Oil at $60 in Chaos: What AI Can Predict in Energy

Brent and WTI crude hovering around $60/barrel feels like a typo when headlines are packed with sanctions, regional conflicts, and political crises in oil-producing states. Ten years ago, many traders would’ve assumed this mix of risks automatically meant $100 oil.

But the market isn’t behaving “as it should.” And that mismatch is exactly why this topic matters for Қазақстандағы мұнай-газ және энергия компаниялары. When price moves don’t map neatly to geopolitical events, planning production, procurement, and cash flow becomes a stress test—unless you have a better way to model reality.

Here’s my stance: geopolitics still matters, but it’s no longer the single dominant price engine. Oil prices are increasingly the output of a crowded system—inventory buffers, spare capacity, demand uncertainty, financial positioning, and policy signals. This is where жасанды интеллект (AI) stops being a buzzword and becomes a practical tool: it helps teams quantify probabilities, detect weak signals early, and run decisions through scenarios that humans can’t compute fast enough.

Why geopolitical chaos isn’t translating into higher oil prices

Answer first: Oil prices stay muted when the market believes disruptions are manageable—because inventories, spare capacity, and demand softness absorb the shock.

The “headline logic” says: conflict + sanctions + unrest = fewer barrels = higher prices. Reality is more conditional. Prices rise sharply only when traders believe there’s a sustained supply deficit that can’t be offset.

Several mechanisms can keep prices capped even when the news looks terrifying:

Buffers are bigger than people think

Commercial inventories and strategic reserves act like shock absorbers. If buyers believe stocks can cover a disruption for weeks or months, panic pricing doesn’t last.

There’s also an underappreciated buffer: product inventories (diesel, gasoline, jet fuel). If refiners and traders have product coverage, crude demand can lag even when crude supply risk rises.

Spare capacity and “risk perception” aren’t the same thing

Some producers (often OPEC core members) maintain spare capacity—barrels that can be brought online. Even if the world is tense, the market asks a colder question: Can someone backfill the missing supply quickly? If yes, the risk premium shrinks.

Demand anxiety can overpower supply anxiety

When global manufacturing slows, when China/Europe demand signals soften, or when interest rates remain restrictive, the market worries about demand destruction.

Oil is priced on expectations. If demand expectations fall faster than supply expectations, prices can slide even in geopolitical noise.

The futures market can mute the “headline premium”

Financial positioning matters. If funds are already long (betting on higher prices), new conflict headlines may not add as much incremental buying. Conversely, if risk-off sentiment dominates, traders reduce exposure broadly—even to commodities—pressuring prices.

Snippet-worthy: “Oil doesn’t rise on bad news. It rises when bad news creates an unfillable barrel gap.”

The hidden forces: what actually sets the oil price day-to-day

Answer first: Short-term prices are driven less by single events and more by the interaction of inventory levels, refinery behavior, shipping flows, and expectations embedded in futures curves.

To plan operations in Kazakhstan, it helps to treat oil as a system—not a headline.

Inventories + refinery margins = immediate demand signals

Refineries buy crude when margins justify it. If cracks (refining margins) compress due to weak product demand, refineries may run lower utilization. That reduces crude buying and caps price.

For upstream operators, this matters because refinery demand doesn’t respond instantly to geopolitical risk—it responds to product consumption and profitability.

Shipping and trade flows reroute faster than expected

Sanctions and conflicts often change routes, not just volumes. A disrupted exporter may discount barrels to keep flows moving; buyers may accept longer routes if economics work. The market adapts—sometimes within weeks.

That adaptability reduces the duration of price spikes.

Policy signals shape the long end of the curve

Energy policy—SPR releases, sanctions enforcement strictness, price caps, export restrictions, carbon policy—affects expectations. The futures curve (backwardation/contango) embeds these beliefs.

For Kazakhstan’s oil and gas leadership teams, the practical takeaway is simple: don’t run budgets on one narrative. Run them on scenarios.

Where AI fits: making sense of markets that don’t “behave”

Answer first: AI improves decisions by turning messy signals—news, shipping data, inventories, macro indicators—into probabilistic forecasts and operational scenarios.

This post is part of our series “Қазақстандағы энергия және мұнай-газ саласын жасанды интеллект қалай түрлендіріп жатыр”, and price uncertainty is one of the most tangible places AI creates value. Not by predicting a single number perfectly, but by improving the quality of choices under uncertainty.

1) Geopolitical event parsing that’s actually usable

Modern NLP models can ingest thousands of sources (official statements, local-language media, sanctions bulletins) and convert them into structured event features:

• Event type (sanctions, pipeline disruption, port closure, strike)
• Severity score
• Likely duration distribution
• Affected assets/routes
• Confidence level based on source credibility

This is better than relying on one breaking headline. Your team gets an evolving “risk map” rather than a panic spike.

