OpenAIโ€™s latest research demonstrates that large language models can actively participate in the intricate process of semiconductor creation, turning a field once dominated by seasoned engineers into a collaborative effort between humans and algorithms. In August 2026 the organization disclosed how its frontier models accelerated the development of a custom inference processor named Jalapeรฑo, a chip built specifically to serve AI workloads with high efficiency. By feeding the model detailed specifications of the target architecture, engineers were able to let the system suggest refinements that shaved down power consumption, boosted performance, and trimmed silicon area without sacrificing functionality. This breakthrough signals a paradigm shift: instead of relying solely on human intuition and years of trialโ€‘andโ€‘error, AI can now generate concrete circuitโ€‘level ideas, test them through simulation, and propose revisions that are ready for tapeโ€‘out. The implication is profound for the entire semiconductor ecosystem, as it promises faster design cycles, lower engineering costs, and the possibility of exploring architectures that might have been deemed too risky or timeโ€‘consuming just a few years ago. These advances come at a time when the industry faces mounting pressure to deliver more compute per watt while navigating the physical limits of lithography and the rising costs of advanced process nodes. By integrating AI into the design flow, companies can tap into a knowledge base that spans countless design iterations, enabling rapid exploration of tradeโ€‘offs that would otherwise require extensive manual effort.

Early in 2025 OpenAI began a focused partnership with Advanced Micro Devices, providing its models to assist in the optimization of AMDโ€™s flagship GPU intellectual property. The goal was to push the powerโ€‘performanceโ€‘area triangle toward more favorable values, a metric that directly influences silicon yield and market competitiveness. Using an early version of the model, engineers observed that even modest modifications to the RTL description could yield measurable gains in clock frequency while keeping power budgets steady. Over the course of several months the model identified opportunities to share hardware resources across pipeline stages, to replace bulky combinational blocks with more compact alternatives, and to restructure data paths so that fewer registers were needed. These changes translated into a reduction of standardโ€‘cell area in critical blocks and a measurable drop in dynamic power. The collaboration demonstrated that AIโ€‘driven suggestions are not merely theoretical; they can be incorporated into productionโ€‘ready design flows, reviewed with existing electronic design automation tools, and validated during silicon bringโ€‘up. This early success laid the groundwork for the deeper investigations reported later in the year, where the same models were applied to a broader set of processor components and to thirdโ€‘party hardware. The partnership also opened the door to joint research on nextโ€‘generation compute units, positioning both companies to coโ€‘develop chips that could meet the demanding requirements of emerging AI services.

Within the Jalapeรฑo architecture the vector unit serves as the engine that executes parallel arithmetic operations, making it a natural target for AIโ€‘guided optimization. The model examined each component of the unit, from the floatingโ€‘point adder and multiplier pipelines to the conversion routines that translate between formats and the reduction logic that aggregates partial results. By exploring alternative logic families and by reโ€‘partitioning computations across pipeline stages, the system discovered ways to reuse existing hardware rather than instantiating dedicated blocks for each operation. For example, it suggested merging the floatingโ€‘point adder with a shared carry chain, thereby cutting the number of distinct transistor devices required. The model also proposed dividing longโ€‘latency multiplyโ€‘accumulate operations into smaller, overlapping subโ€‘operations, which reduced the need for large intermediate storage elements. These refinements collectively trimmed the standardโ€‘cell area of the vector unit by roughly thirteen percent, a gain that was later confirmed on silicon during the bringโ€‘up phase. In addition to area savings, the redesigns lowered the average power per operation, delivering a more energyโ€‘efficient core without compromising throughput. The efficiency gains also extended to the specialโ€‘function unit, where the model identified opportunities to fuse transcendental calculations with nearby arithmetic blocks, reducing instruction latency and freeing pipeline slots for other tasks.

The matrix unit, responsible for dense linearโ€‘algebra operations that underpin most AI inference workloads, presented another fertile ground for modelโ€‘driven redesign. By analyzing the floatingโ€‘point adder and multiplier structures, the system proposed a multiplier layout that required roughly fifteen percent less cell area after synthesis while halving the latency from two clock cycles to a single cycle. A parallel adder revision achieved a similar fifteen percent reduction in area while preserving the original twoโ€‘cycle latency, demonstrating that size and speed are not mutually exclusive. Moreover, the model reโ€‘engineered the lowโ€‘precision dotโ€‘product circuitry, consolidating several small adders into a unified structure and reโ€‘ordering pipeline stages to minimize data movement. These adjustments collectively shaved ten percent off the matrix unitโ€™s standardโ€‘cell footprint and further improved its energy efficiency. The ripple effect was evident across the entire chip, as smaller arithmetic blocks freed up routing resources for other functional units, contributing to a more balanced overall design. Additionally, the model suggested reโ€‘balancing the operand width of the dotโ€‘product unit to match the precision requirements of the targeted AI models, which eliminated unnecessary bits and further reduced power consumption without affecting numerical accuracy. These enhancements not only lowered the silicon area but also contributed to a measurable decrease in the chipโ€™s overall dynamic power, which is critical for maintaining acceptable thermal margins in highโ€‘throughput AI servers.

