According to Woofun AI, Figma’s (FIG.US) AI features have penetrated the workflows of 80% of its large customers, while Arm’s (ARM.US) second-quarter revenue hit a quarterly record of $1.44 billion, setting the industry tone. Despite significant adoption of AI technologies, Figma’s declining gross margins and the cost structure shifts underlying Arm’s business transformation together paint a complex picture of the early commercialization phase for AI infrastructure and application layers. Wall Street executives have openly acknowledged the difficulty in predicting long-term AI returns, and investors are now demanding that companies demonstrate how technological advancements translate into actual profits—not just rely on narrative-driven valuations.
Figma reported its third-quarter 2026 earnings, generating $391.4 million in revenue, a 43% year-over-year increase and $16.4 million above the upper end of prior guidance. The company raised its full-year revenue forecast to $1.497 billion to $1.499 billion and increased its non-GAAP operating profit forecast to $158 million to $163 million.
However, the market did not fully buy in, and the stock dropped by about 7% in after-hours trading. The company’s revenue guidance for the fourth quarter of $402 million to $404 million implies a growth rate slowing from 43% in the third quarter to approximately 33%, which is the direct cause of the post-market pressure. The market is now evaluating not just gross margins, but also the company’s transition to a lower growth pace. According to data compiled by Woofun AI, Figma’s GAAP gross margin for the third quarter was 80%, and its non-GAAP gross margin was 82%, down from 86% in the same period last year. The decline in gross margin was primarily driven by increased inference costs resulting from higher AI usage. The company is increasing the proportion of tasks handled by its proprietary models to improve cost efficiency, but has not disclosed specific cost savings. Figma has demonstrated that large customers are willing to adopt AI consistently; the next step is proving how much incremental profit these usage levels can generate.
If AI积分收入 grows faster than inference costs, Figma could establish a new revenue stream; otherwise, AI could become an ongoing service cost that pressures gross margins.
The usage data for AI products is the most noteworthy part of this earnings report. As of October 31, over 80% of paid customers with an annualized contract value exceeding $10,000 consumed AI credits weekly, and 65% used Figma Agent weekly. Overall usage of the Figma MCP Server grew threefold quarter-over-quarter in Q3, with write-back operations increasing fourfold. The company also launched the public beta of Code Layers in October, allowing users to build and edit code-based prototypes directly on the design canvas. These changes are redefining the boundaries of Figma’s product offerings.
Traditional design tools focus on helping people create interfaces more efficiently, while Agents, MCP, and Code Layers enable software to directly participate in design, code generation, and modification. Figma is striving to evolve from a design editor into a unified platform connecting design and development. The business model is also shifting: previously, Figma primarily generated revenue through seat-based subscriptions; now, AI credits introduce usage-based pricing. The public beta of Code Layers, launched in October, further blurs the line between design and development, with 80% of high-value customers not only using AI-assisted design but increasingly relying on AI for code-level interactions.
This shift from a “tool” to a “platform” has made Figma’s revenue structure more complex and raised the bar for cost control. Although the non-GAAP gross margin of 82% remains high, it has declined from the historical peak of 86%, indicating that cost pressures are emerging during the early stages of scaling AI applications.
Arm announced its second-quarter fiscal year 2027 results ended September 30, with revenue reaching $1.44 billion, a 27% year-over-year increase and the highest ever for the company’s second fiscal quarter. Licensing and other revenue amounted to $773 million, up 50% year-over-year; royalty revenue reached $670 million, up 8% year-over-year. Notably, data center royalty revenue more than doubled. Arm’s business is undergoing deeper transformation: while the company historically generated revenue primarily by licensing its architecture to chip designers, it now directly offers its own designed Arm AGI CPUs. The company stated that it has secured purchase orders from multiple customers extending through calendar year 2027, with initial commercial deliveries planned for the fourth quarter of fiscal year 2027. Management expects this product to contribute up to $2 billion in revenue combined across fiscal years 2027 and 2028. This is a forecast, not confirmed sales. Arm also disclosed that among major hyperscale cloud providers, the market share of Arm-based servers increased from 15% in 2024 to over 50% in 2025.
This statistic applies only to specific large cloud providers and cannot be directly interpreted as global server market share. These figures reflect the growing demand for AI infrastructure. While GPUs handle massive parallel computations, CPUs still manage general-purpose computing, task scheduling, and system administration. As AI agents need to run programs, invoke tools, and execute complex workflows, CPU performance and energy efficiency will also impact overall data center costs. For Arm, both opportunities and risks are expanding. Selling its own designed CPUs could generate higher revenue per customer than licensing its architecture, but it also means the company must assume risks related to chip delivery, supply chains, and product competition. Arm is transitioning from primarily selling chip design intellectual property to also participating directly in the chip product market.
