QumulusAI, Inc. · NASDAQ:QMLS
QumulusAI — Investor Presentation (September 2026)
32 slides from QumulusAI's investor presentation, Sep 22, 2026
Slide 1: Breaking AI's Bottlenecks
// INVESTOR PRESENTATION Breaking AI’s AI compute online in months, not years. We use existing capacity to meet demand now, while simultaneously building the pipeline to power what's next. QumulusAI (NASDAQ: QMLS) :: September 22, 2026
Slide 2: Forward-looking Statements and Disclaimers
// DISCLAIMER Forward-looking Statements and Disclaimers Presentation time in the Company’s subsequent filings with the SEC. These factors include, among others, our EBITDA is defined as net income (loss) before interest expense, net; income tax expense dependence on a limited number of large custom
Slide 3: Why Now, Why Us
// SECTION 04 Why Now, Why Us
Slide 4: It all runs on AI infrastructure—that's the opportunity
NASDAQ QMLS :: 4 // THE OPPORTUNITY Demand for AI infrastructure providers is compounding. It all runs on AI infrastructure that’s the opportunity. Google’s monthly tokens leaped from 480T to 3.2 Big-4 hyperscaler 2026 capex guidance — raised quadrillion year over year for the third straight May ’2
Slide 5: Better models. Better results. Real use cases.
NASDAQ QMLS :: 5 // THE OPPORTUNITY Drivers of Demand Sources: Stanford University Apr 2026 — AI Index Report; Harvard Business Review June 2026 — How People are Really Using AI in 2026. AI USE CASES Better models. Better results. Real use cases. Customer Service/ Software Personal and Call Cent
Slide 6: Demand is real and it's repricing compute
NASDAQ QMLS :: 6 // THE DEMAND SIGNAL Sources: Exponential View, State of the AI Economy Jun 2026 — GenAI revenue; Epoch AI, An Demand is real and it’s repricing compute update on AI's most important number Aug 2026 — OpenAI/Anthropic run rate; CBRE, North America Data Center Trends H1 2026; Cor
Slide 7: Data centers can't come online fast enough
NASDAQ QMLS :: 7 // THE BOTTLENECK Data centers can’t come online fast enough Sources: Lawrence Berkeley National Lab, Queued Up 2026 Edition Jun 2026; Sightline Climate via Bloomberg May 2026 — 2026 US datacenter pipeline; Gartner 2025. GRID INTERCONNECTION 2026 PIPELINE STUCK IN QUEUE Median
Slide 8: The market needs faster, more flexible access to compute
NASDAQ QMLS :: 8 // OUR FACTS MODEL The market needs faster, more flexible access to compute We house the latest models of NVIDIA GPUs in colocation facilities and 50MW capacity data centers to bring GPU capacity online in months, on a quarterly deployment cadence that can accelerate time to value
Slide 9: The Proof
a QumulusAl The Proof
Slide 10: What's inside one megawatt of IT load
NASDAQ QMLS :: 10 // READING THE MEGAWATT Sources: NVIDIA DGX B300 specifications 14.5 kW, 8 GPUs per system); SemiAnalysis datacenter and What’s inside one megawatt of IT load AI cloud research; QumulusAI management estimates for equipment cost and revenue per MW ranges, representative of market a
Slide 11: What's live, what's signed, and what we hold
NASDAQ QMLS :: 11 // POWER FOOTPRINT What’s live, what’s signed, and what we hold POWER BY STATUS Fully sold, coming online ACTIVE Five colocation sites · Marietta, Kansas 8 MW 100% sold City, Denver, Philadelphia 2 In final deployment CONTRACTED Contracted for Q4 3.75 ROFO 7 MW Energiz ing to be
Slide 12: What today's capacity generates, and its potential
NASDAQ QMLS :: 12 // CAPACITY INTO REVENUE What today's capacity generates, and its potential WHAT EACH BLOCK OF OUR CAPACITY IS WORTH PER YEAR 3.75 MW Atlanta · Q4 2026 contracted colocation $75M / yr NOT YET SOLD Potential revenue, at $20M per MW, once sold and fully active — within the $1921M
Slide 13: Q2 RPO of $173.1M; ~$276M pro forma backlog
NASDAQ QMLS :: 13 // REVENUE BACKLOG Q2 RPO of $173.1M; ~$276M pro forma backlog with post-quarter signings EXPECTED REVENUE RECOGNITION BY YEAR — REPORTED Q2 RPO VS. PRO FORMA BACKLOG (ILLUSTRATIVE) $276M Since Q2 2026 quarter-end $173.1M Left: RPO as reported in the Q2 10Q — $34. 4M expected to
Slide 14: Selected news, June–August 2026
NASDAQ QMLS :: 14 // TRACTION & PROOF POINTS Selected news, June–August 2026 AUG 20, 2026 Metro Atlanta site anchors the next 3.75 MW Seven-year colocation, RFS Q4 2026; right of first offer on 7 MW more — a path to 10.75 MW. AUG 11, 2026 DRW becomes a customer for a dedicated B300 cluster Annually
Slide 15: Distributed for speed now, positioned for inference next
NASDAQ QMLS :: 15 // DISTRIBUTED NETWORK Distributed for speed now, positioned for inference next 6 Contracted colocation Powered land AI IXP pod · CNIXP partnership Metro-adjacent colocation sites Atlanta, GA Marietta, GA Kansas City, MO Denver, CO Philadelphia, PA – x2 2 Powered-land sites Oklahom
Slide 16: A network of sites: how we build it, and why it compounds
NASDAQ QMLS :: 16 // SITE BUILD MODEL A network of sites: how we build it, and why it compounds HOW WE BUILD WHAT WE’VE BUILT Find Pockets of 250 MW of existing power. COLOCATION POWERED LAND PARTNER EDGE Metro-adjacent colocation Power we hold ourselves GPU pods at the edge Existing capacity in an
Slide 17: Who We Work With
NASDAQ QMLS :: 17 // TRACTION & PROOF POINTS Who We Work With BLUE CHIP PARTNERS BREAKOUT CLIENTS
Slide 18: Why We Win
a QumulusAl Why We Win
Slide 19: Customers don't open with price. They open with "When can I be running?"
