Stay Ahead, Stay ONMINE

I Tried Making my Own (Bad) LLM Benchmark to Cheat in Escape Rooms

Recently, DeepSeek announced their latest model, R1, and article after article came out praising its performance relative to cost, and how the release of such open-source models could genuinely change the course of LLMs forever. That is really exciting! And also, too big of a scope to write about… but when a model like DeepSeek […]

Recently, DeepSeek announced their latest model, R1, and article after article came out praising its performance relative to cost, and how the release of such open-source models could genuinely change the course of LLMs forever. That is really exciting! And also, too big of a scope to write about… but when a model like DeepSeek comes out of nowhere with a steel chair, boasting similar performance levels to other models, what does performance really mean in this context?

If you follow AI releases, you’ve seen this dance before. Every new model drops with its graphs showing how it’s somehow simultaneously better than GPT-4 on math problems while being smaller and more efficient. But what exactly are these benchmarks measuring? How are they created? And more importantly, how can we cut through the hype to create our own benchmarks for specific use cases?

I wanted to learn more about LLM Benchmarking.

Part 1: What is a Benchmark? (in 3 seconds)

TL:DR — The SATs (multiple, actually) for LLMs.

Part 1.1: What is a Benchmark? (in more than 3 seconds)

Before we dive into the nitty-gritty of specific benchmarks, let’s take a moment to unpack what we even mean by “LLM Benchmark.” Because calling them the “SATs for AI” feels both right and also slightly oversimplified.

LLM benchmarks are, at their core, structured tests used to measure how well large language models perform on certain tasks. These tasks can be anything from identifying if a statement is true or false, to summarizing a legal document, to generating valid Python functions. Think of them as curated obstacle courses specially designed by AI researchers to test every relevant muscle these models might have. These frameworks typically provide a dataset of inputs with known correct outputs, allowing for consistent comparison between models.

Modern benchmarks employ various evaluation methodologies. Classification metrics like accuracy work for tasks with discrete correct answers, while overlap-based metrics (BLEU, ROUGE) evaluate free-form text generation. Some benchmarks use functional testing for code generation, or employ other LLMs as judges to evaluate response quality.

A typical benchmark usually comes packaged as:

  • A standardized dataset of questions, prompts, or tasks (with correct or reference answers).
  • An evaluation protocol specifying how to measure success, like accuracy, F1 score, BLEU/ROUGE for text generation, or pass/fail rates for coding tasks.
  • A leaderboard or some form of comparative scoreboard, often with big flashy graphs.

Some really famous benchmarks include MMLU for testing multitask language understanding, TruthfulQA for assessing factual accuracy, and HumanEval for measuring coding capabilities. Results are pretty often published on public leaderboards, which let’s people perform some transparent comparison between different models.

From the DeepSeek paper: DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

What Makes a Good Benchmark?

  1. A Clear Task Definition: We want tasks that are unambiguous. The more straightforward and well-specified the challenge, the easier it is to trust the results.
  2. Data Integrity: The test set shouldn’t be floating around in the training data. Because if the model’s seen the exact same question 50 times before, the evaluation is about as useful as giving a math quiz to someone who already has the answer key.
  3. Quantifiable Metrics: You need a standard for scoring performance — like how many times the model’s code passes test cases or how close the generated summary is to a “ground-truth” summary.
  4. Task Diversity & Difficulty: If a benchmark is too easy, everyone just ACES it on day one, and we learn… well, nothing. If it’s too niche (like “We test only the model’s ability to count the digits of Pi for 20 minutes”), that’s also not so helpful.

Life Ain’t All about The Grades

Benchmarks capture only a slice of what LLMs can do. In the real world, your chatbot might need to juggle domain knowledge, keep track of conversation context, abide by your company’s policies, and produce fluent, non-offensive replies. No single standardized test out there fully covers that. As we’ll see in the upcoming case studies, the design and execution of a benchmark can heavily shape the picture you get of your model’s performance… and sometimes lead you astray if you’re not careful with how you measure success.

Now that we have a sense of what Llm Benchmarks are designed to accomplish (and where they might fall short), let’s explore a couple of examples to see how people actually build and use them in practice — with mixed results!

Case Study #1: Leetcode as an LLM Benchmark

As a student in the tech space, the word “Leetcode” popping up during my search for cool benchmarks raised by blood pressure by a statistically significant amount. Unlike Leetcode, which sucks, the paper “Performance Study of LLM-Generated Code on Leetcode” was very interesting — it asks a deceptively simple question: can we use Leetcode to benchmark LLM code generation? Their findings reveal both the promise and pitfalls of this approach.

The Benchmark Design

The researchers built a three-stage validation system. Local tests catch basic errors, Leetcode’s judge verifies correctness, and a custom benchmarking setup measures performance. This setup revealed something critical: benchmarking code performance is harder than it looks.

When they compared local measurements to Leetcode’s metrics, they found only a 0.28 correlation. Leetcode’s measurements showed much higher variation (0.089 vs 0.035 locally). Even worse, Leetcode’s rankings proved unstable — identical solutions could drop from the 77th to 54th percentile just based on submission timing.

A Performance Study of LLM-Generated Code on Leetcode,” In 28th International Conference on Evaluation and Assessment in Software Engineering (EASE 2024), Salerno, Italy (2024)

The Real Problems

Three major issues emerged that challenge Leetcode’s viability as a benchmark:

Data Contamination: Using public problems risks LLMs having seen the solutions during training. The researchers had to use only problems from 2023 to mitigate this.

Platform Instability: Leetcode’s metrics drift over time — memory measurements showed a -0.24 correlation with test date. This makes reproducible benchmarking nearly impossible.

Measurement Reliability: The weak correlation between local and platform measurements raises questions about what we’re actually testing.

What It Means for LLM Benchmarking

This study doesn’t just critique Leetcode — it highlights what we need in a code generation benchmark: reproducible measurements, reliable performance metrics, and guaranteed training-test separation. Until we have platforms built specifically for this purpose, we need to be extremely cautious about using competition platforms as benchmarks.

