Fed Minutes Reveal Rate Hike Debates, Twelve Labs Secures $100M, and Geopolitical Shocks Rebound Oil

Fed Minutes Reveal Rate Hike Debates, Twelve Labs Secures $100M, and Geopolitical Shocks Rebound Oil
Three intersecting market signals on July 8–9, 2026 collectively tightened the global financial conditions picture. FOMC minutes from the June meeting revealed a Fed far more divided than the unanimous hold suggested — several members actively advocated for a rate hike given AI infrastructure's structural electricity demand shock (a new supply-side inflationary force not in the original Taylor Rule calculus), and the market probability of a November 2026 cut fell from 35% to 15% in one session. Twelve Labs' $100M Series B — co-led by NEA and NAVER with Amazon participation, tied to a multiyear AWS Trainium chip commitment — is the latest evidence that enterprise video intelligence (search, cataloging, and analysis of video at scale) is becoming its own AI infrastructure category, distinct from generative video. And US airstrikes on Iranian targets pushed Brent crude +3.2% to $78.80/bbl as Strait of Hormuz disruption fears repriced, while gold fell −1.4% to $4,065/oz as the hawkish minutes raised the opportunity cost of the non-yielding metal.
📈 FOMC June Minutes — The AI Energy Demand Wild Card
What the Minutes Revealed Beyond the Unanimous Hold
The unanimity paradox: The June FOMC meeting voted 12-0 to hold rates at 3.50–3.75% — a unanimous decision that markets read as consensus neutrality. The released minutes revealed the opposite: several participants discussed actively raising rates, and the vote was unanimous only because of an implicit agreement to hold pending one more inflation data point, not because of agreement on the direction.
Key concerns cited in the minutes:
| Fed concern | Specific language (paraphrased) | Market implication |
|---|---|---|
| AI electricity demand | "Structural surge in data centre energy demand may constitute a persistent supply-side cost shock" | New inflationary variable — not in standard models |
| Tariff pass-through | "Business survey contacts report ongoing difficulty absorbing tariff-related input cost increases" | Goods inflation floor remains elevated |
| Services inflation persistence | "Non-housing services inflation remains above 4.5% and is not decelerating" | Core PCE stuck above target |
| Geopolitical freight costs | "Shipping disruption contributing 30–50bp to headline CPI — no near-term resolution visible" | Supply-side inflation floor |
| Labour market | "Labour market remains tighter than models suggest for this stage of the cycle" | Wage inflation persistence |
The AI electricity demand argument — why it matters: US data centre electricity consumption:
- 2023: ~130 TWh/year (2.5% of US electricity)
- 2025: ~250 TWh/year (4.8% of US electricity)
- 2028 projection (Goldman Sachs): ~500 TWh/year (9.5% of US electricity) — equivalent to adding Texas's entire electricity demand
The problem: US grid capacity is not growing at the same rate. New data centres are forcing spot power price spikes (+40–70% above long-run averages) in grid-constrained markets (Virginia, Texas, Arizona, Georgia). When electricity price inflation is structural (from capacity constraints), not cyclical, monetary policy cannot reduce it — but the Fed must raise rates anyway if it feeds into core services CPI.