NASDAQ QMLS :: 19 // WHY WE WIN Customers don’t open with price. They open with “When can I be running?” TIME TO A TOKEN-GENERATING CLUSTER HYPERSCALE BUILD Design, approvals, financing Construction 34 years 12 15 months 2 436 months decides most deals today to revenue generation We believe supp
Slide 20: Smaller, distributed sites reach power faster than large campuses
NASDAQ QMLS :: 20 // COMPETITIVE DIFFERENTIATION Smaller, distributed sites reach power faster than large campuses. Pockets of 250 MW of existing or near-ready power, mostly in colocation, brought online in months. Why smaller and distributed sites win on time 1. More sites to choose from, and in m
Slide 21: Validated by NVIDIA. Built for production.
NASDAQ QMLS :: 21 // NVIDIA CLOUD PARTNER Validated by NVIDIA. Built for production. QumulusAI is an official NVIDIA Cloud Partner, recognized as delivering infrastructure purpose-built for modern AI workloads at production scale. Our platform is built and validated against NVIDIA reference architec
Slide 22: Customer testimonials
NASDAQ QMLS :: 22 // FROM THE SOURCE Customer testimonials “ QumulusAI’s workload-optimized infrastructure gives us the performance, efficiency, and scalability we need as we continue expanding reliable GPU compute for customers building AI at scale.” — Jasper Zhang, CEO of Hyperbolic “ This capacit
Slide 23: Team & Governance
a QumulusAl Team & Governance
Slide 24: Michael Maniscalco
NASDAQ QMLS :: 24 // TEAM & GOVERNANCE At the helm: an operator who's scaled GPU cloud Michael Maniscalco Chief Executive Officer :: QumulusAI Michael has built and launched a GPU-as-a-Service business at scale before — the exact business QumulusAI runs today. 01 — FOUNDATION 02 — VENTURE 03 — GPUAA
Slide 25: A team that has built this before
NASDAQ QMLS :: 25 // TEAM & GOVERNANCE A team that has built this before Mike Maniscalco Scott Krosnowski Ankur Chatterjee Ryan DiRocco Stephen Hunton Patrick Gahan Jim Marino CHIEF EXECUTIVE CHIEF FINANCIAL CHIEF INTEGRATION CHIEF TECHNOLOGY CHIEF MARKETING SENIOR VICE PRESIDENT SENIOR VICE PRESIDE
Slide 26: Financial Overview
a QumulusAl Financial Overview
Slide 27: Q2 2026 financial summary
NASDAQ QMLS :: 27 // FINANCIAL DATA ¹ Adjusted EBITDA is a non-GAAP measure; see the reconciliation of net income (loss) to Adjusted EBITDA in this presentation and in the Q2 2026 earnings release at investors.qumulusai.com. Q2 2026 financial summary $ in actual dollars · Three months ended June 30,
Slide 28: Statement of Operations
NASDAQ QMLS :: 28 // FINANCIAL DATA ¹ Q2 2026 includes a $19.2M non-cash loss on issuance of the ATW convertible note, partly offset by $6.2M of fair-value gains; H1 2026 includes $73.9M of such losses. Q2 2025 includes a $14.5M one-time TCM remeasurement gain. Statement of Operations GAAP · $ in ac
Slide 29: Balance sheet & liquidity snapshot
NASDAQ QMLS :: 29 // FINANCIAL DATA Balance sheet & liquidity snapshot Total long-term liabilities at June 30, 2026 include convertible notes at fair value $55.5M) and the additional convertible notes option liability $38.7M, both non-cash fair-value items; finance lease liabilities of $45.8M; an
Slide 30: H1 2026 net loss is almost entirely non-cash
NASDAQ QMLS :: 30 // OPERATING LOSS → NET LOSS BRIDGE H1 2026 net loss is almost entirely non-cash Six months ended June 30, 2026, $ in millions. ¹ Non-cash. Fair-value gains are on the convertible notes and additional-notes option liabilities. ² Interest expense, net; warrant liability fair-value c
Slide 31: Adjusted EBITDA reconciliation
NASDAQ QMLS :: 31 // FINANCIAL DATA Adjusted EBITDA is a non-GAAP measure with limitations as an analytical tool and is not a substitute for net income (loss); see the Q2 2026 earnings release for the full definition, reconciliation and limitations, including finance-lease payments it excludes. Adju
Slide 32: This isn't hyperscale. It's hyperspeed.
// WE ARE THE INFRASTRUCTURE THAT RUNS AI. This isn't hyperscale. It's hyperspeed. > QumulusAI.com