So! We know that not all benchmarks are viable benchmarks — what about a more mainstream one?

Case Study #2: SuperGLUE — Building a Better Language Understanding Benchmark

The SuperGLUE paper tackles a fascinating problem in AI benchmarking: what do you do when models get too good at your tests? When GLUE became insufficient (with models surpassing human performance), the researchers had to rethink how we measure language understanding.

The Benchmark Design

SuperGLUE’s core innovation is its task selection methodology. The researchers collected task proposals from the NLP community and filtered them through a rigorous process: each task needed clear evaluation metrics, public training data, and — most importantly — significant headroom between machine and human performance.

This resulted in eight tasks (I’ve simplified the table from the document here, it’s a little less readable but you should get the sense of what the questions are asking):

SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems, In 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada (2019)

What makes these tasks special is their diversity in format. Unlike GLUE’s focus on sentence classification, SuperGLUE includes coreference resolution, reading comprehension, and more com plex reasoning tasks. Each task measures different aspects of language understanding while maintaining clear, quantifiable metrics.


Part 2: Let’s Build a Physical Reasoning Benchmark: To Cheat at Escape Rooms

After looking at some benchmarks like SuperGLUE and Leetcode, I had an idea: what if we tested LLMs on something completely different — physical reasoning… through escape room puzzles?

It’s a pretty valid idea — escape rooms poses possibilities and consequences for failure — screw up one too many puzzles, and your friends will think you’re pretty stupid, and relegate you to spectator duty. Luckily for us however, they (or the poor employees) don’t know that you can sneak a phone into an escape room — and you know just who to ask for the answers. Today, LLMs face off against the puzzles of a physical escape room.

Note: This is NOT a rigorous academic benchmark (please don’t cite this in papers, why would you even want to do that?), or even close to it, and it’s just supposed to be a fun way to test LLM benchmarking and evaluation. Please do not destroy my prompts, I am aware they are bad.

Why Physical Reasoning?

For real, though… most LLM benchmarks focus on linguistic tasks (like SuperGLUE) or code generation (like Leetcode). And for good reason — these are well-defined domains with clear evaluation metrics. But real-world problem solving often requires understanding physical principles and their interactions. The famous “Can GPT-4 do physics?” debates usually center around mathematical problem-solving, not practical physical reasoning.

Looking at existing benchmarks taught me a few key principles:

  1. Clear evaluation metrics are crucial (from SuperGLUE’s task-specific scores)
  2. Problems should have unambiguous solutions (from HumanEval’s test cases)
  3. The benchmark should test distinct capabilities (from MMLU’s subject categories)

Designing the Problems

I settled on escape room puzzles for two reasons. First, they naturally combine physical reasoning with clear goals. Second, they have unambiguous success conditions — either you solve it through the intended way, or you don’t. Third, and most importantly, they let me include “red herrings” — irrelevant items that test if the LLM can identify what matters physically. Fourth, I just really like doing escape rooms (did I mention that already?),

I am aware that this is more than two reasons, but if LLMs can’t count how many rs’ there are in strawberry, I’m allowed to mess up once in a while too.

Here’s how I structured the five core problems:

Fluid Dynamics (FLUID_001) (Ping pong ball stuck in a tube)

  • Tests understanding of buoyancy and fluid displacement
  • Inspired by classic physics problems but in practical context
  • Includes intentionally irrelevant items (like squishy food models)

Light Properties (UV_001) (UV light on a push numebr lock)

  • Tests understanding of UV fluorescence and material properties
  • Combines multiple physical principles (light, material science)
  • Requires understanding of environmental conditions

Mechanical Understanding (CIPHER_001) (A cipher ring)

  • Tests spatial reasoning and mechanical alignment
  • No red herrings — tests for correlating a dial to a cypher wheel
  • Requires understanding rotational symmetry

Force Application (VAC_001) (Can stuck in hole)

  • Tests understanding of vacuum forces and surface adhesion
  • Multiple possible solution approaches
  • Requires understanding force multiplication

Collaborative Physics (COLLAB_001) (Can two people shimmy a key?)

  • Tests understanding of physical constraints in multi-agent scenarios
  • Requires combining multiple physical principles
  • Tests understanding of tool creation and friction

Sounds really fancy… but it’s just some basic physical puzzles. You can access them on my GitHub.

The Technical Part

The benchmark implementation has three main components:

Problem Definition Layer

Problems are defined in a structured JSON format that enforces consistent evaluation:

{
    "problem_id": "FLUID_001",
    "setup": {
        "scenario": "A ping pong ball is at the bottom of a narrow tube...",
        "available_items": ["bottle of water", "squishy food models"...],
        "constraints": ["tube too narrow for manual retrieval"]
    },
    "physical_principles": ["buoyancy", "fluid displacement"],
    "red_herrings": ["squishy food models", "milk carton"],
    "solution": {
        "steps": ["pour water into tube", "allow ball to float"],
        "key_insights": ["water displaces air", "ping pong ball less dense"]
    }
}

This structure draws from SuperGLUE’s design — each component is clearly separated and machine-readable. The physical_principles field explicitly lists what’s being tested, while red_herrings helps in scoring the LLM’s ability to ignore irrelevant information.

2. Evaluation Framework

The evaluation system uses Python’s asyncio for concurrent testing, with retry logic for a little bit more API stability:

@retry(stop=stop_after_attempt(3), wait=wait_exponential(min=1, max=10))
async def evaluate_response(self, criteria: JudgingCriteria) -> Dict:
    """Evaluate a model's response using GPT-4 as judge."""
    async with aiohttp.ClientSession() as session:
        # ... evaluation logic

The scoring system looks at three components:

Physical Understanding Score (PUS) ∈ [0,2]

  • Measures understanding of relevant physical principles
  • Calculated as normalized sum of demonstrated principles

Solution Path Score (SPS) ∈ [0,2]

  • Evaluates completeness and correctness of solution steps
  • Considers practical feasibility of proposed solutions

Red Herring Handling (RHH) ∈ {0,1}

  • A Binary score for avoiding irrelevant items
  • Tests ability to focus on physically relevant factors

And yes, there are also so many other scoring methods, better and worse, that could be used! For example, RHH could be about how many irrelevant items are used in the solution, or it could be a measure of how viable the use is… the point is that picking these metrics are often times pretty arbitrary, but are very very important to making your benchmark is credible, which mine is very much not.