2) Supply chain visibility from alternative data

In oil markets, tanker movements, port congestion, and refinery runs often tell the story before price does.

AI models can fuse:

• AIS shipping data (tanker location/speed/port calls)
• Satellite indicators (flare intensity, storage tank shadows where available)
• Customs/trade proxies
• Weather and sea-state constraints

Result: earlier detection of real disruptions versus media noise.

3) Probabilistic price bands, not single-point forecasts

Most companies get this wrong: they ask AI, “What will Brent be next month?” The better question is:

• “What’s the 80% confidence interval for Brent?”
• “What’s the probability of a move above $75 if X happens?”
• “Which input shifts our outcome the most?”

Probabilistic forecasting (quantile regression, Bayesian models, ensembles) fits oil’s reality: fat tails, regime changes, sudden jumps.

Snippet-worthy: “In commodities, the forecast number matters less than the range—and what you’ll do if the range breaks.”

4) Decision intelligence for upstream and midstream planning

AI becomes most valuable when tied to actions:

• Production optimization: choosing stable output plans under price uncertainty
• Maintenance scheduling: timing turnarounds when margin risk is lower
• Inventory strategy: crude and product stock targets based on disruption probability
• Hedging strategy support: recommending hedge ratios aligned to budget risk (not speculation)

For Kazakhstan producers, this can mean fewer reactive decisions and more disciplined playbooks.

Practical playbook for Kazakhstan oil & energy teams

Answer first: Start with 3 high-impact use cases—price risk scenarios, disruption detection, and asset optimization—and build the data foundation around them.

If you’re leading operations, trading, finance, or planning, here’s what works in practice.

Step 1: Define the decision you’re improving

AI projects fail when the goal is “better analytics.” Pick a decision with a clock and a cost.

Examples:

• Monthly production and lifting plan
• Quarterly budget assumptions and stress tests
• Spare parts and chemical procurement
• Pipeline/export routing contingencies

Write the success metric in numbers: reduce forecast error by X, cut unplanned downtime by Y%, improve cash flow stability by Z.

Step 2: Build a minimal “market + operations” feature set

You don’t need everything on day one. A strong starter set often includes:

• Brent/WTI, differentials relevant to your crude quality
• Futures curve structure (spreads), volatility
• OECD inventory proxies (where available)
• Refinery utilization proxies and product cracks
• Freight rates indicators
• FX rates relevant to costs and revenue
• Your own: downtime logs, throughput, energy use, maintenance history

The key is combining external market signals with internal operational data.

Step 3: Use scenarios that reflect your real constraints

A useful AI model respects constraints:

• Minimum stable production levels
• Water cut and reservoir limits
• Planned turnaround windows
• Export capacity and quality specs

Pair forecasts with scenario narratives:

1. Base case: stable flows, soft demand
2. Supply shock: disruption in a key region, partial backfill
3. Demand shock: macro slowdown, weaker refinery pull
4. Policy shock: stricter sanctions enforcement / SPR action

Then attach playbooks: If scenario 2 probability > 35%, do X.

Step 4: Put governance around it (or it becomes shelfware)

For LEADS and real adoption, governance is the difference between a pilot and a system.

Minimum governance stack:

• Data ownership (who fixes broken feeds)
• Model monitoring (drift, stability, error tracking)
• Human approval steps for high-stakes actions
• Auditability (why the model suggested an action)

AI in the oil and gas sector isn’t “set and forget.” It’s closer to running an asset: monitor, maintain, improve.

Common questions teams ask (and straight answers)

Answer first: AI helps most when it’s used to manage uncertainty, not eliminate it.

“Can AI really predict oil prices better than the market?”

Not consistently in single-point terms. Markets are efficient enough to humble anyone. AI’s edge is in signal fusion and speed—turning many weak indicators into earlier, more structured probabilities.

“Is geopolitics now irrelevant?”

No. It’s just not sufficient. Geopolitics drives the tail risks; inventories, spare capacity, and demand decide whether those tails become reality.

“What’s the fastest win for a Kazakhstan operator?”

I’d start with disruption monitoring + scenario-based planning tied to a monthly decision cycle. It’s easier to operationalize than a pure price-prediction project and shows value quickly.

What this means for 2026 planning

Oil at ~$60 amid chaos is a reminder that the market is adaptive—and sometimes cynical. It prices probabilities, not headlines. For Kazakhstan’s energy and oil & gas companies, the implication is clear: planning based on one story is a risk.

The better approach is building an AI-supported decision system: continuously update disruption probabilities, quantify price ranges, and translate them into operating plans you can execute.

If you’re mapping your 2026 initiatives in the series “Қазақстандағы энергия және мұнай-газ саласын жасанды интеллект қалай түрлендіріп жатыр”, consider this your context-setting chapter: the world is noisy, prices are non-linear, and your advantage comes from faster, calmer interpretation.

Where would an extra two weeks of early warning help you most—exports, maintenance, procurement, or hedging?