ReadingResearchRadarInvestment framework
Sign in / Sign up
AllTechnologyProductsInvestingPublic marketsOther

Reading

2026-11-0159 posts

Refresh

OpenSearch Foundation: AWS Contribution Share Drops to 72%, Self-Managed Deployments Rise to 60%

Topic · OpenSearchMaterial
2026-11-01 00:02 GMT+8

Bianca Lewis, Executive Director of the OpenSearch Software Foundation, told The New Stack that the project's governance structure has undergone substantive changes since AWS donated OpenSearch to the Linux Foundation in September 2024. Latest data reveals that AWS's share of contribution activities across OpenSearch repositories has dropped from 91% in the first year to 72% in 2025–26. Meanwhile, the proportion of maintainers affiliated with Amazon in the core repository has decreased from 95% in April 2022 to 71% today.

This shift reflects enterprise demand for vendor neutrality. According to the "2026 Open Data Infrastructure Report" published by the Linux Foundation and the OpenSearch Software Foundation, self-managed cloud deployments have grown faster than Amazon's managed OpenSearch Service since 2024. Currently, 60% of surveyed OpenSearch users run self-managed deployments in the cloud, up significantly from 37% in 2024, while usage of Amazon's managed service rose more modestly from 44% to 48%.

Although AWS still accounts for nearly three-quarters of code contributions and holds four of the 15 voting seats on the Technical Steering Committee (TSC), with AWS employee Andrew Ross re-elected as committee chair in October, Lewis argues that legal independence is the foundation for building trust. "When they donated it to the Linux Foundation, that was a watershed moment," Lewis said. "Because they couldn't close OpenSearch and they couldn't change the license. They didn't own the application anymore."

OpenSearch is positioning itself as a central component of enterprise AI infrastructure, particularly for applications like Retrieval-Augmented Generation (RAG). Its latest release, OpenSearch 3.9, enhances vector and hybrid search capabilities to help developers retrieve information for AI applications. However, the report notes that 24% of non-adopters remain hesitant due to concerns about the project's historical ties to AWS.

Sources:The New Stack

Research

The Verge's Sean Hollister: After Modding a $50 Device with Claude Code, I Believe the Value of Dedicated AI Hardware Lies Elsewhere

Material
2026-11-01 00:05 GMT+8

Meta plans to release the Muse Charm this holiday season—a wearable AI pet hardware device. The Verge gadgets editor Sean Hollister didn't buy one; instead, he paid $50 to have a Waveshare ESP32-S3 1.75C touchscreen development board drop-shipped directly from China, using Claude Code to turn it into his own version of the Muse Charm.

This development board is one of the officially recommended Muse Gadget development kits by Meta, using Meta's open-source code and tokens to connect directly to the Muse AI cloud. Hollister plugged the device into his computer, directed Claude at Meta's GitHub repository, and paired it with the Muse mobile app. However, the out-of-the-box experience felt nearly useless: Muse treated his voice as audio clips for transcription rather than commands; responses were slow and silent; the screen displayed only a few words with no place to read full messages; there was only one static avatar named Jolly; and it failed to automatically discover smart home devices.

What truly changed everything was a few rounds of dialogue with Claude Code. When Hollister told Claude he couldn't read or hear Muse's replies, Claude built a conversation history page allowing him to swipe right to view full messages; he integrated ElevenLabs' British AI voice so the device could speak answers aloud. Complaining that the phone's front-facing camera showed an upside-down image during remote photo capture, Claude fixed the issue and added a low-quality Bluetooth real-time preview, enabling creative framing even when not connected to home Wi-Fi.

The electronic pet aspect served as another example. He asked Claude to create a batch of different avatars and add a Tamagotchi-style mini-game where the pet needs feeding, playing, sleeping, and bathing. Finding the game too simple to hold attention, he instructed Claude to add four levels each for hunger, boredom, and dirtiness. He decided that snacks would leave crumbs making the pet dirty, and excessive play would tire it out. Throughout the process, he only needed to describe what he wanted; Claude handled the coding.

He also had Claude spend 15 minutes creating a pocket watch face with three hands that actually tells time, as well as a video player. He downloaded, edited, and loaded Mark Zuckerberg's introduction clip of the Muse Charm onto the small screen.

Hollister explicitly stated he has no programming experience and doesn't know if the code quality is good or if it contains fatal flaws. He also does not trust Meta or Muse, citing prior reports that Muse leaked a reviewer's home address; therefore, he placed the device on a guest Wi-Fi network from the start.

