New in-memory computing chip promises faster processing with lower energy use ⁠26%

By Neetika Walter⁠62%

7/8/2026, 11:28:18 PM

BS Summary: This article contains 16 faulty reasoning types, including Self-Serving Bias, Appeal to Authority, and Biased Writer Voice, with Confirmation Bias as the most egregious example at 18.8% saturation with 100 hits. Analysis detected 707 faulty-reasoning hits from 533 analyzed words, generating a BS Score of 38.1% and a BS Rank of ⁠26% (15,225 of 20,441 articles). This article is better (less manipulative) than 74.50% of the article peer group.

Artificial intelligence chip developers SK hynix and TetraMem have demonstrated a new memory-centric processor that performs AI calculations directly inside memory, a design aimed at cutting energy use and reducing the bottlenecks caused by moving data between processors and memory. 
The two companies announced the completion of a joint technology collaboration centered on an analog in-memory computing (A-IMC) system-on-chip (SoC). 
Their work demonstrates how memory can take on part of the computing workload instead of simply storing data. 
The prototype uses memristor-based in-memory computing to carry out efficient depthwise convolution, a key operation used in many AI inference models. 
By processing data where AI model weights are stored, the architecture reduces the need to repeatedly transfer information between memory and processors. 
The approach targets one of the biggest challenges facing modern AI hardware. 
As AI models grow from billions to trillions of parameters, data movement has become a major source of power consumption, latency, and heat generation inside computing systems. 
Computing inside memory 
Traditional AI chips continuously move data between compute units and memory, consuming both time and energy. 
Analog in-memory computing changes that workflow by performing matrix calculations directly within the memory array, reducing unnecessary data transfers . 
The joint project combines TetraMem’s analog in-memory computing platform with SK hynix’s expertise in advanced memory technologies. 
The companies also integrated emerging memory devices, circuit design, AI architecture, software, and system optimization into a single semiconductor platform. 
“We are honored to celebrate this important milestone together with SK hynix,” said Glenn Ge, CEO and Co-Founder of TetraMem. 
“This achievement demonstrates what can be accomplished through close collaboration across the semiconductor ecosystem.” 
According to the companies, the work goes beyond proving the concept of analog in-memory computing by demonstrating a practical AI system-on-chip that integrates multiple layers of hardware and software engineering. 
Tackling AI bottlenecks 
Growing AI workloads have increased pressure on chipmakers to improve energy efficiency without sacrificing performance. 
Memory-centric computing has emerged as one possible solution because moving data often consumes more energy than the calculations themselves. 
“We believe memory-centric computing and Analog In-Memory Computing will become increasingly important technologies for addressing future AI energy efficiency and thermal challenges, and we look forward to continuing our collaboration with SK hynix,” Ge said. 
The project represents a strategic move for SK hynix beyond traditional memory manufacturing into advanced computing architectures. 
While the company is a major producer of dynamic random-access memory (DRAM) and high-bandwidth memory (HBM) used in standard AI systems, this prototype shifts toward a neuromorphic approach. 
“We are pleased to see the successful outcome of this collaboration and the recognition from Advanced Intelligent Systems,” said Soo Gil Kim, Vice President of SK hynix. 
“This project demonstrates the value of exploring innovative memory technologies and new computing architectures for future AI systems.” 
The research paper was also selected as the cover feature of the journal, highlighting its technical contribution to next-generation AI hardware. 
The companies said they plan to continue working together on memory technologies, computing architectures and system integration for future AI infrastructure. 
The study, “A Memristor-based In-Memory Computing SoC with Efficient Depthwise Convolution,” was published in Advanced Intelligent Systems . 
Article reasoning-pattern comparisonThis article: 18.8%Neetika Walter: 4.0%Interesting Engineering: 3.9%Confirmation Bias18.8%This article: 0.0%Neetika Walter: 0.9%Interesting Engineering: 1.2%Anchoring Bias0.0%This article: 5.1%Neetika Walter: 1.8%Interesting Engineering: 2.5%Availability Heuristic5.1%This article: 3.6%Neetika Walter: 1.1%Interesting Engineering: 1.2%Representativeness Heuristic3.6%This article: 0.0%Neetika Walter: 0.2%Interesting Engineering: 0.3%Hindsight Bias0.0%This article: 7.5%Neetika Walter: 5.3%Interesting Engineering: 5.4%Overconfidence Bias7.5%This article: 5.4%Neetika Walter: 6.9%Interesting Engineering: 6.4%Framing Effect5.4%This article: 0.0%Neetika Walter: 0.1%Interesting Engineering: 0.2%Loss Aversion0.0%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.6%Status Quo Bias0.0%This article: 0.0%Neetika Walter: 0.2%Interesting Engineering: 0.4%Sunk Cost Effect0.0%This article: 12.8%Neetika Walter: 19.0%Interesting Engineering: 16.6%Optimism Bias12.8%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.5%Pessimism Bias0.0%This article: 0.0%Neetika Walter: 0.2%Interesting Engineering: 1.0%Negativity Bias0.0%This article: 15.4%Neetika Walter: 6.8%Interesting Engineering: 4.7%Self-Serving Bias15.4%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.1%Fundamental Attribution Error0.0%This article: 0.0%Neetika Walter: 0.1%Interesting Engineering: 0.0%Actor-Observer Bias0.0%This article: 0.0%Neetika Walter: 2.5%Interesting Engineering: 1.0%In-Group Bias0.0%This article: 0.0%Neetika Walter: 0.1%Interesting Engineering: 0.1%Out-Group Homogeneity Bias0.0%This article: 7.7%Neetika Walter: 4.7%Interesting Engineering: 5.0%Halo Effect7.7%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Horn Effect0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Dunning-Kruger Effect0.0%This article: 0.0%Neetika Walter: 0.5%Interesting Engineering: 1.0%Recency Bias0.0%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.2%Primacy Effect0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Blind-Spot Bias0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Ad Hominem0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Straw Man0.0%This article: 14.6%Neetika Walter: 6.1%Interesting Engineering: 9.0%Appeal to Authority14.6%This article: 2.8%Neetika Walter: 1.2%Interesting Engineering: 1.5%False Dilemma2.8%This article: 0.0%Neetika Walter: 0.1%Interesting Engineering: 0.3%Slippery Slope0.0%This article: 0.0%Neetika Walter: 0.1%Interesting Engineering: 0.1%Circular Reasoning0.0%This article: 3.4%Neetika Walter: 4.0%Interesting Engineering: 5.1%Hasty Generalization3.4%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.2%Red Herring0.0%This article: 0.0%Neetika Walter: 0.5%Interesting Engineering: 0.8%Bandwagon0.0%This article: 0.0%Neetika Walter: 2.8%Interesting Engineering: 2.4%Appeal to Emotion0.0%This article: 0.0%Neetika Walter: 1.0%Interesting Engineering: 1.3%Begging the Question0.0%This article: 8.6%Neetika Walter: 1.6%Interesting Engineering: 2.2%Post Hoc (False Cause)8.6%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Tu Quoque0.0%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.6%Burden of Proof0.0%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.2%Appeal to Nature0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.3%Composition/Division0.0%This article: 0.0%Neetika Walter: 0.4%Interesting Engineering: 0.7%Anecdotal0.0%This article: 0.0%Neetika Walter: 0.1%Interesting Engineering: 0.1%No True Scotsman0.0%This article: 5.3%Neetika Walter: 1.1%Interesting Engineering: 2.7%Ambiguity (Equivocation)5.3%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Middle Ground0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Personal Incredulity0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.2%Special Pleading0.0%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Genetic Fallacy0.0%This article: 0.0%Neetika Walter: 0.9%Interesting Engineering: 1.8%Unattributed Quote0.0%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.7%Quote-first Misdirection0.0%This article: 13.1%Neetika Walter: 2.1%Interesting Engineering: 3.8%Biased Writer Voice13.1%This article: 6.6%Neetika Walter: 0.5%Interesting Engineering: 0.8%Indoctrination6.6%This article: 0.0%Neetika Walter: 0.0%Interesting Engineering: 0.0%Politically Left Leaning Bias0.0%This article: 0.0%Neetika Walter: 0.3%Interesting Engineering: 0.1%Politically Right Leaning Bias0.0%This article: 2.1%Neetika Walter: 15.5%Interesting Engineering: 10.6%Attempt to Sell a Product or S…2.1%