Overall, the AIโ€‘assisted design effort delivered quantifiable improvements across the Jalapeรฑo compute architecture. The vector unit saw a thirteen percent reduction in standardโ€‘cell area, while the matrix unit achieved a ten percent shrinkage, translating into a noticeable contraction of the chipโ€™s total die size. These percentages were derived from postโ€‘synthesis reports that compared the AIโ€‘proposed layouts against the baseline handโ€‘crafted versions. The silicon bringโ€‘up phase confirmed that the revised blocks behaved as expected, with no functional regressions observed during extensive testโ€‘pattern execution. Power measurements indicated a proportional drop in dynamic energy per operation, and timing analyses showed that the critical paths remained within the original performance envelope. Such validation is essential because theoretical area savings alone do not guarantee realโ€‘world benefits; the chip must meet its timing, power, and functionality targets in actual silicon. The successful tapeโ€‘out of Jalapeรฑo thus serves as a concrete proof point that large language models can contribute meaningfully to productionโ€‘grade chip design, moving beyond speculative experiments toward deployable results. The design team also reported that the accelerated development timeline, which shrank what would have been a yearโ€‘long optimization cycle into just five months, allowed the company to allocate resources more flexibly and respond faster to market demands for higher efficiency AI accelerators.

Building on the Jalapeรฑo experience, OpenAI entered a deeper collaboration with AMD to apply its models to the companyโ€™s flagship GPU intellectual property. By feeding the latest model variants the detailed RTL descriptions of the GPUโ€™s compute clusters, the team observed up to fourteen percent improvement in the powerโ€‘performanceโ€‘area metrics relative to previously heavily tuned baselines. The breakthrough came from rewriting portions of the SystemVerilog code that define the circuit topology, a task that traditionally requires many months of expert microarchitect effort. In less than five months the model generated alternative netlists that preserved functional behavior while reducing both transistor count and clockโ€‘toโ€‘clock latency. AMD incorporated several of these modifications into a design that has already progressed to the manufacturing stage, and it added additional changes to its production library for future silicon revisions. The partnership also enabled AMD to verify singleโ€‘digit percentage gains in PPA across key blocks of its upcoming nextโ€‘generation flagship CPU, demonstrating that the same methodology can be transferred across distinct product families. These results suggest that AI can act as a catalyst for accelerating IP development cycles, reducing timeโ€‘toโ€‘market and enabling smaller design teams to compete with larger vendors.

The model demonstrated an impressive capacity to reason about RTL at a level that rivals seasoned microarchitects, enabling it to rewrite circuits that execute arithmetic, control flow, or data movement with confidence. By leveraging Synopsys electronic design automation tools, the system automatically verified that each proposed alteration preserved the functional equivalence of the original block, while also quantifying the impact on power, performance, and area. This dualโ€‘objective verification allowed engineers to explore aggressive optimizations without fearing hidden bugs, a capability that historically required extensive manual simulation and formal checking. The modelโ€™s suggestions often involved subtle changes such as reโ€‘balancing carry chains, reโ€‘ordering pipeline stages, or consolidating adjacent logic cells, all of which were validated through synthesis and timing reports. Moreover, the ability to evaluate tradeโ€‘offs in a systematic way helped the design team prioritize modifications that delivered the highest return on investment, ensuring that the final tapeโ€‘out met stringent commercial targets. This systematic approach also facilitated the generation of comprehensive regression suites, enabling continuous verification as the design evolved and reducing the risk of lateโ€‘stage silicon errata that can be costly to resolve.

AMDโ€™s validation of the modelโ€‘generated designs extended beyond the GPU, reaching the upcoming flagship CPU architecture. Independent testing showed that key blocks of the new processor exhibited singleโ€‘digit percentage improvements in power, performance, and area, confirming that the AIโ€‘crafted RTL modifications translate across different compute units. These gains, while modest in isolation, collectively contribute to a more efficient silicon footprint and higher clockโ€‘frequency headroom, both of which are critical for meeting the performance expectations of nextโ€‘generation dataโ€‘center and edge workloads. Looking ahead, the collaboration anticipates further refinements that could push PPA improvements beyond the current singleโ€‘digit range, especially as the model continues to mature and gain exposure to more sophisticated design constraints. The expectation is that future iterations will enable tighter integration between AI suggestions and the advanced layoutโ€‘aware tools used by semiconductor foundries, thereby unlocking even greater optimization opportunities. This trajectory positions AIโ€‘assisted design as a longโ€‘term enabler for the semiconductor industry, helping companies stay ahead of the curve in an era where each new node brings diminishing returns on raw transistor scaling. Such forwardโ€‘looking improvements could also reduce the overall cost of ownership for customers, as lower power consumption and higher performance translate into smaller cooling solutions and longer device lifespans, reinforcing the economic incentive for AIโ€‘driven design innovation.

