AMD Ryzen AI 9 PRO 465 vs Intel Core Ultra 5 235T Comparison

Where Each One Wins

The recorded benchmark data splits these two processors nearly evenly, with Intel taking 6 wins and AMD taking 5 in the head-to-head PassMark suite. The split is not arbitrary; it follows a clear pattern tied to workload type. The AMD Ryzen AI 9 PRO 465 dominates in integer-heavy and data-processing tasks. Its largest victories come in data compression (385174 vs 295100, a 30.5% advantage), integer math (107173 vs 84244, a 27.2% advantage), and random string sorting (40860 vs 35377, a 15.5% advantage). It also wins in extended instructions (26441 vs 23212, a 13.9% advantage) and edges out in the overall multithread score (31485 vs 30918, a 1.8% advantage).

The Intel Core Ultra 5 235T claims the opposite territory. It wins decisively in prime number finding (318 vs 126, a 60.4% advantage), floating point math (106546 vs 66824, a 37.3% advantage), and physics simulation (2157 vs 1747, a 19% advantage). It also leads in data encryption (23457 vs 19308, a 17.7% advantage) and in single-thread performance (4339 vs 4168, a 3.9% advantage). The pattern suggests AMD's strengths lie in compression, sorting, and general integer throughput, while Intel's architecture handles floating-point, encryption, and single-threaded physics workloads more efficiently. The multithread result is close enough that neither chip can claim overall dominance, but the use-case split is sharp.

Architecture Differences

The two processors come from fundamentally different design philosophies. AMD's Ryzen AI 9 PRO 465 uses the Zen 5 architecture on a 4 nm TSMC process, built on the Gorgon Point codename. It packs 10 cores and 20 threads, relying on simultaneous multithreading to double its thread count. The cache layout uses 80 KB of L1 per core, 1 MB of L2 per core, and 16 MB of L3. The die measures 233 mm². Intel's Core Ultra 5 235T uses Arrow Lake architecture on a 3 nm TSMC process, with the Arrow Lake-S codename. It has 14 cores but only 14 threads, indicating no hyperthreading. Its cache is larger across the board: 192 KB L1 per core, 3 MB L2 per core, and 24 MB of shared L3, on a 243 mm² die with 17,800 million transistors.

The memory and I/O configurations differ as well. AMD supports both DDR5 and LPDDR5X memory with a dual-channel bus and 89.6 GB/s bandwidth, while Intel supports DDR5 only, also dual-channel, but with higher bandwidth at 102.4 GB/s. AMD uses PCIe Gen 4 with 16 CPU lanes; Intel uses PCIe Gen 5 with 20 CPU lanes. The integrated graphics differ: AMD has the Radeon 890M, while Intel has the Arc Xe-LPG Graphics 24EU. AMD's socket is AMD Socket FP8, designed for mobile, while Intel uses Intel Socket 1851 for desktop. The AMD chip has a 28 W TDP, the Intel chip a 65 W TDP. The process node difference (4 nm vs 3 nm) and the cache structure explain much of the benchmark behavior, as the larger L2 and L3 on the Intel side benefit floating-point and single-threaded tasks, while AMD's SMT and thread count drive its integer and compression wins.

Head-to-Head Benchmarks

The biggest single win for AMD appears in data compression, where it scores 385174 against Intel's 295100. That 30.5% gap is the largest advantage any chip holds in this comparison, and it aligns with AMD's higher thread count and SMT capability. Integer math follows with a 27.2% lead (107173 vs 84244), reinforcing the pattern of AMD's strength in parallel integer workloads. Random string sorting adds a 15.5% win (40860 vs 35377), and extended instructions show a 13.9% advantage (26441 vs 23212). The multithread score is nearly even, with AMD leading by just 1.8% (31485 vs 30918), which indicates that the two processors are broadly comparable in total throughput despite their architectural differences.

Intel's largest win is prime number finding, where it scores 318 against AMD's 126, a 60.4% margin. This is the most lopsided result in either direction. Floating point math shows a 37.3% advantage for Intel (106546 vs 66824), and physics simulation a 19% advantage (2157 vs 1747). Data encryption goes to Intel by 17.7% (23457 vs 19308), and single-thread performance is close but favors Intel by 3.9% (4339 vs 4168). The single-thread and singlethread entries are identical as expected. These results show that Intel's per-core efficiency, likely aided by its larger cache and newer process node, carries it in latency-sensitive and floating-point-heavy tasks, while AMD's thread count wins in throughput-oriented integer work. The overall multithread score being within 2% suggests that for mixed workloads, the choice depends heavily on the specific application mix.