His final judgment is that Muse itself is not a reason to buy dedicated computing hardware, because the Muse features on a smartphone offer more functionality than the development board in its out-of-the-box state. The reason to own dedicated hardware is that you can reprogram the entire device to do things the smartphone in your pocket cannot. This raises a question about the upcoming Muse Charm: Will it allow such free programming, or should people who really want to tinker just buy a development kit?

Sources:The Verge

Research

Former IDC President Del Prete: The J-Curve Between Enterprise AI Personal Efficiency and Corporate Profitability

Topic · 企业AI J型曲线Material
2026-11-01 01:12 GMT+8

The massive investment in AI infrastructure is essentially a bet that enterprises will eventually be willing to pay for intelligence. However, a more pressing question arises: What internal changes must occur within enterprises for these investments to truly become self-funding? In the SiliconANGLE Breaking Analysis program, Crawford Del Prete, former president of IDC and now senior adviser at PSG Equity, provided a specific mechanistic explanation: Currently, AI is automating a large number of individual tasks, but it has not yet changed how work is organized within an organization. Consequently, there is a J-curve delay between individual productivity improvements and the company's income statement.

Del Prete’s observations stem from his daily interactions with private equity portfolio companies and a wide range of enterprise clients. He described a seemingly contradictory phenomenon: On one hand, quantitative-oriented clients have 'insatiable' demand for compute capacity—they are willing to purchase all available GPU capacity because AI enables them to create analyses and deliverables previously impossible. On the other hand, broader enterprise customers are beginning to use AI in expected areas such as coding and customer service, and it is starting to permeate various specific tasks within organizations. However, 'jobs consist of hundreds or even thousands of tasks; we are starting to automate these tasks without changing the way work is orchestrated within an organization.'

This is the source of the J-curve delay. Host Vellante introduced the research framework by Brynjolfsson, Rock, and Syverson regarding how intangible investments affect productivity: The adoption phase involves initial investments in tools and skills, followed by process redesign; both steps result in negative cumulative cash flow. Only when integration spans the entire organization does the curve begin to rise steeply. Del Prete drew an analogy to the PC era—actual growth in productivity data became visible only after thirteen or fourteen years following the introduction of PCs, with the key turning point occurring during the internet connectivity period from 1996 to 2002. He also pointed out a significant difference from the dot-com era: While dark fiber laid down back then remained idle for a long time, 'every GPU brought online today is at 100% utilization.'

Del Prete believes the specific path to crossing this curve is workflow redesign, rather than simply automating existing processes. He contrasted this with previous Robotic Process Automation (RPA): RPA was difficult because you had to codify human steps into code; AI differs in that it starts 'from the bottom up'—employees solve pain points they have faced for a long time using natural language, then encode those solutions into replicable projects or skills, which are then shared across the organization. He described this as 'an extremely flexible, adaptable, and usable form of RPA that never existed before.'

At the level of software pricing, Del Prete observed the most significant change is the migration from seat-based to usage-based models. For non-SaaS companies introducing AI, the biggest challenge is 'how to discretely monetize AI'—the answer is usually not to sell AI separately, but to embed it into core products, demonstrating value through cross-selling, upselling, and price increases, ultimately launching incremental features exclusive to AI. He remains skeptical about outcome-based pricing because 'it becomes very episodic and project-specific, so you don’t get that stickiness.'

Sources:SiliconANGLE

Research

PolyAI CEO Nikola Mrkšić: Why Voice Remains the Primary Channel for Business Contact

Topic · PolyAIQuick take
2026-10-31 23:57 GMT+8

Nikola Mrkšić, co-founder and CEO of PolyAI, recently appeared in a clip from the 'Disruptive' podcast by Barclays Eagle Labs, hosted by Luke Christoforidis. In this segment, Mrkšić discusses why the prevailing view in Silicon Valley that 'voice is dead' is detached from the reality of how people actually reach out to businesses.

PolyAI builds enterprise voice AI agents that answer customer calls. This resource is suitable for professionals seeking to understand the current application status and market positioning of voice AI in business scenarios. While the full argumentation details are not provided, the direct perspective from the founder of a leading voice AI company offers a reference point for evaluating the continued importance of voice technology in enterprise communications.