Market repricing post-minutes:
| Metric | Pre-minutes (July 7) | Post-minutes (July 8) | Change |
|---|---|---|---|
| Nov 2026 rate cut probability | 35% | 15% | −20pp |
| Dec 2026 rate cut probability | 52% | 31% | −21pp |
| US 10Y Treasury yield | 4.04% | 4.12% | +8bps |
| DXY | 103.8 | 104.5 | +0.7% |
| S&P 500 | −0.1% | −0.3% | Minor equity consolidation |
| Gold price | $4,123/oz | $4,065/oz | −1.4% |
The rate hike vs hold calculus — what tips the September decision:
| Data point | Bullish (supports hold) | Bearish (supports hike) |
|---|---|---|
| June CPI (released July 11) | <3.8% → hold reinforced | >4.2% → hike debate intensifies |
| June PCE (released July 31) | <3.5% core → relief | >3.8% → hike back on table |
| June employment (released July 5) | NFP <180K → labour cooling | NFP >230K → tighter → hike risk |
| Q2 GDP advance (released July 30) | <2.0% → rate sensitivity concerns | >2.8% → economy can absorb hike |
🤖 Twelve Labs $100M Series B — Video Intelligence as Infrastructure
The Enterprise Video Intelligence Market
What Twelve Labs builds — the distinction from generative video:
| Company/Product | Category | Core capability | Customer |
|---|---|---|---|
| Runway, Sora, Kling | Generative video | Text → video creation | Creators, entertainment |
| Twelve Labs (Marengo, Pegasus) | Video intelligence | Video → insights, search, cataloging | Enterprise (media, security, sports, legal) |
| Veritone | Video AI + content | Audio/video transcription + AI metadata | Broadcasters |
| Clarifai | General vision AI | Image/video classification | Government, retail |
The video intelligence market is fundamentally different from generative video: instead of creating content, it understands content at scale — enabling search across video repositories ("find all footage where a person in a blue jacket enters through the left door"), automated cataloging of broadcast libraries, and forensic analysis of security footage.
Twelve Labs' model stack:
| Model | Function | Use case |
|---|---|---|
| Marengo | Multimodal video search and retrieval | Enterprise: find specific moments in large video archives |
| Pegasus | Video-to-text generation | Generate chapters, summaries, insights from long videos |
| Pegasus-Pro (in development) | Real-time video analysis | Security, live broadcasting |
The AWS Trainium partnership — strategic implications:
| Deal element | What it means |
|---|---|
| Training workloads on AWS Trainium | Twelve Labs trains its large models on AWS custom chips (Trainium 2) — not NVIDIA A100s/H100s |
| First-access to new models on AWS | New Twelve Labs model releases debut on AWS Bedrock platform before other clouds |
| Amazon participation in round | Amazon is both investor and compute provider — aligned incentives |
| Trainium cost advantage | AWS Trainium typically 30–40% cheaper per training FLOP than comparable NVIDIA GPU instances |
The AWS developer lock-in strategy: AWS has invested in >15 AI foundation model companies with the same structure: strategic investment + Trainium compute commitment + first-access to new models. The pattern creates a "model garden" in AWS Bedrock that differentiates AWS from Azure (OpenAI) and GCP (Google DeepMind) by offering enterprise-grade specialised models across vision, language, and reasoning.
Total funding trajectory:
| Round | Date | Amount | Lead | Total raised |
|---|---|---|---|---|
| Seed | 2021 | $4.7M | — | $4.7M |
| Series A | 2022 | $12.5M | NEA | $17.2M |
| Series A-2 | 2023 | $50M | NEA | $67.2M |
| Series B | July 1, 2026 | $100M | NEA + NAVER | ~$167M |
🛢️ Brent Crude and the Strait of Hormuz Risk Premium
The Iran Strike Impact on Oil Markets
Strait of Hormuz flow significance:
| Commodity | Daily transit volume | % of global trade |
|---|---|---|
| Crude oil + condensate | ~18–20 million bpd | ~21% |
| LNG | ~25 bcf/day | ~25% |
| Petroleum products | ~3–4 million bpd | ~15% |
Any closure or disruption to the Strait would affect a quarter of the world's traded oil — a structural geopolitical risk premium is always embedded in Brent prices.