The Verdict

The data indicates two distinct profiles. The AMD Ryzen AI 9 PRO 465 is the stronger choice for workloads involving data compression, integer math, sorting, and extended instruction sets. Its 30.5% lead in compression and 27.2% lead in integer math are substantial, and its slim multithread advantage confirms it can handle general parallel tasks at least as well as the Intel part. The 20 threads and SMT support give it a clear edge in heavily threaded integer operations.

The Intel Core Ultra 5 235T is the better pick for floating-point math, prime number generation, physics simulation, and encryption. Its 60.4% win in prime finding and 37.3% win in floating point are decisive, and the 3.9% single-thread lead indicates better per-core performance for lightly threaded applications. The higher memory bandwidth (102.4 GB/s vs 89.6 GB/s) and larger L3 cache (24 MB vs 16 MB) likely contribute to these results.

For a user prioritizing compression, database-style integer work, or sorting algorithms, the AMD chip is clearly ahead. For scientific computing, physics, encryption, or single-threaded responsiveness, the Intel chip wins. The two are nearly tied in overall multithread performance, so the deciding factor should be the specific workload mix. The AMD part operates at a lower TDP (28 W vs 65 W), which matters for mobile deployments, while the Intel part targets desktop use with a launch MSRP of $247. Neither processor is universally superior; the recorded data supports a workload-based selection.

FAQ

Q: Which processor is faster in multi-threaded workloads?

A: The AMD Ryzen AI 9 PRO 465 leads in the PassMark multithread test with a score of 31485 versus 30918 for the Intel Core Ultra 5 235T, a 1.8% advantage.

Q: How large is the single-thread performance gap?

A: The Intel Core Ultra 5 235T scores 4339 in the single-thread test compared to 4168 for the AMD Ryzen AI 9 PRO 465, giving Intel a 3.9% lead.

Q: Which chip dominates floating-point calculations?

A: The Intel Core Ultra 5 235T wins floating point math decisively with 106546 versus 66824 for AMD, a 37.3% advantage.

Q: What is the biggest benchmark difference in either direction?

A: Intel's win in prime number finding is the largest gap, scoring 318 versus AMD's 126, a 60.4% margin. AMD's largest win is data compression at 385174 versus 295100, a 30.5% margin.

Q: How do the core and thread counts compare?

A: The AMD Ryzen AI 9 PRO 465 has 10 cores and 20 threads, while the Intel Core Ultra 5 235T has 14 cores and 14 threads. AMD uses simultaneous multithreading, Intel does not.

Q: Which processor has more cache?

A: The Intel Core Ultra 5 235T has larger caches: 192 KB L1 per core, 3 MB L2 per core, and 24 MB shared L3. The AMD Ryzen AI 9 PRO 465 has 80 KB L1 per core, 1 MB L2 per core, and 16 MB L3.

Specification Differences

| Specification | AMD Ryzen AI 9 PRO 465 | Intel Core Ultra 5 235T |

| --- | --- | --- |

| Cores | 10 | 14 |

| Threads | 20 | 14 |

| Base Clock | 2.00 GHz | 2.20 GHz |

| Boost Clock | 5.00 GHz | 5.00 GHz |

| TDP | 28 W | 65 W |

| Socket | AMD Socket FP8 | Intel Socket 1851 |

| Architecture | Zen 5 | Arrow Lake |

| Codename | Gorgon Point | Arrow Lake-S |

| Process Node | 4 nm | 3 nm |

| Die Size | 233 mm² | 243 mm² |

| L1 Cache | 80 KB (per core) | 192 KB (per core) |

| L2 Cache | 1 MB (per core) | 3 MB (per core) |

| L3 Cache | 16 MB | 24 MB (shared) |

| Memory Support | DDR5, LPDDR5X | DDR5 |

| Memory Bandwidth | 89.6 GB/s | 102.4 GB/s |

| PCIe | Gen 4, 16 Lanes (CPU only) | Gen 5, 20 Lanes (CPU only) |

| Integrated Graphics | Radeon 890M | Arc Xe-LPG Graphics 24EU |

| Market Segment | Mobile | Desktop |

| Release Date | 2026-01-04 | 2025-01-06 |

| Transistors | Not specified | 17,800 million |