Additionally, I did not want to rewrite any code after. Sue me.

3. Model Interface Layer

The benchmark supports multiple LLM backends through a common interface:

class ModelInterface:
    """Interface for different LLM APIs."""
    async def generate_response(self, prompt: str) -> str:
        raise NotImplementedError
class GPT4Interface(ModelInterface):
    async def generate_response(self, prompt: str) -> str:
        # GPT-4 specific implementation
class ClaudeInterface(ModelInterface):
    async def generate_response(self, prompt: str) -> str:
        # Claude specific implementation

Two models… I can’t really afford any more, please understand.

Let’s Test It!

So after some refilling of API balances, I was ready to go. I ran the benchmark, and after totally not wasting credits on like, 5 different errors, the results were in. And they were… interesting! I used GPT4 and Claude 3.5 Haiku (I am aware these are not similarly spec-ed models) as our victims.

The Challenge: A can is stuck in a hole that almost exactly matches its volume. A combination is written on the bottom of the can that needs to be retrieved.

The inspiration
"VAC_001": {
        "problem_id": "VAC_001",
        "setup": {
            "available_items": [
                "ruler",
                "notebook",
                "suction cup",
                "UV light"
            ],
            "constraints": [
                "can fits too tightly for finger grip",
                "hole is same diameter as can",
                "can is clean and dry"
            ]
        },
        "physical_principles": [
            "vacuum suction",
            "surface adhesion",
            "force multiplication"
        ],
        "red_herrings": [
            "ruler",
            "notebook",
            "UV light"
        ],
        "solution": {
            "steps": [
                "inspect suction cup and can surface",
                "ensure can surface is clean",
                "firmly press suction cup onto can top",
                "create vacuum seal",
                "pull can out using suction cup"
            ],
            "key_insights": [
                "suction cup creates vacuum seal",
                "vacuum force allows extraction",
                "other items don't provide necessary force"
            ]
        }

GPT-4’s Solution

Physical Principles: The main physical principle involved in this solution is the concept of air pressure differential, which is the same principle that allows a suction cup to stick to a surface. Air pressure is higher on one side of the suction cup (outside) compared to the other (inside), and this difference in pressure causes the cup to stick to the surface.

Solution Steps:

  1. Attach the suction cup to the top of the can. The flat surface of the can should provide a good contact for the suction cup to adhere to.
  2. Once the suction cup is firmly attached, gently pull upwards. The air pressure differential should cause the suction cup (and thus the can) to be lifted out of the hole.
  3. Once the can is out of the hole, remove the suction cup from the top of the can.
  4. Use the UV light to illuminate the combination written on the bottom of the can. Some inks only become visible under UV light, which is why this step is necessary.
  5. Write down the combination on the notebook to ensure it’s not forgotten.

Required Items:

  • Suction cup: This is required to adhere to the can and pull it out of the hole, as the can fits too tightly for finger grip.
  • UV light: This is required to reveal the combination written on the bottom of the can, as it may be written in ink that only becomes visible under UV light.
  • Notebook: This is required to record the combination once it’s visible, ensuring it’s not forgotten.”

Not bad at all! GPT-4 manages to figure out the physical principle behind the solution (even if it is very wordy), and manages to step-by-step describe how to solve the problem. It does end up using the UV light and notebook, but that’s more of a problem on my end, not punishing the use of unnecessary items in the prompt enough. Also, the uses for these items seem decently reasonable… so why not!

Haiku-3.5’s Solution

Physical Principles: The key physical principles involved in this scenario are the concept of volume and the use of suction to extract the can from the hole. Since the can fits almost exactly into the hole, it is likely that the can is stuck due to the tight fit and the lack of any air gaps or space between the can and the hole. This tight fit creates a vacuum-like effect, making it difficult to grip and pull the can out.

Solution Steps:

  1. Measure the diameter of the can using the ruler. This will help determine the size of the hole and confirm that the can fits tightly within it.
  2. Inspect the bottom of the can to locate the combination written on it. Use the UV light to illuminate the bottom of the can and make the combination visible, if it is not already clearly visible.
  3. Attach the suction cup to the top of the can. The suction cup will create a temporary seal on the can’s surface, allowing you to grip and pull it out of the hole.
  4. Gently pull the can upwards using the suction cup. The suction force should be strong enough to overcome the vacuum-like effect and extract the can from the hole.
  5. Once the can is out, record the combination written on the bottom of the can in the notebook.

Required Items:

  • Ruler: To measure the diameter of the can and confirm the tight fit in the hole.
  • UV light: To illuminate the bottom of the can and make the combination visible, if it is not already clearly visible.
  • Suction cup: To create a temporary seal on the can’s surface and provide the necessary grip to pull the can out of the hole.
  • Notebook: To record the combination written on the bottom of the can.

On the other hand, Haiku’s solution was.. certainly a solution of all time.

First, it wants to measure the diameter of the can using the ruler??? To determine the size of the hole and confirm that the can fits tightly within it? Why would we need to do this? And do we need a ruler for that?

Second, it tells us to inspect the bottom of the can to locate the combination, when the entire problem is about not being able to pull the can out of the hole conventionally. This might just be an issue of order, but now I truly understand my friends’ feelings whenever I would tell them “just fix it man” to their numerous problems.

But it eventually does get the solution. So… not the worst.

Here’s a fancy radar graph of the results!

We see that both models are pretty similar in their capabilities, with GPT-4 being slightly better in physical understanding and solution path, and Haiku being slightly better in red herring handling. Overall though, both models kind of suck. Dang.

There are also only… 5 questions.

If you’d like to see the full breadth of questions, they’re on my GitHub.