Sources:youtube.com

Research

PolyAI Co-founder Nikola Mrkšić Discusses AI's Impact on Jobs in Podcast Clip

Topic · PolyAI就业影响Quick take
2026-10-31 23:54 GMT+8

Nikola Mrkšić, co-founder of PolyAI (poly.ai), discusses the profound impact of artificial intelligence on the labor market in a recent clip from the tech interview show 'The Unlock with Oliver Bruce'. Mrkšić, originally from Belgrade, references his mother's experience working at the National Bank of Serbia as a personalized example to explain the magnitude of the AI wave.

According to the video title and publisher description, Mrkšić notes that his mother's department originally had 40 people, but only 2 remained. He uses this case to argue that AI represents "the same wave at far greater scale." This viewpoint provides a specific, tangible entry point for understanding how AI technology reshapes workforce structures in traditional industries.

For readers focused on AI industry investment and social implications, the value of this clip lies in its first-hand observation from the founder of a company specializing in enterprise voice AI agents. Although a full transcript or detailed argumentation is not provided, the resource clearly identifies the guest, their affiliation, and the direction of their core thesis, making it suitable as quick reference material for understanding AI-driven job displacement.

Sources:youtube.com

Research

PolyAI Co-founder Nikola Mrkšić Discusses Leaving Apple Over Siri Dissatisfaction

Topic · PolyAIQuick take
2026-10-31 23:54 GMT+8

PolyAI co-founder Nikola Mrkšić recently appeared in a clip from the tech interview show 'The Unlock with Oliver Bruce,' sharing his experience of leaving Apple to found PolyAI. Mrkšić previously worked on Siri within Apple, noting in the description that persistent feedback about Siri's poor performance ('People kept telling me Siri's sh*t') motivated him to leave and build voice AI focused on doing one job really well.

This narrative provides a micro-view of AI talent mobility and competitive dynamics within major tech companies. Mrkšić's startup, PolyAI, currently builds voice AI agents that answer customer calls for enterprises. Although the clip does not elaborate on technical arguments or market analysis, it clearly identifies the motivation of former Apple employees turning to vertical voice interaction development due to dissatisfaction with general-purpose assistant experiences. For professionals interested in AI industry investment, particularly in voice technology commercialization, this background story reveals a potential driver behind the shift from general large models to specialized tools for specific scenarios.

Sources:youtube.com

Research

OpenAI Plans to Train Models on Previously Off-Limits Daybreak Blue Cybersecurity Session Data

Topic · OpenAI安全隔离争议Material
2026-11-01 00:38 GMT+8

According to The Decoder, OpenAI plans to begin using session data from its Daybreak Blue cybersecurity product for model training starting November 29. The change was confirmed via an email to members of the Trusted Access program. Previously, this data was excluded from training even when users had turned off the "improve the model for everyone" setting.

The backlash was swift. OpenAI followed up with a second email promising a separate opt-out mechanism. However, the report notes that interactions flagged during safety reviews can still be used for safety purposes.

The move has researchers worried about a repeat of the Buckmaster-Alpöge case. In that incident, mathematicians Tristan Buckmaster and Levent Alpöge uploaded Navier-Stokes equation drafts to OpenAI's Codex. OpenAI soon presented its own proof but denied using their inputs for training. Yet the suspicion alone was enough to show researchers what could go wrong.

Sources:The Decoder

Research

US DOJ Indicts Five in $200 Million Student Loan Fraud Ring Using AI to Fake Identities

Topic · AI助学贷款欺诈案Material
2026-10-31 23:37 GMT+8

The U.S. Department of Justice (DOJ) announced on October 31, 2026, that the first results of Operation Ghostbusters have led to the indictment of five defendants and the seizure of more than $750,000 in cryptocurrency tied to a network associated with Kenyan nationals. This marks the first international seizure of its kind in Kenya.

According to the agency, bad actors used stolen identities to sign up for student loans and used AI to complete the application, making it appear that the fake student was attending online classes. While schools received the majority of the proceeds, the remaining amount was funneled into bank accounts of the fraudsters, who then used mules to move money to overseas conspirators. Unwitting victims who skipped college are saddled with debt they never signed up for.

Multiple community colleges and universities across the United States started suspecting fraud in the Federal Student Aid (FAFSA) program in 2021. Estimates suggest the U.S. government lost over $200 million due to this activity. Currently, more than 40 open fraud cases involve 26 FBI field offices and 20 U.S. Attorney’s offices.

