Latest news with #AutoGPT
Forbes
06-05-2025
- Business
- Forbes
PayPal's Big Bet On AI Agents And Payments Like Stablecoins
Alex Chriss, CEO of PayPal, on stage at PayPal Dev Days Sandy Carter At its 2025 Dev Days event, PayPal made one of the most significant announcements in recent fintech history: it's building a Financial Operating System (Financial OS) tailored specifically for AI agents—autonomous software that can browse, buy, negotiate, refund, and optimize transactions for users. This isn't just another product launch. It's a foundational shift. With over 430 million users and $1.5 trillion in annual payment volume, PayPal is already a global giant. But now, it's setting its sights on becoming the infrastructure layer for agentic commerce—a fast-emerging model where intelligent agents handle financial activity independently, on behalf of people and businesses. The timing couldn't be better. AI agents are gaining momentum fast. Per Gartner, they project that by 2026, 40% of enterprise workflows will be automated by agents. By 2030, the AI agent economy is expected to exceed $22 billion in value. We're already seeing growth explode on platforms like AutoGPT and across open-source agent ecosystems on GitHub. Jeremiah Owyang, partner at Blitzscaling Ventures, captured this future well when he said, "In a world run by AI agents, websites may become obsolete. We won't browse—we'll instruct. The interfaces of the future are conversations, not clicks." That may sound far off, but PayPal is getting ready now. Its Financial OS is built for a new kind of two-sided marketplace—where agents don't just support buyers and sellers, they are the buyers and sellers. As part of this vision, PayPal's support for cryptocurrency—including its own stablecoin, PayPal USD (PYUSD)—gives developers and agents more flexibility in how transactions happen. PYUSD offers a stable, programmable on-chain currency alongside fiat, opening the door to more inclusive and global agent-driven commerce. It's no longer just about apps and clicks; it's about AI-to-AI negotiation, execution, and trust. Why AI Agents Are the Future of Commerce For PayPal And The Industry Let's be clear—AI agents aren't just glorified bots. These are autonomous, intelligent software entities that learn, make decisions, and act on behalf of users. Imagine this: your agent scans a product in-store, compares prices across platforms, places an order, tracks delivery, and handles a return—without you lifting a finger. PayPal sees these agents as the next generation of economic actors. And it's designing its infrastructure to become their financial backbone. PayPal Dev Days in California PayPal What Makes PayPal's Agentic OS Different While many companies are dipping into AI, PayPal is going all in—rebuilding its core systems to support intelligent agents from the ground up. It's launching tools like the Agent Toolkit and MCP (Multiparty Compute Protocol) server to give developers everything they need to build secure, compliant agent workflows. This is more than automation. It's about intelligent, seamless commerce built on PayPal's trusted rails. Agents can now manage payments, invoices, refunds, and even compliance in one streamlined flow. For developers, that means less stitching together APIs. For businesses and consumers, it means smarter, faster, frictionless transactions. As PayPal CEO Alex Chriss put it, "One of PayPal's greatest advantages is our two-sided network. While most fintechs are built for either merchants or consumers, we've built a platform that serves both—and now, we're extending that foundation to support intelligent agents as well." The result? AI Agents on both sides of a transaction can communicate, complete purchases, resolve issues, and protect against fraud—without human intervention. Powering the Ecosystem: PayPal, Google, AWS, and Microsoft At Dev Days, we saw what this looks like in action. Google showed how PayPal's tools power shopping agents using Gemini, enabling personalized, conversational commerce. AWS demonstrated digitizing physical goods into shoppable storefronts with its Nova models. And Microsoft revealed how Azure AI is building agent-powered retail flows. These aren't just demos—they're live use cases. Together, they signal that PayPal isn't just building a payments layer. It's creating a full-stack intelligent commerce platform. One that empowers agents to manage subscriptions, initiate purchases, and optimize behavior—autonomously. To do that, a Financial OS needs more than just payment processing. It needs programmable wallets, digital identity, agent-to-agent protocols, and real-time APIs. Think of it as the financial nervous system of the AI economy—with PayPal delivering the trust, Google and Microsoft providing the intelligence, and AWS bringing the global scale. PayPal's AI Story Wasn't First So Where Do Visa and Mastercard Fit? Traditional players like Visa and Mastercard are quickly moving into agentic commerce, with Visa's Intelligent Commerce APIs and Mastercard's Agent Pay demonstrating their commitment to enabling AI-powered transactions. These are meaningful steps—and signal that the race to support intelligent agents is now in full swing. PayPal may still hold a strategic edge. It was one of the first to articulate a full Financial Operating System designed for AI agents, and its two-sided network—serving both merchants and consumers—offers a native advantage for agent-to-agent flows. The company's integration of its stablecoin, PYUSD, adds programmability and global reach that could power both fiat and onchain agents. Critically, PayPal isn't starting from zero. Its $4 billion acquisition of Honey is now proving interesting. At the time, Honey had over 17 million monthly active users and a browser extension that helped shoppers find coupons and deals in real time. More than just a coupon tool, Honey offered AI-driven shopping insights, price tracking, and behavioral data—elements that are now highly valuable in training and powering intelligent agents. Agents powered by PayPal won't just transact—they'll anticipate needs, hunt for savings, and personalize offers in real time using the data and tooling inherited from Honey. That gives PayPal a unique blend of payments infrastructure, programmable money, and consumer intelligence—three foundational pillars for building agentic commerce at scale. As this next chapter of autonomous commerce unfolds, it's no longer just about who can process a transaction—it's about who can enable the smartest agents to do it all. What PayPal's AI Agent Means for Businesses and Developers For developers, this new Financial OS simplifies what used to be complex. Instead of manually coding checkout, refunds, or