LLM-as-a-Judge

By the way, the method I used to generate the evaluations, LLM-as-a-judge, has gained significant traction in the AI community, particularly after the work of Zheng et al. in their 2023 paper “Judging LLM-as-a-Judge.” The technique has proven remarkably effective, achieving over 80% agreement with human evaluators in tasks ranging from code assessment to dialogue quality evaluation!

Here’s where my experiment gets kind of cool (arguably, maybe, subjectively) — I used this methodology and had GPT-4 judge other LLMs’ physical reasoning abilities. Yes, I’m using an AI to judge other AIs.

Why does this work? Well, judging a response is actually a simpler task than generating one. When GPT-4 generates a solution to a physical puzzle, it needs to:

  • Understand the physical principles involved
  • Plan a sequence of steps
  • Consider all constraints
  • Generate a coherent explanation

But when judging, it only needs to check if specific criteria are met in an existing solution. The evaluation prompt is very focused:

def _create_evaluation_prompt(self, criteria: JudgingCriteria) -> str:
    return f"""You are an expert judge evaluating an LLM's understanding of physical reasoning puzzles.
Evaluate based on three criteria:
2. Physical Understanding Score (0-2): Does the solution correctly apply relevant physical principles?
3. Solution Path Score (0-2): Are the steps complete and feasible?
4. Red Herring Handling (0-1): Does it avoid using irrelevant items?
Scenario: {criteria.scenario}
Physical Principles Required: {criteria.correct_principles}
Solution Given: {criteria.model_response}
"""

To validate this approach, I followed the validation framework suggested by Zheng et al., performing spot-checks of GPT-4’s evaluations against my own judgments. Surprisingly (or perhaps unsurprisingly, given the broader research on LLM evaluation), it was remarkably consistent in identifying both correct physical understanding and flawed reasoning.

Is this perfect? Absolutely not. There’s something philosophically weird about using one LLM to evaluate another. But in practice, it can work surprisingly well — just like how I moan and groan about the visual presentation of a dish on Masterchef, while setting my kitchen aflame trying to microwave a hot dog.

What I Learned

Building this benchmark taught me several things about benchmark design:

Clear Metrics Matter: Even for complex tasks like physical reasoning, you need unambiguous scoring criteria.

Red Herrings Are Powerful: Including irrelevant items reveals a lot about an LLM’s reasoning process.

Context Control is Hard: Ensuring LLMs don’t “hallucinate” additional physical context is challenging.

Is this a perfect benchmark? Not even close. Please don’t rub it in. Is it scientifically rigorous? Definitely not. But it’s been a fascinating exploration into an aspect of LLM capabilities, and sometimes the best we can learn can come from just trying things out and seeing what happens.

Now, if you’ll excuse me, I will be sneaking in a phone with an internet connection into my next escape room, for reasons that I am legally unmotivated to disclose.

[1] L. Zheng, W.-L. Chiang, Y. Sheng, S. Zhuang, Z. Wu, Y. Zhuang, Z. Lin, Z. Li, D. Li, E. P. Xing, H. Zhang, J. E. Gonzalez, I. Stoica, “Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena,” Proceedings of the 37th Conference on Neural Information Processing Systems (NeurIPS 2023), Datasets and Benchmarks Track (2023)

[2] T. Coignion, C. Quinton, R. Rouvoy, “A Performance Study of LLM-Generated Code on Leetcode,” In 28th International Conference on Evaluation and Assessment in Software Engineering (EASE 2024), Salerno, Italy (2024)

[3] A. Wang, Y. Pruksachatkun, N. Nangia, A. Singh, J. Michael, F. Hill, O. Levy, S. R. Bowman, “SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems,” In 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada (2019)

[5] DeepSeek-AI, D. Guo, D. Yang, H. Zhang, J. Song, R. Zhang, R. Xu, Q. Zhu, S. Ma, P. Wang, X. Bi, X. Zhang, X. Yu, Y. Wu, Z.F. Wu, Z. Gou, Z. Shao, Z. Li, Z. Gao et al., “DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning,” arXiv preprint arXiv:2501.12948 (2025)

[6] Unless otherwise stated, all images are created by the author.

Shape
Shape
Stay Ahead

Explore More Insights

Stay ahead with more perspectives on cutting-edge power, infrastructure, energy,  bitcoin and AI solutions. Explore these articles to uncover strategies and insights shaping the future of industries.

Shape

Fluent Bit vulnerabilities could enable full cloud takeover

Attackers could flood monitoring systems with false or misleading events, hide alerts in the noise, or even hijack the telemetry stream entirely, Katz said. The issue is now tracked as CVE-2025-12969 and awaits a severity valuation. Almost equally troubling are other flaws in the “tag” mechanism, which determines how the records are

Read More »

NFL, AWS drive football modernization with cloud, AI

AWS Next Gen Stats: Initially used for player participation tracking (replacing manual photo-taking), Next Gen Stats uses sensors to capture center-of-mass and contact information, which is then used to generate performance insights. Computer vision: Computer vision was initially insufficient, but the technology has improved greatly over the past few years.

Read More »

Oil Creeps Higher Alongside Global Equities

Oil crept higher alongside global equities, bouncing off a one-month low as the White House signaled optimism about a peace deal between Russia and Ukraine, which could bring back Moscow’s barrels into an already saturated market.  West Texas Intermediate advanced 1.2% to settle above $58 a barrel, recouping most of the previous day’s losses. Volumes are still trending lower ahead of Thursday’s Thanksgiving holiday in the US.  Steve Witkoff, US President Donald Trump’s envoy, will lead a delegation for talks in Russia next week on ending the nearly four-year long war, a Kremlin official said. Ukrainian leader’s chief of staff said negotiations in Geneva had laid a “good foundation.” Yet any peace deal still face the same obstacles as in the past: What satisfies Ukraine is likely a deal-breaker for Russia, and vice versa. Much of Russia’s oil and fuel is subject to heavy Western sanctions, with US restrictions on the two biggest producers kicking in last week. However, China, India and Turkey have been eager buyers of the discounted crude, so the impact on global prices from any lifting of curbs is hard to gauge. “Minute adjustments between the US, Russia, Ukraine and the EU on proposed peace deals have been carefully digested by the market,” Standard Chartered analysts including Emily Ashford wrote in a note. “Any positive signs of collaboration or agreement have resulted in short-term sell-offs, while the dialing-back of enthusiasm has bolstered prices.” In the US, meanwhile, the Energy Information Administration reported on Wednesday that overall crude inventories climbed by 2.8 million barrels, while gasoline and distillate inventories also expanded. That did little to assuage growing oversupply fears. Oil has retreated by more than a fifth since the middle of June as the Organization of the Petroleum Exporting Countries and its allies restored barrels, while producers outside of