Sources:Tom's Hardware

Research

Report: OpenAI raises just $11 billion of its $30 billion target as investors pull back

Topic · OpenAI估值与融资Material
2026-10-31 17:22 GMT+8

According to a report by The Decoder citing the Wall Street Journal, OpenAI has raised less than half of its targeted $30 billion in its latest private funding round, securing approximately $11 billion so far. The company's valuation landed at $1.25 trillion, falling short of the $1.4 trillion it had aimed for.

The shortfall is reportedly driven by fading interest from major institutional investors amid a broader tech sell-off triggered by weak quarterly results from Google, Amazon, and Intel. In response, OpenAI plans to raise up to $20 billion from family offices and high-net-worth individuals. The company is also reportedly planning an initial public offering early next year.

This development contrasts with rival Anthropic, which is pushing to go public as early as November. A strong IPO could leave Anthropic with well over $100 billion in capital, while OpenAI continues to push price cuts and more efficient models to hold its market share.

Sources:The Decoder

Research

"ImageNet Classification with Deep Convolutional Neural Networks": How ReLU and Multi-GPU Parallelism Enabled AlexNet

Verified 2026-11-01 17:02 GMT+8

Long-term reading · 《ImageNet Classification with Deep Convolutional Neural Networks》(2012)

In 2012, the paper "ImageNet Classification with Deep Convolutional Neural Networks" (commonly known as AlexNet), co-authored by Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton of the University of Toronto (Canada), marked a turning point for deep learning in the field of computer vision. The core contribution of this paper was to demonstrate that through specific architectural optimizations and hardware utilization, deep convolutional neural networks could achieve performance far surpassing traditional methods on large-scale image datasets.

The paper first addressed the issue of training efficiency. Traditional saturated non-linear activation functions (such as tanh or sigmoid) were slower during gradient descent training. The authors adopted unsaturated neurons called Rectified Linear Units (ReLUs), defined as f(x) = max(0, x). The text notes, "Deep convolutional neural networks with ReLUs train several times faster than their equivalents with tanh units." This improvement made it possible to train large networks with 60 million parameters within a limited time frame.

Secondly, to address the problem of model size exceeding the memory limits of a single GPU, the authors implemented a multi-GPU parallel training strategy. Since a single GTX 580 GPU had only 3GB of memory, and the 1.2 million training images were sufficient to train a network too large to fit on a single card, they stated, "we spread the net across two GPUs." This parallelization scheme not only broke through hardware bottlenecks but also improved computational efficiency by reducing communication overhead, ultimately lowering top-1 and top-5 error rates by 1.7% and 1.2%, respectively.

Additionally, the paper mentioned the Dropout regularization method to reduce overfitting in fully connected layers, as well as Local Response Normalization to assist generalization. These combined techniques enabled AlexNet to achieve a top-5 test error rate of 15.3% in the ILSVRC-2012 competition, significantly leading the runner-up's 26.2%.

For today's readers, understanding AlexNet is less about replicating its specific numerical results and more about mastering the engineering mindset used to resolve the "contradiction between data scale and computing resources." For example, consider a modern AI team facing similar challenges: they possess a new domain dataset containing millions of samples but are constrained by cloud GPU quotas. Drawing on the approach of AlexNet, they should not focus solely on the mathematical elegance of the algorithm itself but should prioritize how to modify activation functions to accelerate convergence (such as adopting ReLU variants) and how to design cross-device data/model parallel strategies to break through single-point computing power limits. This methodology of tightly integrating algorithmic innovation with hardware constraints remains the foundation for building efficient AI systems today.

We recommend all readers interested in the history of computer vision and deep learning engineering practices to read the paper "ImageNet Classification with Deep Convolutional Neural Networks" (2012). It is not only the original source of AlexNet but also demonstrates how systematic innovation can achieve performance leaps in resource-constrained environments. We suggest starting with the abstract and Section 3 (Architecture), paying close attention to the specific implementation details of ReLU and multi-GPU training.

Sources:《ImageNet Classification with Deep Convolutional Neural Networks》(2012)

Research

The Final Act of the Superbubble: Jeremy Grantham on Market Irrationality Phases and Bear Market Rally Traps

Verified 2026-11-01 17:02 GMT+8

Long-term reading · 《Entering the Superbubble’s Final Act》(2022)

In an investment career, truly significant market events are few, and "superbubbles" are among the most decisive. Jeremy Grantham points out that these events differ from ordinary bull-bear transitions; they possess unique common characteristics and are often accompanied by dramatic phase changes in investor behavior. The key to understanding this concept lies in identifying extreme valuation states that deviate more than 2.5 standard deviations from the trend, rather than merely focusing on daily market fluctuations.