compliance, you can build those flows with natural language or streamlined APIs. For businesses—especially startups and SMBs—it's a game changer. You can deploy agents to handle repetitive tasks like order management or post-sale support, freeing up your team to focus on strategy and innovation. I had the pleasure of talking with Michelle Gill, General Manager, Small Business & Financial Services. As General Manager of PayPal's Small Business and Financial Services Group, Michelle Gill is responsible for bringing together the products and services that help small business owners run and grow their business. Michelle Gill told me that 'Because PayPal serves over 70% of the largest merchants across North America and Europe, plus millions of SMBs, we can verify the merchants with whom consumers interact.' Michelle Gill, General Manager, Small Business & Financial Services, keynoting at PayPal Dev Days. Michael O'Donnell photographer And for consumers? The experience becomes seamless. Your agent knows your preferences, your wallet, and your needs—finding the best deals, handling the logistics, and keeping you secure, all in the background. From Automation to Autonomy: The Big Shift For PayPal PayPal's vision isn't just about faster payments. It's about shifting who (or what) is doing the transacting. Its Financial OS is designed not for a new UI—but for a new user type. One that doesn't need screens, clicks, or keyboards to drive commerce. This shift is massive. It raises important questions about trust, regulation, and control—but it also offers new possibilities for scale, efficiency, and accessibility. For a company that pioneered online payments, PayPal is once again leading a new wave of digital transformation. Final Take For The PayPal Dev Conference With this bold move, PayPal isn't just riding the AI wave—it's helping shape it. By releasing open tools, supporting intelligent protocols, and activating a global network of agents, PayPal is betting that the next trillion-dollar economy won't be driven by clicks. It'll be driven by AI agents. The question is—are you building for them yet? Did you enjoy this story about PayPal and AI Agents? Don't miss my next one: Use the blue follow button at the top of the article near my byline to follow more of my work.
Forbes
30-04-2025
- Business
- Forbes
AI As Your Next CTO? The Future Of Autonomous Software Engineering
Amy Gu, CEO and Cofounder at Dynamsoft. getty With the near-universal adoption of AI coding tools, are we hurtling toward a future where autonomous coding agents replace developers? As AI evolves from coding assistant to reshaping roles that tech leadership traditionally held, could AI ever function as CTO? What do tech leaders need to consider as the workforce evolves to maximize the gains but retain creativity, innovation and ethical governance? AI is no longer just a coding assistant. It is now capable of designing, optimizing and even deploying code. Early tools focused on automating repetitive tasks like syntax correction and autocompletion. Today's tools such as GitHub Copilot, OpenAI Codex and AWS CodeWhisperer can not only write and debug but also identify inefficiencies and suggest architectural improvements. AI testing tools can not only generate test cases but also detect vulnerabilities and compliance issues much earlier in the development cycle. We've seen accelerated growth of "self-learning software engineering" tools in recent years, driven by more powerful and context-aware large language models (LLMs), advances in natural language understanding and widespread adoption of AI development assistants. In contrast to earlier rule-based tools, these tools continuously refine their output by learning from billions of lines of code, developer interactions and usage patterns. These tools are context-aware and personalized—meaning they adapt and improve over time based on your codebase, coding style and team practices. Traditional development follows a reactive approach—code, debug and iterate. With AI, we are shifting toward a proactive development model—continuously optimizing and anticipating issues such as bottlenecks or security vulnerabilities before they become critical issues. This switch in paradigm means teams can build more resilient, future-proof software. The next step in AI-driven development is AI agents capable of full-cycle development—from planning to writing to testing and deploying code—with minimal human intervention. These agents go beyond simply assisting developers and act as digital team members. Tools like AutoGPT, Cognition Labs' Devin and Devika are designed to take initiative and work independently from start to finish. While self-learning tools boost developer productivity, autonomous agents reshape the developer role altogether. As AI takes on more of the traditional coding roles, developers will need to shift toward strategic problem-solvers, customer champions and domain experts. Keeping the "human in the loop" will become increasingly important to ensure creativity, customer-centric design and ethical oversight. Teams will need to rebalance from traditional developers to more AI engineers, prompt engineers and AI trainers. With a dramatically reduced time to market, product managers will have more time to gather customer insights, prototyping and iterating to ensure product-market fit. The key to tech leadership success will be balancing efficiency with creativity, keeping control over critical decisions and ensuring ethical oversight. The role of CTO will shift toward product vision, ethics and business alignment. Build an AI-first development culture and upskill developers to work alongside AI tools. As junior coding and test roles diminish, evolve toward more senior full-stack technical roles, UX experts and product managers. Consider product managers as mini-CEOs with a broader view of the vision, road map and business plan for their products. Use the efficiency gains of AI to broaden testing. • Shift-Left: Test earlier for compliance, risk and accessibility in parallel to development. • Shift-Right: Test later to ensure performance in real-world conditions. What level of autonomy does it offer? Can decisions be audited? How will it integrate into current DevOps? What are the data privacy, compliance and security implications? Will it scale with architectural and business goals? Avoid the "black box dilemma" by ensuring transparency in AI decision making and auditability of AI-generated code. Make explainable AI (XAI) a priority. Introduce processes and checkpoints to identify bias, security vulnerabilities and accountability in AI-driven coding. Ensure compliance with emerging global AI regulations such as the EU AI Act and any U.S. AI policies. The question is not whether AI will replace CTOs but how it can amplify the impact of strong tech leadership. AI is less of a replacement and more of a multiplier—extending team capabilities, accelerating development life cycles and freeing up human thinking for higher-level problem-solving and innovation. Just as cloud computing redefined infrastructure and deployment, AI is redefining software development—automating manual tasks, predicting rather than reacting and uncovering hidden insights. The most successful CTOs will be those who can reshape their development culture, orchestrate a new workforce combining human and AI roles and ensure transparency and ethical stewardship. Those who can proactively adapt and leverage AI will replace those who don't. Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Associated Press