Read More »

Venture Global Signs 20 Year LNG Deal With Tokyo Gas

Venture Global Inc. signed an agreement with Tokyo Gas Co. to supply the utility with liquefied natural gas for 20 years, the US producer’s fourth long-term contract with a Japanese company. The deal will cover 1 million tons of LNG a year from 2030, according to a statement on Wednesday. Venture Global has signed supply agreements with companies from Japan to Spain over the past six months totaling 7.75 million tons a year, the US producer added. Venture Global is constructing its third export facility — CP2 — in Louisiana, and operates two others, Plaquemines LNG and Calcasieu Pass. The Arlington, Virginia-based producer is one of the largest US suppliers, boosting the the country’s position as the biggest exporter of the fuel globally. The company has signed recent deals with Japan’s Mitsui & Co., Spain’s Naturgy Energy Group SA and Greek entity Atlantic-See LNG Trade SA. Venture Global is currently facing arbitration from customers including Shell Plc over a dispute around the start of long-term contracts for supply from Calcasieu Pass. The facility began exporting cargoes in 2022, but contracts did not begin until earlier this year. A tribunal panel ruled in favor of Venture Global in regards to Shell, but the oil giant is now appealing the ruling in New York Supreme Court. WHAT DO YOU THINK? Generated by readers, the comments included herein do not reflect the views and opinions of Rigzone. All comments are subject to editorial review. Off-topic, inappropriate or insulting comments will be removed.

Read More »

Saudi Aramco Is Said to Pick Citi for Oil Storage Terminals Stake Sale

Saudi Aramco has chosen Citigroup Inc. to help arrange a potential multibillion-dollar stake sale in its oil export and storage terminals business, according to people familiar with the matter. The US investment bank was selected in recent days after a pitching process that drew proposals from several other Wall Street lenders, the people said, asking not to be identified as the matter is private.  The mandate is a win for Citigroup, whose Chief Executive Officer Jane Fraser has made a renewed effort to win business from large corporates and sovereign wealth funds in the Middle East. Aramco had tapped JPMorgan Chase & Co. as a sell-side adviser when it previously sold stakes in its oil and gas pipeline infrastructure in separate transactions.  The Saudi oil giant is expected to kick-off a formal sale process as early as next year and is likely to see interest from large infrastructure funds, the people said. Discussions are at an early stage and no final decisions have been made on the timing or structure of the transaction, they said. Representatives for Citigroup and Aramco declined to comment. Aramco is considering options including selling an equity stake in the business, Bloomberg News reported this week. It aims to raise billions of dollars from such a sale, people familiar with the matter said at the time.  The plans are part of a broader attempt by the firm to sell a range of assets, including potentially part of its real estate portfolio.  Oil prices have dropped about 16% this year and while the impact of that drop on Aramco’s earnings has been tempered by higher output, the firm has delayed some projects and looked to sell assets to free up cash for investments.  The deals now being considered would mark a step up from previous transactions that were focused on stakes

Read More »

BlackRock Looks to Double Saudi Investments

BlackRock Inc. aims to rapidly grow its investments in Saudi Arabia and the wider Middle East in the next few years as it looks to tap into a rush of activity in areas from infrastructure to artificial intelligence. The world’s largest asset manager has already invested more than $35 billion in the kingdom across equities, fixed income and infrastructure, and now has four investment teams in Riyadh focused on strategies across the Middle East, according to Kashif Riaz, who heads BlackRock’s Financial Markets Advisory business in the Middle East and its Riyadh-based investment management platform. “We think we’ve just gotten started with the theme of the Middle East as an investment destination,” he said in an interview in the Saudi capital on Monday night. When asked about the expected level of future Saudi investments, he said that “double to triple is kind of the range I would talk about.” Riaz sees the strongest opportunities in infrastructure as Saudi Arabia shells out hundreds of billions of dollars on projects to develop the non-oil economy and serve a growing population. The kingdom is, for example, expanding the Riyadh metro, building one of the world’s largest airports and rushing to build data centers through its new AI champion Humain. “The bulk of capital deployment has been in energy infrastructure but I think that’ll broaden to transportation, things around digital infrastructure, data centers, et cetera,” Riaz said. That suggests more Saudi deals to come for BlackRock and its Global Infrastructure Partners division, which recently led an $11 billion deal involving Saudi Aramco’s natural gas facilities. The unit also recently partnered with investors including Abu Dhabi’s MGX to buy Aligned Data Centers in a $40 billion deal, as BlackRock and Middle East nations race to claim a stake in the global AI boom.  BlackRock established itself in Saudi

Read More »

Cheapest US Stations Drop Gas to Sub-$2 Ahead of Thanksgiving

GasBuddy announced, in a blog posted on its website on Tuesday, that the United States’ cheapest stations dropped the gasoline price to $1.99 per gallon ahead of Thanksgiving for the first time in four years. “GasBuddy … today reported the first sub-$2 per gallon gas prices available without discounts or as part of a temporary promotion, making them the lowest prices seen in the U.S. since 2021,” GasBuddy said in the blog.  The company noted in the blog that four stations located in Midwest City, Oklahoma, stood at $1.99 per gallon for regular unleaded gasoline early Monday morning on November 24. “It’s pretty compelling to see gas prices this low, falling ahead of Thanksgiving, and it signals what more Americans could experience in the coming months,” Patrick De Haan, head of petroleum analysis at GasBuddy, said in the blog. “Lower seasonal demand, falling oil prices, and rising OPEC output are all pushing prices down. While a few stations have recently dipped below $2 through temporary offers and promotions, this marks the first time we’ve seen a regular sub-$2 price,” he added. “Prices at this level may fluctuate, but more locations in low-cost states like Texas, Mississippi, and others across the Gulf region are likely to follow before the typical spring rebound we’ll likely see in 2026,” De Haan continued. In the blog, GasBuddy stated that falling oil prices and healthy refinery output have combined to ease pressure on consumers at the pump over the past several months, highlighting that West Texas Intermediate (WTI) crude oil recently traded near $58 per barrel. Fewer Americans Plan to Hit the Road In a separate blog posted on GasBuddy’s website on November 18, GasBuddy warned that fewer Americans plan to hit the road for Thanksgiving this year, “even as gas prices remain near the

