Confidential · Investor Brief
May 2026
A Brief Concerning

XportL.

Every AI company is paying for computers that spend a third of their time doing nothing.

We get those computers back to work.

Opportunity Zone Fund Available
The Ask
$2,000,000
SEED · 18-MONTH RUNWAY
Stage
Working prototype
CLOSED LOOP VALIDATED · XPORTL.IO
i.
§ 01 · The Problem
XportL · Investor Brief
Picture this

Imagine the world's most expensive kitchen.

Four thousand chefs are cooking one enormous meal together. Each costs a hundred dollars an hour. The kitchen costs four hundred thousand dollars an hour to operate.

But the recipe requires constant coordination—blending sauces, passing ingredients, syncing timing. Every few seconds, all four thousand chefs stop chopping and crowd the hallway to hand things to each other.

The hallway traffic takes a third of the workday.

A typical workday for a $2.50/hr GPU
Actually cooking55%
Standing in the hallway, waiting35%
Cleaning, setup, breaks10%

Multiply across every AI company on Earth—$400 billion of computers over three years—

$140B
paid to computers standing in the hallway
≈ 30% of projected ~$467B 2025 AI infrastructure spend
Measured: 5% average GPU utilization across 23,000 clusters — Cast AI, 2026
01
§ 02 · Why It Happens
XportL · Investor Brief
Why no one has fixed this

The hallway has no traffic cop.

The software that tells the chefs when and how to coordinate was written eight years ago, when kitchens were smaller and simpler.

It makes one plan at the start of the workday and never changes it. Meanwhile, the kitchen has grown ten times bigger.

It is a fast car
with the steering locked.

The mismatch, in three lines
  1. 01 The hallway gets congested thousands of times per second.
  2. 02 The coordinator updates its plan once every eight hours.
  3. 03 Every second in between is wasted money you've already paid for.

XportL is the traffic cop. We watch the hallway, predict the jams, and reroute the chefs—every millisecond, automatically.

02
§ 03 · Two Faces of the Same Problem
XportL · Investor Brief
The problem has two faces

A clogged hallway is not the only way to waste a kitchen.

FACE ONE
The crowded hallway.

Thousands of chefs trying to cross a hallway that wasn't built for them. Bottlenecks. Backed-up queues. Routes the coordinator never updates.

We watch traffic, predict jams, reshape routing. The hallway flows again.

In our prototype
+98%
bandwidth recovered under congestion
FACE TWO
The slow chef.

One chef is moving at three-quarter speed. Not sick — not gone — just slow. Every other chef in the kitchen waits at the every coordination point for the slow one.

The whole kitchen slows down to match. NVIDIA's own tools say "keep running" — they can't tell.

In our prototype
+59%
throughput recovered from degraded GPU
The platform thesis

One telemetry pipeline. Two distinct recoveries. The same intelligence layer that solves congestion solves fail-slow — and tomorrow it solves more.

"We saw fail-slow detection ship in one engineering day on top of the platform that handles congestion. That is what platform leverage looks like."
03
§ 04 · How XportL Works
XportL · Investor Brief
What we built

Three steps. All running at once. All invisible to the customer.

STEP ONE
We watch.

We install a small piece of software that listens to every interaction between computers—millions per second.

Nothing the customer is doing changes. We just listen.

STEP TWO
We predict.

A small AI model forecasts which routes will be jammed moments from now—a window that's slow for nanoseconds but extraordinary for collective routing.

A weather forecast for traffic jams.

STEP THREE
We reroute.

We change the route through a vendor-supported extension point. No application code modification. No kernel drivers.

The customer never knows we did anything. Their work finishes sooner.

What the customer sees

A training run that used to take 30 days—now finishes in 25. On the same hardware.

$4–8M
saved per major training run
04
§ 05 · Under the Hood
XportL · Investor Brief
For the technical read

Specific. Bounded. Verifiable.

Not a kernel module. Not a binary patch. Not a vendor SDK fork. XportL operates inside the standard collective communication runtime through a vendor-supported extension point — the same surface any cluster administrator can use today.

What we observe What we predict What we change
Per-rank collective duration
microsecond resolution
Link utilization
~100ms horizon · <5ms inference
Collective algorithm
Ring ↔ Tree
Switch queue depth, ECN marks
per port, per spine
Per-link congestion score
graph neural network
Transport protocol
low-latency ↔ pipelined
Rank-relative byte-rate variance
nanoseconds per byte
Degraded ranks
~3 second detection
Ring order, chunk count
degraded ranks rotated out
Active-collective state
in-flight handle polling
Stuck collectives
~3 seconds vs the 30-minute default
Scheduler cordon
Slurm · Kubernetes · audit-only
No customer code change

Training frameworks (PyTorch DDP, Megatron, DeepSpeed) need no modification. The integration surface is at the runtime layer below.