14-04-2025
- Automotive
- Associated Press
ECARX Unveils ECARXperience: Revolutionizing In-Vehicle Interaction with Adaptive AI-Powered Driving Experience
SHANGHAI, April 14, 2025 (GLOBE NEWSWIRE) -- ECARX Holdings Inc. (Nasdaq: ECX) ('ECARX' or the 'Company'), a global mobility tech provider, today announced the launch of ECARXperience, an advanced generative in-vehicle HMI system powered by its proprietary AutoGPT AI large model application, designed to transform the driving experience with a dynamic, generative interface that evolves in real-time to address driver needs, delivering smarter, safer, and more personalized interactions. ECARXperience simplifies and enhances how drivers and passengers interact with vehicles by introducing an adaptive interface that adjusts in real-time to driving conditions and user preferences. The system customizes displays based on factors like weather, time of day, and road conditions while proactively improving safety and comfort with features such as fatigue detection, child monitoring, and tailored route recommendations. It seamlessly integrates navigation, safety systems, and advanced driver assistance into one cohesive interface, making driving smarter and more intuitive. With emotionally intelligent virtual assistants that interpret voice tone and gestures, ECARXperience delivers natural interactions that prioritize user needs. Together, these features transform the driving experience from passive functionality to active engagement, setting a new standard for intelligent cockpits. Mr. Ziyu Shen, Chairman and CEO of ECARX, commented, 'With ECARXperience, we are redefining what it means to interact with a vehicle. By harnessing the power of AutoGPT, this innovation is designed to scale seamlessly across models and brands, enabling automakers to offer differentiated and personalized driving experience in a highly competitive market. As part of our long-term R&D roadmap, we will integrate ECARXperience into our broader technology solutions matrix, helping global vehicle brands remain at the forefront of technological change and driving the sustainable development of the automotive industry.' About ECARX ECARX (Nasdaq: ECX) is a global automotive technology provider with capabilities to deliver turnkey solutions for next-generation smart vehicles, from the system on a chip (SoC), to central computing platforms, and software. As automakers develop new electric vehicle architectures from the ground up, ECARX is developing full-stack solutions to enhance the user experience, while reducing complexity and cost. Founded in 2017 and listed on the Nasdaq in 2022, ECARX now has over 1,900 employees based in 12 major locations in China, UK, USA, Sweden and Germany. The co-founders are two automotive entrepreneurs, Chairman and CEO Ziyu Shen, and Eric Li (Li Shufu), who is also the founder and chairman of Zhejiang Geely Holding Group — with ownership interests in global brands including Lotus, Lynk & Co, Geely Galaxy, Polestar, smart, and Volvo Cars. ECARX also works with other well-known automakers, including Volkswagen Group, FAW and Dongfeng Peugeot-Citroën. To date, ECARX products can be found in over 8.1 million vehicles worldwide. Forward-Looking Statements This release contains statements that are forward-looking statements within the meaning of the U.S. Private Securities Litigation Reform Act of 1995. These statements are based on management's beliefs and expectations as well as on assumptions made by and data currently available to management, appear in a number of places throughout this document and include statements regarding, amongst other things, results of operations, financial condition, liquidity, prospects, growth, strategies, and the industry in which we operate. The use of words 'expects,' 'intends,' 'anticipates,' 'estimates,' 'predicts,' 'believes,' 'should,' 'potential,' 'may,' 'preliminary,' 'forecast,' 'objective,' 'plan,' or 'target,' and other similar expressions are intended to identify forward-looking statements. These forward-looking statements are not guarantees of future performance and are subject to a number of risks and uncertainties that could cause actual results to differ materially, including, but not limited to statements regarding our intentions, beliefs, or current expectations concerning, among other things, results of operations, financial condition, liquidity, prospects, growth, strategies, future market conditions or economic performance and developments in the capital and credit markets and expected future financial performance, and the markets in which we operate. For a discussion of these and other risks and uncertainties that could cause actual results to differ materially from those expressed in any forward-looking statement, see ECARX's filings with the U.S. Securities and Exchange Commission. ECARX undertakes no obligation to update or revise forward-looking statements to reflect subsequent events or circumstances, except as required by applicable law. Investor Contacts: [email protected] Media Contacts: [email protected]
Forbes
11-04-2025
- Business
- Forbes
The AI Coordination Revolution You Haven't Heard About Yet
