widened; earn-out fair value volatility; fragmented BESS market; execution required to shift manufacturing to U.S. for sensitive accounts .
- Trading lens: Near-term catalysts include BESS project progression, additional OEM contracts for AI-for-Science, and subscription software traction; stock sensitivity likely tied to evidence of order conversion and revenue cadence vs guide .
- Medium-term thesis: If SES can scale high-margin AI and materials revenue while leveraging CM for hardware, the blended margin structure could improve; monitoring segment mix (BESS vs EV vs drones/robotics) will be critical for model accuracy .