At the Hong Kong International Financial Leaders Investment Summit, financial leaders including David Solomon, CEO of Goldman Sachs (GS.US), Peng Zhao, CEO of Citadel Securities, and Jim Coulter, founding partner of TPG (TPG.US), discussed the long-term development of AI. Comments noted that the most noteworthy aspect of the summit was not the optimistic outlook from financial institutions toward AI, but rather the public acknowledgment by multiple executives of significant uncertainty in long-term predictions. Solomon stated that short-term demand for computing power remains strong, but the long-term trajectory is difficult to determine.
Peng Zhao raised a more direct question: the pace of technological development has outstripped investment cycles, forcing capital markets to make massive investment decisions without reliable long-term forecasts. Coulter reminded investors that the experience of the dot-com bubble showed that even correctly identifying a technology that will change the world can still lead to financial losses if the wrong company is bet on. This reveals a structural contradiction in AI investing: data centers, chips, and power infrastructure require capital investments years in advance, while model costs, application methods, and ultimate business models can change within months. The investment horizon for infrastructure does not align with the pace of technological iteration.
At the welcome dinner on November 2, Chinese AI companies offered another perspective. During a closed-door dialogue on “Technology for Good,” moderated by Shen Nanpeng, Moonshot’s founder Yang Zhilin addressed the debate between open-source and closed-source models, discussed recursive self-improvement capabilities of models, and explored pathways for generalization into the physical world. Compared to simply increasing computational investment, embedding AI into existing industries may provide a more direct path to validating commercial value—Hong Kong’s Financial Secretary, Paul Chan, noted at the luncheon that factories, hospitals, and logistics networks on the Chinese mainland are fertile ground for AI applications. However, application value does not automatically translate into investment returns. Model providers, cloud computing companies, chip manufacturers, and application developers may all create value, yet profits are unlikely to be evenly distributed. This is currently the most critical investment question in the AI industry: while the market believes AI will transform the economy, it remains uncertain who will capture returns commensurate with their capital investment.
The combined market value of the five largest U.S. tech giants increased by nearly $800 billion in a single day. Following the Democrats' win of both chambers of Congress, shares of Amazon (AMZN.US), Apple (AAPL.US), Microsoft (MSFT.US), Alphabet (GOOGL.US), and Meta (META.US) rose collectively. Investors believe that tech giants with existing large-scale computing infrastructure may benefit if new data center projects face restrictions. Amazon rose approximately 6.1%, and Apple increased about 5.3%, contrasting with declines among some data center operators. With the election results clarified and oil prices plunging, the Dow Jones Index surged 1,569 points, or 3.03%, on November 4, while the S&P 500 rose about 1.6%. However, the energy sector declined 4.77% due to the sharp drop in oil prices, making it the worst-performing sector that day, with non-cyclicals and financials leading the gains.
Brent crude fell approximately 8.7% to $94.62 per barrel. The decline in energy prices eased market concerns about inflation, but conflicts in Iran and political shifts in the U.S. could still influence future movements. American voters have begun directly rejecting the expansion of AI data centers. In the 2026 midterm elections, voters in multiple regions supported candidates or policies restricting data center construction. Texas Democratic gubernatorial candidate Gina Hinojosa unexpectedly defeated incumbent Governor Abbott, becoming the state’s first Latino governor and the first Democratic governor since 1990. Her platform includes pausing approvals for new data centers, making her victory a focal point for the AI infrastructure industry. Issues surrounding data centers’ high electricity and water consumption, as well as their impact on residential electricity prices, are evolving from local disputes into national political issues.
Uber (UBER.US) has encountered developments in the autonomous driving sector, but its strategy is not one of contraction. Waymo Co-CEO Tekedra Mawakana has confirmed that the exclusive Robotaxi partnership with Uber in Austin and Atlanta will end in January 2028, transitioning to Waymo’s own app for ride requests—earlier than the original agreement’s scheduled end date of May 2028. Uber stated that launching a dedicated entry for Robotaxis within its own app “was not Uber’s choice,” and emphasized its continued collaborations with multiple autonomous driving companies: during its Q3 earnings call, the company noted that approximately 700 fully driverless vehicles will be operational by year-end, exceeding 2,000 by 2027. The real pressure stems from supply constraints and the core ride-hailing business. CEO Dara Khosrowshahi noted that elevated fuel prices and limited driver supply have hindered supply growth across multiple international markets; Q3 ride-hailing gross bookings reached $29.512 billion, up 18% year-over-year, but revenue fell 2% to $7.52 billion due primarily to a business model shift in the UK market. Overall, Q3 total gross bookings rose 20%, and non-GAAP EPS reached $0.88, up 35% year-over-year—both exceeding market expectations. The key to commercializing autonomous driving remains operational efficiency, not merely technological maturity.
California’s Proposition 40 proposed a one-time 5% wealth tax on residents with a net worth exceeding $1 billion; preliminary results show 55.6% of votes opposed, indicating the proposal is likely to fail. The proposition previously drew significant attention from Silicon Valley entrepreneurs and investors, with debate centering on the trade-offs between wealth taxation, capital migration, and local government finances.