Coordinated, orchestrated and ready to go. Breakthroughs don't always arrive with a bang or a flash of inspiration. Sometimes, they come in the form of quietly made connections, wires carefully laid between sparks we had already had. The AI revolution has given us no shortage of sparks. So many, in fact, that it's easy to lose count. It started with large language models, unlocking the power of general-purpose text generation. Just as we were getting comfortable with prompting, another shift appeared: AI agents, built not just to respond, but to take action on our behalf. From customer service to code generation, we saw tools like LangChain and AutoGPT explode in use cases, with one company after another dashing to find use for their Agents. But we quickly ran into a wall: these agents didn't talk to each other. They acted alone without coordination, like ships passing in the night. Given how rapidly things have evolved with AI it would be a fool's errand to try to predict the future, but one thing is for sure: the next frontier in AI isn't going to be as simple as creating increasingly smart agents. Instead, it will require unlocking their ability to collaborate, just like humans do. And that's where agentic orchestration comes in. Remember how just a few years ago autocomplete felt magical? Now, instead of making guesses at the next word we have AI agents drafting our emails, optimizing our workflows, and handling customer engagements for us. This acceleration from prediction to action has been stunning, and it has happened faster than most people get promoted on the job. Just like we've reached the limits of training data available for foundational model providers, we're also getting closer to reaching the limits of siloed agents. Most enterprises now face a growing web of tools with the average enterprise having more than 360 SaaS applications deployed at any given time. Although each promises to be a part of the digital transformation, few work in concert to actually make life easier for clients. The same holds for the agentic revolution as well, where the sum of the parts is becoming a morass clients can find difficult to move forward effectively in. Enterprises are building agents on different platforms ranging from LangChain, Salesforce, Anthropic to OpenAI and the challenge isn't the quality of each, it's their lack of cohesion and ability to coordinate. 'Many of our clients come with agents,' says Matt Wood, who helps lead PwC's AI efforts. "They are already seeing results, and next they want them organized. There's so much breadth in how clients work with these. Once we got structure in place, plans, playbooks, the ability to iterate, that's when we saw 10x improvement in how fast and well agents could be built.' This is where agentic orchestration comes in. The concept is exactly what it sounds like: creating the conditions for many agents to collaborate toward a shared goal. Instead of a single AI doing a single task, it's multiple AI systems coordinating across departments, domains, and decisions. PwC's new agent OS is designed to do just that, and it is the first mover in a space that is likely to see many entrants in the near future. What solutions like PwC's agent OS do is connect agents, regardless of their tech stack, into structured, adaptive workflows. Then, it lets them talk to each other, exchanging information about the task and goals, all the while the humans remain on the loop managing it all as necessary. When done right, orchestration sets boundaries and limits on the AI, allowing them to focus, if not specialize almost as if they had read Adam Smith's example of the pin factory. 'We found that the more constrained the agents are, the better they perform,' says Wood. 'Agents do better when they are asked to go on deliberate expeditions with a clear goal instead of exploration. Once we shifted our mindset from building agents to orchestrating them, we built 250+ across every function. And more importantly, they worked together.' The results aren't just theoretical. As Paul Griggs, PwC's US Senior Partner, puts it: ' we're helping organizations scale AI agents with confidence, speed, and purpose and it's a real game-changer for how our clients' work gets done.' Clearly there's great power in coordination between agents, and there's space for humans to pitch in too. Much of the hype around AI is focused on the rather natural, albeit somewhat overblown, fears of AI replacing humans. For those who are ready to welcome our robotic overlords and submit to a life of involuntary leisure, Paul offers reassurance that us humans still have our role to play. 'There's no question whether the future of work is in human-AI collaboration.' To understand what the future of work will look like, it helps to understand the evolving relationship between humans and machines. Overall, there will be three main modes of interacting with agentic AI: These distinctions are more than academic, but getting the nuances across to an audience that is still coming to terms with AI is its own challenge. 'You see peoples' eyes gloss over when you use terms such as 'multi-agent orchestration,'' says IBM's Karl Haller. 'But if you show them how their Standard Operating Procedures (SOPs) can be turned into code, and the efficiency improvements that come from that, they get it pretty fast.' Haller sees the real opportunity of orchestration in codifying business logic. 'The future of AI is where your Visio diagram becomes executable code. It's where the question, answer, and action are all linked. That's when an agent goes from being a tool to being a teammate.' It's no wonder that every major firm seems to be building some flavor of orchestration layer. But when you ask around, as analyst Usman Sheikh, Managing Director at High Output Ventures, seeing the full picture remains difficult. 'There's a flurry of announcements,' Sheikh tells us. 'But when I ask people at these firms to provide actual case studies or examples they are far and few between. The marketing is often ahead of the implementation.' This insight brings us back to a key point in the AI revolution that many companies are still grappling with: trust. 'We get a lot of questions about agents and whether they can be trusted to do the work right,' says Matt Wood. 'Orchestration with the right guardrails and the knowledge sources connected is what makes the difference. That's what scales, because users can see that it's trustworthy instead of companies just marketing it as such.' If an AI agent books your flight, handles your taxes, or flags a compliance issue, you need to know it's acting in alignment with enterprise standards. That's why baking in governance and guardrails will be part-and-parcel in making any orchestration system work. 'We made a priority of governance with PwC's agent OS', Wood explains. 'It integrates with our risk frameworks, ensuring oversight without friction, and we've opened up the black box of AI as much as possible. Non-technical teams can use drag-and-drop interfaces, agents can be edited, updated, and monitored without a PhD.' A single agent is a tool. A fleet of agents, orchestrated with precision and aligned to outcomes, that's infrastructure. And for organizations ready to move beyond the hype and into real AI transformation, the future is already arriving. It's agentic. It's orchestrated. And it's ready to push us humans consistently ahead of the curve.