API-stable across versions

The extension surface is documented and supported by the vendor. New runtime versions ship; XportL keeps working.

Cross-vendor by design

NVIDIA and AMD reference stacks expose equivalent extension surfaces. XportL's roadmap covers both.

Telemetry overhead is bounded: <0.5% of fabric bandwidth, single-digit milliseconds of host CPU per second per rank. Never pauses the training process. Never touches application memory.

05
§ 06 · The Savings, Calculated
XportL · Investor Brief
The math, by customer size

What XportL saves a customer—in dollars.

Modeled annual savings · 15% recovery of wasted time
Cluster ProfileGPUsAnnual ComputeXportL Saves
Single Pod
small lab · startup
1,024$22.4M$1.18M
Mid-Tier Cluster
neocloud · enterprise lab
4,096$89.7M$4.71M
Frontier Cluster
major AI lab · sovereign
16,384$358.8M$18.8M
Hyperscale Data Center
Meta · xAI scale
100,000$2.19B$115M

Math: GPU-hours/yr × $2.50/hour × 35% communication waste × 15% recovery.

How XportL gets paid

We charge 25% of the dollars saved.

The customer keeps three quarters of every dollar we recover. Aligned incentives, no winner-loser dynamic.

Revenue per frontier customer
$4.7M/yr
from a single 16K-GPU customer

Three frontier customers alone = roughly $14M ARR.
Ten = $47M ARR.

06
§ 07 · The Market
XportL · Investor Brief
The market, in three sizes

Even a tiny slice of this is a very large company.

The whole pie
$45B
per year, by 2027

Total dollars the AI industry will waste on hallway-waiting time every year.

The slice we can reach
$12B
per year

Neoclouds, sovereign labs, frontier labs, enterprise on-prem. Sovereign AI alone is +$100B of pipeline through 2030.

What we plan to capture
$420M
per year, by year five

A 3.5% capture. Just 90 frontier customers at $4.7M ARR each.

Who actually buys this

The universe of buyers is fewer than 200 companies globally. Every name is already in the news.

Neoclouds (Nebius, Crusoe, Lambda, Together — the wedge customer). Sovereign AI labs (HUMAIN, UAE Stargate, Aleria — newly unlocked by November 2025 GB300 export approval). Frontier labs (Cohere, Mistral, Reka — the prestige references). Enterprise on-prem (JPMorgan-class — the cybersecurity expansion).

~200
customers · the entire TAM
+$100B sovereign pipeline through 2030
07
§ 08 · The Platform
XportL · Investor Brief
What this becomes

XportL is not an optimization tool.
It is per-rank intelligence for AI clusters.

One telemetry pipeline. One graph model of cluster state. One control plane that can act on what it learns. Three customer problems that all reduce to the same shape: watch every rank, find the one that doesn't belong, do something about it.

YEAR ONE · SHIPPING
Performance.

Recover wasted GPU time. Demonstrated capabilities: congestion routing, fail-slow detection, stuck-collective detection, scheduler cordon (Slurm + K8s), checkpoint webhook, chargeback rollup, hardware drift detection, hierarchical ring construction.

See SKU · Fix SKU
$25–250 / GPU / year
YEAR TWO · ADJACENT
Integrity.

Detect anomalous collective traffic patterns: gradient manipulation, weight exfiltration, insider sabotage during training runs. No production tooling exists for this today.

Guard SKU
cybersecurity multiples · 2–3×
YEAR THREE · FRONTIER
Hardware.

Programmable DPU/SmartNIC firmware that pushes XportL's intelligence directly into the fabric silicon. Multi-year, multi-million-dollar contracts.

Build SKU
Enterprise · seven figures

What this seed funds: Year One. Performance optimization, first paying customer, published benchmark. The Integrity and Hardware lines are the long-term arc the seed earns the right to pursue — not what we are pitching, but what makes this a venture-scale outcome rather than a feature.

08
§ 09 · The Field
XportL · Investor Brief
Who else is here

The category is active.
The cross-vendor commercial operator is still us.

In the months we've been building, four credible technical efforts have published into this space — which validates the wedge and compresses it. None of them is a commercial, cross-vendor, deployable-anywhere product. That position remains open, and the audit-of-record story below it is uncontested.

ACADEMIC · OSS
NCCLbpf

UC Santa Cruz, March 2026. Verified eBPF policy execution. Closest technical cousin. Self-deploy.

Commercial?
No — OSS only
HYPERSCALER · OSS
Meta NCCLX

100k+ GPU collective framework. Built for Llama4. Library replacement, not a control plane.