Read More »

Rystad Warns ‘Volatility Far from Over’ for Energy Markets

In a market update sent to Rigzone by the Rystad Energy team late Tuesday, Rystad Energy’s Head of Geopolitical Analysis, Jorge Leon, warned that “volatility is far from over” for energy markets. That Rystad update highlighted that “U.S. and Ukrainian officials said they have reached broad agreement on the basic terms of a peace proposal” but added that “questions remain regarding crucial details and Russia’s willingness to accept it”. “Energy markets responded to the potential geopolitical breakthrough with a dip in both oil and gas prices, followed by a quick reversal,” the update stated, noting that “U.S. talks with Russian officials are reportedly ongoing in Abu Dhabi”. Leon pointed out in the update that “the news of a revised peace proposal and ongoing negotiations, with Ukraine’s agreement in principle … prompted an immediate reaction across energy markets”. He highlighted that oil prices initially fell by around two percent, and that European natural gas prices dropped by a similar amount, “as traders priced in a lower risk of further escalation”. “However, those early declines have already partially reversed, underscoring how sensitive markets remain to every headline,” Leon said in Tuesday’s update. The Rystad Energy head went on to state in the update that, “crucially, the devil is still in the details”. “What are being described as ‘minor points to be resolved’ will determine whether this proposal is accepted by both sides, especially as hostilities have continued in the form of deadly strikes,” Leon said. “Key issues, including territorial arrangements, security guarantees and implementation timelines, could require several additional rounds of revision. Above all, while some outlets have reported on Ukraine’s agreement in principle, the world is still waiting for an official response by both parties,” he added. “Early signals, including comments from Russian Foreign Minister Sergei Lavrov, indicate that Moscow could

Read More »

Microsoft loses two senior AI infrastructure leaders as data center pressures mount

Microsoft did not immediately respond to a request for comment. Microsoft’s constraints Analysts say the twin departures mark a significant setback for Microsoft at a critical moment in the AI data center race, with pressure mounting from both OpenAI’s model demands and Google’s infrastructure scale. “Losing some of the best professionals working on this challenge could set Microsoft back,” said Neil Shah, partner and co-founder at Counterpoint Research. “Solving the energy wall is not trivial, and there may have been friction or strategic differences that contributed to their decision to move on, especially if they saw an opportunity to make a broader impact and do so more lucratively at a company like Nvidia.” Even so, Microsoft has the depth and ecosystem strength to continue doubling down on AI data centers, said Prabhu Ram, VP for industry research at Cybermedia Research. According to Sanchit Gogia, chief analyst at Greyhound Research, the departures come at a sensitive moment because Microsoft is trying to expand its AI infrastructure faster than physical constraints allow. “The executives who have left were central to GPU cluster design, data center engineering, energy procurement, and the experimental power and cooling approaches Microsoft has been pursuing to support dense AI workloads,” Gogia said. “Their exit coincides with pressures the company has already acknowledged publicly. GPUs are arriving faster than the company can energize the facilities that will house them, and power availability has overtaken chip availability as the real bottleneck.”

Read More »

What is Edge AI? When the cloud isn’t close enough

Many edge devices can periodically send summarized or selected inference output data back to a central system for model retraining or refinement. That feedback loop helps the model improve over time while still keeping most decisions local. And to run efficiently on constrained edge hardware, the AI model is often pre-processed by techniques such as quantization (which reduces precision), pruning (which removes redundant parameters), or knowledge distillation (which trains a smaller model to mimic a larger one). These optimizations reduce the model’s memory, compute, and power demands so it can run more easily on an edge device. What technologies make edge AI possible? The concept of the “edge” always assumes that edge devices are less computationally powerful than data centers and cloud platforms. While that remains true, overall improvements in computational hardware have made today’s edge devices much more capable than those designed just a few years ago. In fact, a whole host of technological developments have come together to make edge AI a reality. Specialized hardware acceleration. Edge devices now ship with dedicated AI-accelerators (NPUs, TPUs, GPU cores) and system-on-chip units tailored for on-device inference. For example, companies like Arm have integrated AI-acceleration libraries into standard frameworks so models can run efficiently on Arm-based CPUs. Connectivity and data architecture. Edge AI often depends on durable, low-latency links (e.g., 5G, WiFi 6, LPWAN) and architectures that move compute closer to data. Merging edge nodes, gateways, and local servers means less reliance on distant clouds. And technologies like Kubernetes can provide a consistent management plane from the data center to remote locations. Deployment, orchestration, and model lifecycle tooling. Edge AI deployments must support model-update delivery, device and fleet monitoring, versioning, rollback and secure inference — especially when orchestrated across hundreds or thousands of locations. VMware, for instance, is offering traffic management

Read More »