Associated Press
11-03-2025
- Automotive
- Associated Press
ECARX Announces Fourth Quarter and Full Year 2024 Unaudited Financial Results
SHANGHAI, March 11, 2025 (GLOBE NEWSWIRE) -- ECARX Holdings Inc. (Nasdaq: ECX) ('ECARX' or the 'Company'), a global mobility tech provider, today announced unaudited financial results for the quarter and full year ended December 31, 2024. Ziyu Shen, ECARX Chairman and CEO, commented, '2024 was a remarkable year for ECARX, marked by significant milestones and progress across our business. We continued to execute on our strategic vision, further strengthening the unique value proposition we offer global automakers who are seeking cutting-edge, cost-effective solutions to distinguish themselves from the competition. We are capitalizing on this growing demand with our innovative product portfolio, diverse customer base, and strategic global partnerships. This is clearly reflected in our strong performance for full year 2024, with revenue increasing 18% year-over-year and our operating loss consistently narrowing. As a result, we hit breakeven at EBITDA level during the fourth quarter, which strengthens our confidence in reaching full-year breakeven in 2025.' 'Total shipments reached a record high of 2 million last year, up 33% year-over-year, with over 700 thousand shipped during the fourth quarter alone. Growth during the quarter was driven by robust demand for multiple Geely models powered by our solutions. Geely recently launched Galaxy EX5, the first overseas model integrating the Antora 1000 computing platform and demonstrating our ability to deliver this solution for both the PRC and international markets. With the addition of Volkswagen Group, our global customer base has now expanded to 18 automakers across 28 brands. This project win is a milestone for us and our solutions, which will be deployed in vehicles across EMEA and the Americas, expanding our global reach and demonstrating to a broader audience how our full-stack solutions can be customized for deployment in different markets. Additionally, we integrated DeepSeek into our AutoGPT in-vehicle AI large language model during the quarter to further enhance the in-vehicle experience and keep our automaker partners at the forefront of a rapidly evolving technological landscape. As we look ahead, we remain optimistic about the opportunities presented by the rise of software-defined vehicles. The US$20 million share repurchase program we announced late last year underscores this optimism and reflects our commitment to delivering long-term value for our shareholders. We are also proactively exploring capital-raising opportunities, including through equity financing, to support our strategic objectives. As a part of this effort, we have recently filed a registration statement on Form F-3, allowing us to swiftly capitalize on favorable market conditions as they arise.' Fourth Quarter 2024 Financial Results: Total revenue was RMB1,940.7 million (US$265.9 million), up 4% year-over-year ('YoY'). Sales of goods revenue was RMB1,524.9 million (US$208.9 million), up 16% YoY, the increase in sales revenue was mainly attributable to a RMB540 million increases in the sales volume of automotive computing platform products, primarily driven by an increase in the sales volume of Antora series and Makalu platform digital cockpits, offset by a RMB309 million decline from changes in the per unit price. Additionally, there was a RMB55 million increase from SoC core modules unit price changes, offset by RMB27 million and RMB46 million decline from decreased sales volume of SoC modules and automotive merchandise and other products, respectively. Software license revenue was RMB90.2 million (US$12.4 million), down 3% YoY, primarily attributable to a decrease in the sales volume of navigation and operating software compared to the same period last year. Service revenue was RMB325.6 million (US$44.6 million), down 31% YoY, principally as a result of a decrease in the total value of design and development contracts for automotive computing platforms completed during the fourth quarter compared to the same period last year. Total cost of revenue was RMB1,529.8 million (US$209.5 million), up 6% YoY, primarily driven by an increase in the sales volume of automotive computing platform products, partially offset by a decrease resulting from reduced design and development activities relating to contracts for automotive computing platforms. Gross profit was RMB410.9 million (US$56.4 million), down 4% YoY, which resulted in a gross margin of 21%. The decrease in gross margin was attributable to the penetration pricing strategy adopted to drive automotive computing platform revenue growth, and a shift in the overall revenue mix compared to the same period last year. Research and development expenses were RMB342.1 million (US$46.9 million), down 28% YoY, primarily driven by improved R&D efficiencies and synergies through the reallocation and integration of research and development resources. Selling, general and administrative expenses and others, net were RMB157.0 million (US$21.5 million), down 39% YoY, primarily attributable to improved global operating efficiencies and lower share-based compensation expenses incurred during the fourth quarter of 2024. Net loss was RMB39.5 million (US$5.3 million) significantly down compared with RMB326.7 million during the same period last year, primarily attributable to the reduction in total operating expenses, as well as a gain of RMB130.0 million from partial sale of an equity investment. Adjusted EBITDA (non-GAAP) gain was RMB74.4 million (US$10.3 million), compared with adjusted EBITDA (non-GAAP) loss of RMB236.2 million in the same period last year. For more information on the non-GAAP financial measure, please see 'Non-GAAP Financial Measure' and the table captioned 'Unaudited Reconciliation of GAAP and Non-GAAP Results.' Total cash as of December 31, 2024 was RMB367.4 million (US$50.3 million) including RMB43.4 million of restricted cash. Full Year 2024 Financial Results: Total revenue was