533 words analyzed.

Speakers

2speakers21%attributed speech419writer words
Voice mapSelect a segment to jump to its words
Writer's voice • 11 words • 100.0% coverageWriter's voice • 40 words • 100.0% coverageWriter's voice • 20 words • 0.0% coverageWriter's voice • 18 words • 0.0% coverageWriter's voice • 21 words • 0.0% coverageWriter's voice • 22 words • 0.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 27 words • 0.0% coverageWriter's voice • 3 words • 0.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 20 words • 0.0% coverageWriter's voice • 17 words • 0.0% coverageWriter's voice • 20 words • 0.0% coverageGlenn Ge • 20 words • 0.0% coverageGlenn Ge • 14 words • 0.0% coverageWriter's voice • 30 words • 100.0% coverageWriter's voice • 3 words • 0.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 19 words • 0.0% coverageGlenn Ge • 35 words • 100.0% coverageWriter's voice • 17 words • 0.0% coverageWriter's voice • 28 words • 0.0% coverageSoo Gil Kim • 27 words • 0.0% coverageSoo Gil Kim • 18 words • 0.0% coverageWriter's voice • 21 words • 0.0% coverageWriter's voice • 21 words • 0.0% coverageWriter's voice • 18 words • 0.0% coverage
Selected voice

Glenn Ge

100%flagged-word coverage
69 attributed words61% of attributed speech72% writer coverage
0%27.5%55.0%Indoctrination+50.7 ptsWriter: 0.0%Glenn Ge: 50.7%50.7%Biased Writer Voice-16.7 ptsWriter: 16.7%Glenn Ge: 0.0%0.0%Attempt to Sell a Product -2.6 ptsWriter: 2.6%Glenn Ge: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…
Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.