Grantham divides the bursting process of a superbubble into four stages: first, bubble formation; second, the initial correction caused by flaws in the economic or political environment; third, the highly deceptive "bear market rally"; and finally, the ultimate decline triggered by deteriorating fundamentals. The bear market rally occurs after the initial devaluation but before a clear economic recession. Historical data shows that such rallies typically recover more than half of the initial drop, luring less vigilant investors back into the market, after which the market turns downward with even greater momentum.

Taking the cases of 1929, 1973, and 2000 as examples, the bear market rallies during these periods were extremely fast and large in magnitude. For instance, within two months after the burst of the tech bubble in 2000, the Nasdaq recovered 60% of its initial losses. The psychological mechanism behind this phenomenon is the anchoring effect: investors see stock prices fall significantly from their highs and mistakenly assume they have become "cheap," thereby ignoring the fact that valuations remain far above long-term averages. In contrast, new highs in normal bull markets are often accompanied by hesitation and slow progress due to the lack of historical price references.

For today's readers, the core question is how to distinguish between normal market corrections and the prelude to a superbubble burst. Grantham emphasizes that when the market is in a period of extreme optimistic irrationality (which accounts for approximately 12% of the time), traditional risk management strategies may fail. At such times, institutional managers often tend to follow the crowd and become "permanent bulls," as this is the best choice for career safety. However, for independent investors, identifying this collective behavioral phase change and being wary of the subsequent rapid rally is key to avoiding severe losses in the late stages of the bubble.

"Entering the Superbubble’s Final Act" was published by Jeremy Grantham in 2022. This document is suitable for mature investors who wish to deeply understand extreme market cycles, particularly how to identify the "bear market rally" trap during the bubble bursting process. It is recommended to start reading from the sections on the definition of superbubbles and their four-stage life cycle, paying close attention to the specific citations of historical data and the analytical logic.

Sources:《Entering the Superbubble’s Final Act》(2022)

Research

You’re all caught up in this view

Sign in / Sign up
ReadingResearchRadarInvestment framework

Del Prete also proposed a three-layer software model to explain AI’s impact on industry dynamics: the execution layer (the computational work itself), the judgment layer (automatically generating insights from data), and the accountability layer (who is responsible for the results). He believes the execution and judgment layers are being absorbed by AI, while 'the accountability layer is where companies are truly focusing today'—who answers the phone at 2 a.m., who is responsible for compliance data. This implies that future software companies need to be as efficient as possible in execution and judgment, while enabling humans to take on more decision-making responsibility than ever before.

Regarding observable indicators for whether AI is approaching self-funding, Del Prete listed several data sets he actually tracks in his portfolio: How net retention and gross retention change with the rollout of AI features; win/loss analysis—whether customers are switching to more AI-native competitors or building internally; whether AI features are attracting new customers; and whether each incremental customer receives a better experience due to ecosystem data. He particularly emphasized tracking the 'next best action': Whether tickets are reopened, invoices are disputed, or recommendations are accepted—these follow-up interactions constitute a two-way customer value flywheel.

The framework maintained jointly by Vellante and Del Prete is their summer 2026 AI bubble forecast research, which marked 2029 as the highest-risk year—if infrastructure supply catches up before productivity monetization materializes, the risk of a bubble bursting rises to dangerous levels. Their assessment at the time was that supply constraints and healthy capital funding appetites could provide conditions for a soft landing. Del Prete acknowledges 'these productivity numbers take a long time,' but adds 'I have never seen a technology embraced by the world so rapidly.' The tension between these two points is the core of this article: Adoption speed is unprecedented, but the speed of organizational redesign has never been faster than the technology itself.

It is important to clarify that this is an attribution argument based on Del Prete’s observations from his PE portfolio and decades of technology cycle experience, not an independently verified predictive model. The J-curve is a conceptual analogy, and 2029 is a conditional judgment (dependent on the relative speeds of supply and monetization). The clients Del Prete interacts with are already high-adoption enterprises, limiting their representativeness for the broader market. His observations provide specific conditions and trackable indicators for judging whether enterprise AI is approaching self-funding, but they do not constitute a definitive prediction of whether a bubble will burst.