RMB5,561.3 million (US$761.9 million), up 18% compared to RMB4,699.6 million in 2023. Sales of goods revenue was RMB4,405.5 million (US$603.6 million), up 33% compared to RMB3,311.5 million in 2023, the increase in sales revenue was primarily due to a RMB1,358 million increase in the sales volume of automotive computing platform products, primarily driven by an increase in the sales volume of digital cockpits with Geely Auto and Geely ecosystem brand-penetration, and the ramping up of autonomous driving control unit (ADCU) sales volume, partially offset by a RMB224 million decrease from changes in the per unit price. Additionally, there was a RMB132 million increase from SoC core modules unit price changes, offset by a RMB96 million and RMB75 million decline from decreased sales volume of SoC modules and automotive merchandise and other products, respectively. Software license revenue was RMB306.0 million (US$41.9 million), down 31% compared to RMB444.8 million in 2023, primarily driven by a decrease in the sales volume of navigation and operating software and a decrease in intellectual property license revenue. Service revenue was RMB849.8 million (US$116.4 million), down 10% compared to RMB943.3 million in 2023, principally as a result of a decrease in the total value of design and development contracts for automotive computing platforms completed during the year. Total cost of revenue was RMB4,407.1 million (US$603.8 million), up 29% compared to RMB3,427.0 million in 2023, primarily driven by an increase in sales volume of automotive computing platform products. The increased sales volume contributed to the increase in the cost of goods sold. The decrease in cost of services was mainly impacted by the decrease in service revenue. Gross profit was RMB1,154.2 million (US$158.1 million), down 9% compared to RMB1,272.6 million in 2023, representing a gross margin of 21% (compared to 27% in 2023). Research and development expenses were RMB1,259.7 million (US$172.6 million), down 0.4% compared to RMB1,264.3 million in 2023. Selling, general and administrative expenses and others, net were RMB776.7 million (US$106.4 million), down 16% compared to RMB926.1 million in 2023, primarily driven by improved global operating efficiencies and lower share-based compensation expenses incurred during the year. Net loss of RMB989.9 million (US$135.6 million), down 3% compared to RMB1,015.6 million in 2023, primarily attributable to the reduction in total operating expenses and the gain from partial sale of an equity investment during 2024, partially offset by the decrease in gross margin attributable to the penetration pricing strategy implemented to stimulate revenue growth, as well as an increase in interest expense. Adjusted EBITDA (non-GAAP) loss of RMB591.6 million (US$81.0 million), a 17% improvement from Adjusted EBITDA (non-GAAP) loss of RMB710.6 million in 2023. Fourth Quarter 2024 and Recent Business Highlights: Expanding Global Footprint and International Partnerships Over 8.1 million vehicles on the road incorporating ECARX technology as of December 31, 2024. Total shipments reached a record high of 2 million units in 2024, an increase of 33% from 2023. Nomination awarded from Volkswagen Group for multiple vehicles under the Volkswagen and Skoda brands through its Global Entry Infotainment initiative set to launch internationally throughout 2027 to 2028. Secured 3 new design wins, with two deploying the Galena computing platform and one deploying the Makalu computing platform. Fuyang smart production facility continues to ramp up capacity, with 60,000 Antora computing platform units produced in December 2024. Technological Advancements and Product Updates Galaxy EX5, the overseas version of the Galaxy E5 in China, was launched by Geely in early 2025, being the first overseas vehicle to deploy the Antora 1000 computing platform. ECARX Skyland Pro ADAS solution has been integrated into Geely's G-Pilot unified intelligent driving system, which was recently launched in Galaxy E8 and Xingyao 8. Hongqi Tiangong 05 was launched in February 2025, deploying the first intelligent cockpit jointly-developed under the Company's strategic partnership with FAW Group. Lynk & Co Z10 STARBUFF was launched in November 2024, integrating the Makalu computing platform and providing users with an immersive gaming experience to play triple-A games on the go. Spotlighted the power and versatility of innovative ECARX solutions including the AutoGPT in-vehicle AI large language model application that was later integrated with DeepSeek, ADAS technologies, and the Galena, Antora and Makalu computing platforms at CES 2025. ECARX Cloudpeak hypervisor, a core component of the Cloudpeak software stack, received ISO 26262 ASIL-D certification in January 2025. Showcased the Skyland Pro ADAS solution at AutoSens China in November 2024, demonstrating how it empowers automakers with high-performance, cost-effective intelligent driving solutions. Represented the auto sector at the Saudi Arabia National Quality Conference in November 2024 and received the SASO Award for contributions to generative AI quality assurance, a reflection of the impact that AutoGPT is having on the industry. This press release contains certain forecast or projection which represents the current and preliminary view of the Company's management and is therefore subject to changes and uncertainties. See 'Forward-Looking Statements.' Conference Call and Webcast Details ECARX will host a webcast of its earnings conference call today, Tuesday, March 11, 2025, at 8:00 a.m. EDT. To access the webcast, visit the News and Events section of the ECARX Investor Relations website, or visit the following link – To join the earnings call by telephone, participants must preregister at A replay of the webcast and presentation materials will be available on the Company's Investor Relations website under the results and reports section following the event. About ECARX ECARX (Nasdaq: ECX) is a global automotive technology provider with capabilities to deliver turnkey solutions for