Networks, AI, and metaversing

Our first, conservative, view says that AI’s network impact is largely confined to the data center, to connect clusters of GPU servers and the data they use as they crunch large language models. It’s all “horizontal” traffic; one TikTok challenge would generate way more traffic in the wide area. WAN costs won’t rise for you as an enterprise, and if you’re a carrier you won’t be carrying much new, so you don’t have much service revenue upside. If you don’t host AI on premises, you can pretty much dismiss its impact on your network. Contrast that with the radical metaverse view, our third view. Metaverses and AR/VR transform AI missions, and network services, from transaction processing to event processing, because the real world is a bunch of events pushing on you. They also let you visualize the way that process control models (digital twins) relate to the real world, which is critical if the processes you’re modeling involve human workers who rely on their visual sense. Could it be that the reason Meta is willing to spend on AI, is that the most credible application of AI, and the most impactful for networks, is the metaverse concept? In any event, this model of AI, by driving the users’ experiences and activities directly, demands significant edge connectivity, so you could expect it to have a major impact on network requirements. In fact, just dipping your toes into a metaverse could require a major up-front network upgrade. Networks carry traffic. Traffic is messages. More messages, more traffic, more infrastructure, more service revenue…you get the picture. Door number one, to the AI giant future, leads to nothing much in terms of messages. Door number three, metaverses and AR/VR, leads to a message, traffic, and network revolution. I’ll bet that most enterprises would doubt

Read More »

Microsoft’s Fairwater Atlanta and the Rise of the Distributed AI Supercomputer

Microsoft’s second Fairwater data center in Atlanta isn’t just “another big GPU shed.” It represents the other half of a deliberate architectural experiment: proving that two massive AI campuses, separated by roughly 700 miles, can operate as one coherent, distributed supercomputer. The Atlanta installation is the latest expression of Microsoft’s AI-first data center design: purpose-built for training and serving frontier models rather than supporting mixed cloud workloads. It links directly to the original Fairwater campus in Wisconsin, as well as to earlier generations of Azure AI supercomputers, through a dedicated AI WAN backbone that Microsoft describes as the foundation of a “planet-scale AI superfactory.” Inside a Fairwater Site: Preparing for Multi-Site Distribution Efficient multi-site training only works if each individual site behaves as a clean, well-structured unit. Microsoft’s intra-site design is deliberately simplified so that cross-site coordination has a predictable abstraction boundary—essential for treating multiple campuses as one distributed AI system. Each Fairwater installation presents itself as a single, flat, high-regularity cluster: Up to 72 NVIDIA Blackwell GPUs per rack, using GB200 NVL72 rack-scale systems. NVLink provides the ultra-low-latency, high-bandwidth scale-up fabric within the rack, while the Spectrum-X Ethernet stack handles scale-out. Each rack delivers roughly 1.8 TB/s of GPU-to-GPU bandwidth and exposes a multi-terabyte pooled memory space addressable via NVLink—critical for large-model sharding, activation checkpointing, and parallelism strategies. Racks feed into a two-tier Ethernet scale-out network offering 800 Gbps GPU-to-GPU connectivity with very low hop counts, engineered to scale to hundreds of thousands of GPUs without encountering the classic port-count and topology constraints of traditional Clos fabrics. Microsoft confirms that the fabric relies heavily on: SONiC-based switching and a broad commodity Ethernet ecosystem to avoid vendor lock-in and accelerate architectural iteration. Custom network optimizations, such as packet trimming, packet spray, high-frequency telemetry, and advanced congestion-control mechanisms, to prevent collective

Read More »

Land & Expand: Hyperscale, AI Factory, Megascale

Land & Expand is Data Center Frontier’s periodic roundup of notable North American data center development activity, tracking the newest sites, land plays, retrofits, and hyperscale campus expansions shaping the industry’s build cycle. October delivered a steady cadence of announcements, with several megascale projects advancing from concept to commitment. The month was defined by continued momentum in OpenAI and Oracle’s Stargate initiative (now spanning multiple U.S. regions) as well as major new investments from Google, Meta, DataBank, and emerging AI cloud players accelerating high-density reuse strategies. The result is a clearer picture of how the next wave of AI-first infrastructure is taking shape across the country. Google Begins $4B West Memphis Hyperscale Buildout Google formally broke ground on its $4 billion hyperscale campus in West Memphis, Arkansas, marking the company’s first data center in the state and the anchor for a new Mid-South operational hub. The project spans just over 1,000 acres, with initial site preparation and utility coordination already underway. Google and Entergy Arkansas confirmed a 600 MW solar generation partnership, structured to add dedicated renewable supply to the regional grid. As part of the launch, Google announced a $25 million Energy Impact Fund for local community affordability programs and energy-resilience improvements—an unusually early community-benefit commitment for a first-phase hyperscale project. Cooling specifics have not yet been made public. Water sourcing—whether reclaimed, potable, or hybrid seasonal mode—remains under review, as the company finalizes environmental permits. Public filings reference a large-scale onsite water treatment facility, similar to Google’s deployments in The Dalles and Council Bluffs. Local governance documents show that prior to the October announcement, West Memphis approved a 30-year PILOT via Groot LLC (Google’s land assembly entity), with early filings referencing a typical placeholder of ~50 direct jobs. At launch, officials emphasized hundreds of full-time operations roles and thousands

Read More »

The New Digital Infrastructure Geography: Green Street’s David Guarino on AI Demand, Power Scarcity, and the Next Phase of Data Center Growth

As the global data center industry races through its most frenetic build cycle in history, one question continues to define the market’s mood: is this the peak of an AI-fueled supercycle, or the beginning of a structurally different era for digital infrastructure? For Green Street Managing Director and Head of Global Data Center and Tower Research David Guarino, the answer—based firmly on observable fundamentals—is increasingly clear. Demand remains blisteringly strong. Capital appetite is deepening. And the very definition of a “data center market” is shifting beneath the industry’s feet. In a wide-ranging discussion with Data Center Frontier, Guarino outlined why data centers continue to stand out in the commercial real estate landscape, how AI is reshaping underwriting and development models, why behind-the-meter power is quietly reorganizing the U.S. map, and what Green Street sees ahead for rents, REITs, and the next wave of hyperscale expansion. A ‘Safe’ Asset in an Uncertain CRE Landscape Among institutional investors, the post-COVID era was the moment data centers stepped decisively out of “niche” territory. Guarino notes that pandemic-era reliance on digital services crystallized a structural recognition: data centers deliver stable, predictable cash flows, anchored by the highest-credit tenants in global real estate. Hyperscalers today dominate new leasing and routinely sign 15-year (or longer) contracts, a duration largely unmatched across CRE categories. When compared with one-year apartment leases, five-year office leases, or mall anchor terms, the stability story becomes plain. “These are AAA-caliber companies signing the longest leases in the sector’s history,” Guarino said. “From a real estate point of view, that combination of tenant quality and lease duration continues to position the asset class as uniquely durable.” And development returns remain exceptional. Even without assuming endless AI growth, the math works: strong demand, rising rents, and high-credit tenants create unusually predictable performance relative to