July 8 oil market response to US airstrikes:
| Oil metric | Pre-strike (July 7) | Post-strike (July 8 close) | Move |
|---|---|---|---|
| Brent crude (ICE) | $76.35/bbl | $78.80/bbl | +3.2% |
| WTI crude (NYMEX) | $71.85/bbl | $74.26/bbl | +3.4% |
| Brent-WTI spread | $4.50 | $4.54 | Stable |
| Geopolitical risk premium (estimated) | $4–6/bbl | $7–9/bbl | +$2–3/bbl added |
Why gold fell despite geopolitical escalation: Gold's typical safe-haven behaviour (rally on geopolitical risk) was overridden by two hawkish forces:
| Gold headwind | Mechanism |
|---|---|
| FOMC minutes hawkishness | Higher-for-longer rates → opportunity cost of holding gold (non-yielding) increases → gold sold |
| DXY strengthening | Gold priced in USD → DXY +0.7% → gold's non-USD price rises → USD price falls to compensate |
| Magnitude | DXY +0.7% typically → gold −0.8 to −1.0% → explains most of the −1.4% move |
Copper's range-bound position:
| Copper driver | Direction | Strength |
|---|---|---|
| Chile/Indonesia mine disruptions (supply constraint) | ↑ bullish | Strong |
| AI data centre + grid electrification demand | ↑ bullish | Strong |
| China industrial activity (weak) | ↓ bearish | Moderate |
| High borrowing costs (compressing capex) | ↓ bearish | Moderate |
| Net effect | → Range-bound | ~$6.00–6.40/lb range |
At $6.06/lb, copper is exactly at the midpoint of this range — two strong bullish forces offsetting two moderate bearish forces.
Oil market near-term scenarios:
| Scenario | Brent target | Probability |
|---|---|---|
| Strait of Hormuz closure (15+ days) | $110–125/bbl | ~8% |
| Escalation but no closure | $82–90/bbl | ~30% |
| De-escalation / ceasefire | $72–76/bbl | ~40% |
| Full peace talks resume | $68–72/bbl | ~22% |
📌 The Bottom Line
- fomc-june-minutes-hawkish-hold-rate-hike-debate: Unanimous hold masks deep divisions; key Fed concerns: AI electricity demand as structural supply-side inflation shock (US data centres projected 500 TWh/year by 2028 = +Texas's entire grid), tariff pass-through, services inflation >4.5%, Hormuz freight +30–50bp CPI; Nov 2026 cut probability fell 35%→15%, Dec 31%→21%; 10Y yield +8bps to 4.12%, DXY to 104.5; decision hinges on June CPI (July 11), PCE (July 31), NFP — >4.2% CPI puts September hike back on table.
- twelve-labs-100m-series-b-aws-trainium-video-ai: Video intelligence ≠ generative video: Twelve Labs searches, catalogs, and extracts insights from existing video; Marengo (multimodal search) + Pegasus (video-to-text); Series B $100M, NEA+NAVER co-lead, Amazon participant; AWS Trainium deal: all training workloads on Trainium (30-40% cheaper than NVIDIA), new models debut on AWS Bedrock first; AWS pattern: >15 AI investments with same structure → Bedrock model garden differentiation vs Azure (OpenAI) + GCP (Google); total raised $167M.
- brent-crude-us-iran-strikes-gold-copper-commodities: Hormuz: 21% of global crude + 25% of LNG daily; July 8 US airstrikes: Brent +3.2% to $78.80, WTI +3.4% to $74.26, geopolitical premium +$2-3/bbl to $7-9/bbl; gold −1.4% to $4,065 (hawkish minutes opportunity cost + DXY +0.7% override safe-haven bid); copper range-bound at $6.06 (Chile/Indonesia supply + AI demand bullish vs China weakness + high rates bearish); oil scenarios: Hormuz closure $110-125 (8%), escalation no closure $82-90 (30%), de-escalation $72-76 (40%).
📬 Stay Updated
Get global market analysis delivered to your inbox every week. Subscribe to our free newsletter →
Disclosure: This post contains affiliate links. If you purchase through our links, we earn a small commission at no extra cost to you. We only recommend products we believe in.
Enjoyed this post?
Get our weekly digest delivered free.
Share this post:
Knowelth is reader-supported. We may earn a commission from links in this article at no extra cost to you. Read our disclosure.