To assess how generative models have progressed in handling complete processor architectures, OpenAI selected two openโ€‘source 32โ€‘bit RISCโ€‘V cores for a systematic evaluation: PicoRV32, a compact core intended for FPGAโ€‘based prototyping, and Ibex, the embedded processor employed in the OpenTitan security chip. Each core was presented to the models in its original hardware description language, and the task was to increase its instruction throughputโ€”the number of instructions executed per secondโ€”while respecting strict constraints on area and correctness. The evaluation budget limited each model run to roughly 112,000 tokens, a figure chosen to reflect the amount of context a model can comfortably process in a single pass. Designs that failed simulation checks or exceeded sixteen times the original coreโ€™s area received a score of zero, ensuring that only viable architectures were considered. For valid designs, the team measured instructions per clock cycle on a set of hidden workloads, combined this metric with an estimated maximum clock frequency derived from Yosys, and then calculated an overall instructionsโ€‘perโ€‘second score. A perfect score of one represented the best achievable throughput under the given constraints, allowing for direct comparison across model versions and core types. The methodology also incorporated a geometric mean across the workload set to smooth out variability, and the token budget was deliberately set to challenge the models without overwhelming them, thereby revealing their true capability limits.

Across the evaluation, GPTโ€‘6.2 Astra consistently achieved the highest average scores on both PicoRV32 and Ibex, especially when operating with the larger token budget of approximately 112,000 tokens per run. On PicoRV32, its mean score reached roughly twice that of the earlier GPTโ€‘6 Astra model at comparable token limits, indicating substantial improvement in generating efficient pipeline optimizations. For Ibex, GPTโ€‘6.2 Astraโ€™s score continued to climb as the token budget increased, showing a clear upward trend that outperformed GPTโ€‘6.1 Sol and GPTโ€‘5.6 Sol at the same budgets. The charts presented in the research illustrate that the modelโ€™s ability to reason about architectural tradeโ€‘offs grew with each generation, allowing it to propose more aggressive yet still viable microarchitectural changes. These results underscore that the latest models are not only better at reโ€‘using existing structures but also at inventing novel configurations that achieve higher throughput without breaching the area or correctness limits imposed by the evaluation framework. The performance gap between the topโ€‘scoring model and its predecessors grew by more than thirty percent, highlighting the rapid advancement in the modelsโ€™ architectural reasoning capabilities and their increasing suitability for realโ€‘world chip design tasks. This progress suggests that AI can soon become a standard tool in the semiconductor design ecosystem.

In the second phase of the assessment, each model was given highโ€‘level requirements and code templates that described how a core should interface with the surrounding system, rather than being handed the complete RTL of an existing core. This approach allowed the models to decide which architectural features to retain and which to discard, such as performance counters or rarely used instructions that have minimal impact on the selected workloads. By omitting these legacy elements, the models could concentrate on the core functional units that directly affect instruction throughput, often resulting in more compact and efficient designs. Consequently, all five models achieved higher scores when constructing a core from the written specifications compared to attempting to improve the preโ€‘existing RTL, with GPTโ€‘6.2 Astra showing the most pronounced gain on the Ibex core. The 9,881 scored runs across both tasks and both cores demonstrated that the freedom to redesign rather than merely tweak existing structures unlocks greater design space, leading to more innovative and higherโ€‘performing architectures. These findings imply that designers should consider providing flexible specification inputs to AI assistants, as this can dramatically expand the range of viable solutions and accelerate the discovery of highโ€‘efficiency microarchitectures. In practice, this approach shortens development cycles and reduces the need for extensive manual optimization effort.

For engineers and organizations seeking to harness AI in chip design, the key takeaway is to integrate generative models early in the design flow, treating them as collaborative partners rather than mere novelty tools. Begin by defining clear performance targets and providing the model with highโ€‘quality RTL or specification documents that capture the essential functional and geometric constraints. Use established EDA platforms, such as those from Synopsys, to run automated verification after each AIโ€‘generated modification, ensuring that power, performance, and area goals remain met. Adopt a phased approach: start with small, wellโ€‘bounded blocks like the vector or matrix units, validate the modelโ€™s suggestions, and then scale the technique to larger subsystems such as entire CPU cores or GPU clusters. Finally, monitor the modelโ€™s evolution, as newer generations promise even tighter integration with layoutโ€‘aware tools and more accurate tradeโ€‘off analysis, making AI an increasingly indispensable component of modern semiconductor development pipelines. By embedding AI assistance into existing design reviews and continuous integration pipelines, teams can achieve faster iteration cycles, reduce manual debugging effort, and ultimately deliver higherโ€‘performance chips to market with lower development costs. Adopting these practices now positions companies to stay competitive as the semiconductor industry embraces AIโ€‘driven innovation and faces tighter timelines for new product introductions for sustained longโ€‘term growth.