Read More »

Microsoft will invest $80B in AI data centers in fiscal 2025

And Microsoft isn’t the only one that is ramping up its investments into AI-enabled data centers. Rival cloud service providers are all investing in either upgrading or opening new data centers to capture a larger chunk of business from developers and users of large language models (LLMs).  In a report published in October 2024, Bloomberg Intelligence estimated that demand for generative AI would push Microsoft, AWS, Google, Oracle, Meta, and Apple would between them devote $200 billion to capex in 2025, up from $110 billion in 2023. Microsoft is one of the biggest spenders, followed closely by Google and AWS, Bloomberg Intelligence said. Its estimate of Microsoft’s capital spending on AI, at $62.4 billion for calendar 2025, is lower than Smith’s claim that the company will invest $80 billion in the fiscal year to June 30, 2025. Both figures, though, are way higher than Microsoft’s 2020 capital expenditure of “just” $17.6 billion. The majority of the increased spending is tied to cloud services and the expansion of AI infrastructure needed to provide compute capacity for OpenAI workloads. Separately, last October Amazon CEO Andy Jassy said his company planned total capex spend of $75 billion in 2024 and even more in 2025, with much of it going to AWS, its cloud computing division.

Read More »

John Deere unveils more autonomous farm machines to address skill labor shortage

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Self-driving tractors might be the path to self-driving cars. John Deere has revealed a new line of autonomous machines and tech across agriculture, construction and commercial landscaping. The Moline, Illinois-based John Deere has been in business for 187 years, yet it’s been a regular as a non-tech company showing off technology at the big tech trade show in Las Vegas and is back at CES 2025 with more autonomous tractors and other vehicles. This is not something we usually cover, but John Deere has a lot of data that is interesting in the big picture of tech. The message from the company is that there aren’t enough skilled farm laborers to do the work that its customers need. It’s been a challenge for most of the last two decades, said Jahmy Hindman, CTO at John Deere, in a briefing. Much of the tech will come this fall and after that. He noted that the average farmer in the U.S. is over 58 and works 12 to 18 hours a day to grow food for us. And he said the American Farm Bureau Federation estimates there are roughly 2.4 million farm jobs that need to be filled annually; and the agricultural work force continues to shrink. (This is my hint to the anti-immigration crowd). John Deere’s autonomous 9RX Tractor. Farmers can oversee it using an app. While each of these industries experiences their own set of challenges, a commonality across all is skilled labor availability. In construction, about 80% percent of contractors struggle to find skilled labor. And in commercial landscaping, 86% of landscaping business owners can’t find labor to fill open positions, he said. “They have to figure out how to do

Read More »

2025 playbook for enterprise AI success, from agents to evals

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More 2025 is poised to be a pivotal year for enterprise AI. The past year has seen rapid innovation, and this year will see the same. This has made it more critical than ever to revisit your AI strategy to stay competitive and create value for your customers. From scaling AI agents to optimizing costs, here are the five critical areas enterprises should prioritize for their AI strategy this year. 1. Agents: the next generation of automation AI agents are no longer theoretical. In 2025, they’re indispensable tools for enterprises looking to streamline operations and enhance customer interactions. Unlike traditional software, agents powered by large language models (LLMs) can make nuanced decisions, navigate complex multi-step tasks, and integrate seamlessly with tools and APIs. At the start of 2024, agents were not ready for prime time, making frustrating mistakes like hallucinating URLs. They started getting better as frontier large language models themselves improved. “Let me put it this way,” said Sam Witteveen, cofounder of Red Dragon, a company that develops agents for companies, and that recently reviewed the 48 agents it built last year. “Interestingly, the ones that we built at the start of the year, a lot of those worked way better at the end of the year just because the models got better.” Witteveen shared this in the video podcast we filmed to discuss these five big trends in detail. Models are getting better and hallucinating less, and they’re also being trained to do agentic tasks. Another feature that the model providers are researching is a way to use the LLM as a judge, and as models get cheaper (something we’ll cover below), companies can use three or more models to

Read More »

OpenAI’s red teaming innovations define new essentials for security leaders in the AI era

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More OpenAI has taken a more aggressive approach to red teaming than its AI competitors, demonstrating its security teams’ advanced capabilities in two areas: multi-step reinforcement and external red teaming. OpenAI recently released two papers that set a new competitive standard for improving the quality, reliability and safety of AI models in these two techniques and more. The first paper, “OpenAI’s Approach to External Red Teaming for AI Models and Systems,” reports that specialized teams outside the company have proven effective in uncovering vulnerabilities that might otherwise have made it into a released model because in-house testing techniques may have missed them. In the second paper, “Diverse and Effective Red Teaming with Auto-Generated Rewards and Multi-Step Reinforcement Learning,” OpenAI introduces an automated framework that relies on iterative reinforcement learning to generate a broad spectrum of novel, wide-ranging attacks. Going all-in on red teaming pays practical, competitive dividends It’s encouraging to see competitive intensity in red teaming growing among AI companies. When Anthropic released its AI red team guidelines in June of last year, it joined AI providers including Google, Microsoft, Nvidia, OpenAI, and even the U.S.’s National Institute of Standards and Technology (NIST), which all had released red teaming frameworks. Investing heavily in red teaming yields tangible benefits for security leaders in any organization. OpenAI’s paper on external red teaming provides a detailed analysis of how the company strives to create specialized external teams that include cybersecurity and subject matter experts. The goal is to see if knowledgeable external teams can defeat models’ security perimeters and find gaps in their security, biases and controls that prompt-based testing couldn’t find. What makes OpenAI’s recent papers noteworthy is how well they define using human-in-the-middle

Read More »