next-generation smart vehicles, from the system on a chip (SoC), to central computing platforms, and software. As automakers develop new electric vehicle architectures from the ground up, ECARX is developing full-stack solutions to enhance the user experience, while reducing complexity and cost. Founded in 2017 and listed on the Nasdaq in 2022, ECARX now has over 1,900 employees based in 12 major locations in China, UK, USA, Sweden, Germany and Malaysia. The co-founders are two automotive entrepreneurs, Chairman and CEO Ziyu Shen, and Eric Li (Li Shufu), who is also the founder and chairman of Zhejiang Geely Holding Group — with ownership interests in global brands including Lotus, Lynk & Co, Geely Galaxy, Polestar, smart, and Volvo Cars. ECARX also works with other well-known automakers, including Volkswagen Group, FAW and Dongfeng Peugeot-Citroën. To date, ECARX products can be found in over 8.1 million vehicles worldwide. Forward-Looking Statements This release contains statements that are forward-looking statements within the meaning of the U.S. Private Securities Litigation Reform Act of 1995. These statements are based on management's beliefs and expectations as well as on assumptions made by and data currently available to management, appear in a number of places throughout this document and include statements regarding, amongst other things, results of operations, financial condition, liquidity, prospects, growth, strategies, and the industry in which we operate. The use of words 'expects,' 'intends,' 'anticipates,' 'estimates,' 'predicts,' 'believes,' 'should,' 'potential,' 'may,' 'preliminary,' 'forecast,' 'objective,' 'plan,' or 'target,' and other similar expressions are intended to identify forward-looking statements. These forward-looking statements are not guarantees of future performance and are subject to a number of risks and uncertainties that could cause actual results to differ materially, including, but not limited to statements regarding our intentions, beliefs, or current expectations concerning, among other things, results of operations, financial condition, liquidity, prospects, growth, strategies, future market conditions or economic performance and developments in the capital and credit markets and expected future financial performance, and the markets in which we operate. For a discussion of these and other risks and uncertainties that could cause actual results to differ materially from those expressed in any forward-looking statement, see ECARX's filings with the U.S. Securities and Exchange Commission. ECARX undertakes no obligation to update or revise forward-looking statements to reflect subsequent events or circumstances, except as required by applicable law. Translation of results into U.S. dollars This announcement contains translations of certain Renminbi (RMB) amounts into U.S. dollars (US$) at a specified rate solely for the convenience of the reader. Unless otherwise noted, the translation of RMB into US$ has been made at RMB7.2993 to US$1.00, the noon buying rate in effect on December 31, 2024 as set forth in the H.10 Statistical Release of The Board of Governors of the Federal Reserve System. We make no representation that any Renminbi or U.S. dollar amounts could have been, or could be, converted into U.S. dollars or Renminbi, as the case may be, at any particular rate, or at all. Non-GAAP Financial Measure The Company uses adjusted EBITDA (non-GAAP) in evaluating its operating results and for financial and operational decision-making purposes. Adjusted EBITDA is defined as net loss excluding interest income, interest expense, income tax expense, depreciation of property and equipment, amortization of intangible assets, and share-based compensation expenses. The Company presents this non-GAAP financial measure because it is used by the management to evaluate the Company's operating performance and formulate business plans. The Company believes that the non-GAAP measure helps identify underlying trends in its business that could otherwise be distorted by the effects of certain expenses that are included in net loss. The Company also believes that the use of the non-GAAP measure facilitates investors' assessment of its operating performance. Adjusted EBITDA (non-GAAP) should not be considered in isolation or construed as alternatives to net loss or any other measures of performance or as indicators of the Company's operating performance. Investors are encouraged to compare the Company's historical adjusted EBITDA (non-GAAP) to the most directly comparable GAAP measure, net loss. Adjusted EBITDA (non-GAAP) presented here may not be comparable to similarly titled measures presented by other companies. Other companies may calculate similarly titled measures differently, limiting their usefulness as comparative measures to the Company's data. The Company encourages investors and others to review the financial information in its entirety and not rely on a single financial measure. For more information on the non-GAAP financial measure, please see the table captioned 'Unaudited Reconciliation of GAAP and Non-GAAP Results' set forth at the end of this press release. Investor Contacts: Rene Du Media Contacts: [email protected] ECARX Holdings Inc. Unaudited Condensed Consolidated Balance Sheets As of December 31, 2023 As of December 31, 2024 Millions, except otherwise noted RMB RMB USD ASSETS Current assets Cash 571.8 324.0 44.4 Restricted cash 27.1 43.4 5.9 Short-term investments 137.9 130.5 17.9 Accounts receivable – third parties, net 285.8 221.1 30.3 Accounts receivable – related parties, net 1,572.7 1,373.8 188.2 Notes receivable 54.6 16.8 2.3 Inventories 160.8 233.9 32.0 Amounts due from related parties 74.1 35.4 4.8 Prepayments and other current assets 443.6 452.5 62.1 Total current assets 3,328.4 2,831.4 387.9 Non-current assets Long-term investments 301.0 15.8 2.2 Operating lease right-of-use assets 125.2 132.7 18.2 Property and equipment, net 120.8 160.3 22.0 Intangible assets, net 179.3 309.8 42.4 Goodwill — 25.7 3.5 Other non-current assets – third