Cross-vendor?
NVIDIA-only
CLOUD-LOCKED
Alibaba C4 / ACCL

Production at >80k GPUs. Real-time anomaly detection + traffic engineering. Inside Alibaba Cloud.

Available to
Alibaba customers
CLOUD-LOCKED
AWS Checkpointless

SageMaker HyperPod EKS, December 2025. Solves restart-waste through model redundancy. AWS-only, NeMo-only.

Available to
AWS customers
NEOCLOUD-INTERNAL
CoreWeave Mission Control

Tooling CoreWeave runs for its own fleet. Not a product they sell. Nebius and Crusoe each operate similar internal stacks.

Available to
One cloud each
FROM THE OPERATOR SIDE · DECEMBER 2025

xAI's in-house C training stack confirms the thesis from the frontier. Reported scope: 220k GB300 superchips, 800G NICs, purpose-built communication layer, exact-mapped to a known fleet. At their scale the gap between generic frameworks and cluster-specific optimization is worth massive engineering investment — they amortize that work across one $6B fleet. XportL is the productized form of that work, for every operator who can't justify the C rewrite.

XPORTL · THE UNCONTESTED POSITION
Commercial. Cross-vendor. Deployable anywhere collectives run.

Every alternative above is either an open-source library you self-operate or locked to one cloud. The neocloud, sovereign lab, and enterprise on-prem customer that wants a commercial product across heterogeneous fleets has one option.

Below the runtime tuning — where the open-source efforts compete — sits the audit log, the 90-day reliability ledger, the MTTF-aware checkpoint advisor, the per-node failure recurrence scoring. That layer is uncontested.

Detection latency
~3 seconds
Reliability ledger retention
90 days
The real moat — 18 months out

Open-source frameworks live inside one cluster. They cannot accumulate longitudinal multi-customer reliability data. XportL, deployed across customers, becomes the only place where GPU-SKU-by-firmware failure cohorts exist as queryable data.

That dataset is the actuarial basis for AI infrastructure SLAs. Customers buy the platform. Insurers and procurement buy the data.

Full landscape, IP posture, & data-moat strategy in the dataroom · § 13
09
§ 10 · Why Now
XportL · Investor Brief
A rare moment

Three doors opening at once.

For two years, every dollar of AI investment went to building the kitchens. The next two years reward whoever can make those kitchens cook faster.

i
The market is wide open.

Every new AI data center is being built on a kind of network where Nvidia has no special advantage. Greenfield. Anyone's to win.

ii
The status quo is broken.

Coordinator software designed in 2017. AI training has grown a thousand-fold since. The gap is now obvious.

iii
Customers are desperate.

Every GPU-hour is sold months in advance. The phone rings on its own.

10
§ 11 · The Team
XportL · Investor Brief
Who's building this

Two operators. One has shipped companies. One has shipped infrastructure.

SJ
Steve Johnson
Co-Founder · Chief Executive

Serial entrepreneur with a public-market exit. Took prior venture from inception through IPO.

Deep operator experience in brand strategy, GTM, capital formation, execution discipline.

Owns at XportL
Capital · Sales · Partnerships · Investor relations · Brand
WT
Will Trent
Founder · Chief Technology Officer

Linux systems engineer and entrepreneur. Built the working prototype: closed control loop, GNN predictor, real NCCL plugin validated on NVIDIA hardware.

Deep expertise in kernel-level instrumentation, distributed systems, production infrastructure.

Owns at XportL
Architecture · Engineering · Customer deployment · Tech hires
11
§ 12 · What the $2M Buys
XportL · Investor Brief
Lean. Focused. Eighteen months.

What $2M buys.

Four senior engineers. Real-hardware deployment. A paying first customer. Reserve to raise the next round from strength.

Opportunity Zone Fund
XportL is structured to receive capital through a qualified Opportunity Zone fund.

Investors deploying capital-gains proceeds may be eligible for deferral, basis adjustment, and exclusion of gain after the statutory holding period. Details on request.

Engineering team
Four senior engineers
$1.4M
70% of the round
Real hardware testing
Validation on real clusters
$240K
12% of the round
Operations, IP & reserve
Legal, accounting, FTO opinion, cash buffer
$360K
18% of the round
Total raise
$2,000,000
12
§ 13 · The Invitation
XportL · End of Brief
The Invitation

The world is buying computers.
We make those computers work harder.

Two million dollars. Eighteen months. A small team that has already built the prototype, validated the plugin on real NVIDIA hardware, and is ready to put it in front of real customers.

We're taking a small group of seed investors who understand that the pickaxes get rich, not just the gold miners.

Investment
SAFE · OZ Fund
Minimum
$100,000
Timeline
60 days
Web
xportl.io
Steve Johnson · CEO
sjohnson@stoanhedge.com
Will Trent · CTO
techonthego@gmail.com
13