parties 28.2 28.3 3.9 Other non-current assets – related parties 224.3 267.3 36.6 Total non-current assets 978.8 939.9 128.8 Total assets 4,307.2 3,771.3 516.7 LIABILITIES Current liabilities Short-term borrowings 1,200.0 1,360.0 186.3 Accounts payable - third parties 1,820.7 1,617.4 221.6 Accounts payable - related parties 312.8 512.6 70.2 Notes payable 10.0 142.0 19.5 Amounts due to related parties 35.7 177.9 24.4 Contract liabilities, current - third parties 0.6 6.4 0.9 Contract liabilities, current - related parties 207.0 150.5 20.6 Operating lease liabilities, current 35.1 40.9 5.6 Convertible notes payable - current — 470.6 64.5 Accrued expenses and other current liabilities 614.5 626.6 85.8 Income tax payable 15.8 20.4 2.8 Total current liabilities 4,252.2 5,125.3 702.2 Non-current liabilities Contract liabilities, non-current - related parties 134.0 37.3 5.1 Convertible notes payable, non-current 455.7 — — Operating lease liabilities, non-current 107.6 121.4 16.6 Warrant liabilities, non-current 5.1 8.8 1.2 Provisions 90.9 110.1 15.1 Other non-current liabilities - third parties 48.8 98.0 13.4 Other non-current liabilities - related parties 44.5 — — Deferred tax liabilities — 15.2 2.1 Total non-current liabilities 886.6 390.8 53.5 Total liabilities 5,138.8 5,516.1 755.7 SHAREHOLDERS' DEFICIT Ordinary Shares — — — Additional paid-in capital 6,096.7 6,214.3 851.4 Treasury Shares, at cost — (7.4) (1.0) Accumulated deficit (6,670.7) (7,603.0) (1,041.6) Accumulated other comprehensive loss (344.6) (363.5) (49.8) Total deficit attributable to ordinary shareholders (918.6) (1,759.6) (241.0) Non-redeemable non-controlling interests 87.0 14.8 2.0 Total shareholders' deficit (831.6) (1,744.8) (239.0) Liabilities and shareholders' deficit 4,307.2 3,771.3 516.7 ECARX Holdings Inc. Unaudited Condensed Consolidated Statement of Comprehensive Loss Three Months Ended December 31 Full Year Ended December 31 2023 2024 2024 2023 2024 2024 Millions, except share data and per share data, or otherwise noted RMB RMB USD RMB RMB USD Revenue Sales of goods revenue 1,313.0 1,524.9 208.9 3,311.5 4,405.5 603.6 Software license revenue 92.6 90.2 12.4 444.8 306.0 41.9 Service revenue 468.8 325.6 44.6 943.3 849.8 116.4 Total revenue 1,874.4 1,940.7 265.9 4,699.6 5,561.3 761.9 Cost of goods sold (1,167.6) (1,367.2) (187.3) (2,734.0) (3,874.8) (530.8) Cost of software licenses (21.8) (50.7) (6.9) (120.3) (128.2) (17.6) Cost of services (255.4) (111.9) (15.3) (572.7) (404.1) (55.4) Total cost of revenue (1,444.8) (1,529.8) (209.5) (3,427.0) (4,407.1) (603.8) Gross profit 429.6 410.9 56.4 1,272.6 1,154.2 158.1 Research and development expenses (473.4) (342.1) (46.9) (1,264.3) (1,259.7) (172.6) Selling, general and administrative expenses and others, net (257.1) (157.0) (21.5) (926.1) (776.7) (106.4) Total operating expenses (730.5) (499.1) (68.4) (2,190.4) (2,036.4) (279.0) Loss from operation (300.9) (88.2) (12.0) (917.8) (882.2) (120.9) Interest income 7.6 5.7 0.8 30.5 22.2 3.0 Interest expense (21.2) (60.4) (8.3) (79.3) (133.8) (18.3) Share of results of equity method investments (7.4) 116.5 16.0 (43.1) 40.3 5.5 Foreign currency exchange gains/(losses) 3.5 (4.0) (0.5) (10.6) (7.6) (1.0) Others, net (12.2) (8.5) (1.2) 1.1 (27.3) (3.7) Loss before income taxes (330.6) (38.9) (5.2) (1,019.2) (988.4) (135.4) Income tax benefit/(expense) 3.9 (0.6) (0.1) 3.6 (1.5) (0.2) Net loss (326.7) (39.5) (5.3) (1,015.6) (989.9) (135.6) Net loss attributable to non-controlling interests 28.0 3.2 0.4 75.0 57.6 7.9 Net loss attributable to ECARX Holdings Inc. ordinary shareholders (298.7) (36.3) (4.9) (940.6) (932.3) (127.7) Net loss (326.7) (39.5) (5.3) (1,015.6) (989.9) (135.6) Other comprehensive loss: Foreign currency translation adjustments, net of nil income taxes 16.1 (10.4) (1.4) 41.3 (18.9) (2.6) Comprehensive loss (310.6) (49.9) (6.7) (974.3) (1,008.8) (138.2) Comprehensive loss attributable to non-redeemable non-controlling interests 28.0 3.2 0.4 75.0 57.6 7.9 Comprehensive loss attributable to ECARX Holdings Inc. (282.6) (46.7) (6.3) (899.3) (951.2) (130.3) Loss per ordinary share — Basic and diluted loss per share, ordinary shares (0.89) (0.11) (0.01) (2.79) (2.77) (0.38) Weighted average number of ordinary shares used in computing loss per ordinary share — Weighted average number of ordinary shares 337,442,347 333,819,732 337,407,225 336,641,846 Adjusted EBITDA We use adjusted EBITDA in evaluating our operating results and for financial and operational decision-making purposes. Adjusted EBITDA is defined as net loss excluding interest income, interest expense, income tax expense, depreciation of property and equipment, amortization of intangible assets, and share-based compensation expenses. Adjusted EBITDA should not be considered in isolation or construed as alternatives to net loss or any other measures of performance or as indicators of our operating performance. Investors are encouraged to compare our historical adjusted EBITDA to the most directly comparable GAAP measure, net loss. Adjusted EBITDA presented here may not be comparable to similarly titled measures presented by other companies. Other companies may calculate similarly titled measures differently, limiting their usefulness as comparative measures to our data. We encourage investors and others to review our financial information in its entirety and not rely on a single financial measure. ECARX Holdings Inc. Unaudited Reconciliation of GAAP and Non-GAAP Results Three Months Ended December 31 Full Year Ended December 31 2023 2024 2024 2023 2024 2024 Millions, except otherwise noted RMB RMB USD RMB RMB USD Net Loss (326.7) (39.5) (5.3) (1,015.6) (989.9) (135.6) Interest income (7.6) (5.7) (0.8) (30.5) (22.2) (3.0) Interest expense 21.2 60.4 8.3 79.3 133.8 18.3 Income tax (benefit)/expense (3.9) 0.6 0.1 (3.6) 1.5 0.2 Depreciation of property and equipment 14.3 15.9 2.2 54.0 56.8 7.8 Amortization of intangible assets 14.0 23.7 3.2 31.8 91.5 12.5 Share-based compensation expenses 52.5 19.0 2.6 174.0 136.9 18.8 Adjusted EBITDA (236.2) 74.4 10.3 (710.6) (591.6) (81.0)
