Supply chain · Deep dive

Auger

Dave Clark's post-Amazon act: an AI operating layer that sits on top of every supply chain system a company already owns and makes the boring decisions for them.

emerging

The question that decides it: Does an orchestration layer survive on top of data it does not own — once SAP and Blue Yonder ship agents onto the systems of record they already control?

Snapshot

Auger sells an autonomous operating layer for supply chains: software that sits above a company’s ERP, warehouse, transportation and demand-planning systems, unifies their data, and lets AI agents and optimization models make routine operational decisions and push them back into the systems of record. It was founded in 2024 by Dave Clark — the man who built Amazon’s fulfillment network over 23 years — and his wife Leigh Anne Clark, and has raised roughly $150M, most recently a $50M Series B led by Eclipse on 9 July 2026. At that date: ~130 employees in Bellevue, Washington, three named customers (Meta’s Reality Labs, Fanatics, Kimberly-Clark), no disclosed revenue. Clark’s stated goal is half of US GDP flowing through the platform by 2030 with revenue above $1B — which tells you how the company is being built, and how to read it.

Founding story

The Clark biography is the asset, and Auger’s entire go-to-market rests on it. He joined Amazon in 1999 out of business school via the Pathways operations program, ran a Kentucky fulfillment center by 2001, took over Northeast warehouses in 2003, became VP of North American operations by 2010, SVP of Worldwide Operations in 2013, and CEO of Worldwide Consumer in 2021. Subordinates nicknamed him “the Sniper,” reportedly after he described lurking in warehouse shadows to catch workers slacking — a detail that has followed him for twenty years and cuts both ways. He owned the buildout of Amazon’s fulfillment and last-mile network — the most consequential physical infrastructure in modern retail — and left in July 2022.

Then came Flexport. Named incoming CEO in June 2022, he ran the company alongside founder Ryan Petersen for six months, took sole control in March 2023, and was out by September 2023 — under a year. Petersen publicly framed the ouster around overhiring and lost spending discipline (FreightWaves, September 2023); CNBC’s reconstruction (2 October 2023) described standoffs, politics and spin. Clark hit back at a logistics conference, saying his real regret was picking the wrong founder and calling Flexport’s layoffs brutal and disrespectful. Two days after his exit, Flexport rescinded dozens of job offers. This is not a footnote — it is the largest question mark over Auger. Clark has run an enormous operation inside a system he helped design, and he has once tried to run someone else’s company and failed inside twelve months. Auger is the first time he has built from zero.

He moved back to the Seattle area from Texas, launched Auger in summer 2024, and staffed it from Amazon’s senior ranks — corp-dev chief Alex Ceballos as president and CFO, Amazon’s longtime logistics chief scientist Russell Allgor as chief scientist — with hires from Johnson & Johnson, Microsoft and Salesforce filling out product, data and go-to-market.

How it works

Mechanically, Auger is an ingestion layer, an ontology, an agent/optimizer stack and a write-back path. Ingestion is deliberately unfussy. Auger’s public position is that the golden-record model is a lie: rather than making customers clean their data first, it takes raw operational streams — ERP tables, spreadsheets, legacy APIs — as they are and normalizes them inside its own semantic model. The product is built on Azure, stores data in Microsoft OneLake, and extends Microsoft Fabric with supply-chain-specific reasoning.

The ontology — Auger’s term of art — encodes how a specific customer’s supply chain actually behaves: contractual commitments, capacity ceilings, cost curves, service levels, financial trade-offs. Clark’s argument is that a frontier model cannot run a supply chain because it does not know these constraints, and that many pure-technology companies have died trying.

A decision, in practice, looks like the demo Clark ran for GeekWire in July 2026: a supplier misses a delivery commitment and there is not enough product to go around. Auger detects the shortfall, ranks customers against contractual and financial rules, reallocates inventory, and writes the revised plan back into the existing systems. Adjacent decisions are the same shape — rerouting shipments, resetting production targets, rebalancing inventory across nodes, adjusting POs. Most tools raise an alert and wait for a human; Auger executes and escalates only what falls outside its guardrails. Every autonomous action is logged and auditable, hard constraints are enforced at execution time, and Clark’s framing is that Auger is not a tool but a new employee.

The exception loop is the commercial hinge. At Fanatics, Clark said in July 2026 that about 85% of decisions in the process Auger manages now run without a human, targeting the mid-90s. That is a per-process figure, not a whole-supply-chain one — one workflow at one customer is heavily automated, which is not the same as Fanatics’ supply chain being autonomous.

Product and business overview

The platform is sold as one thing — the autonomous supply chain OS — in three components. The ontology and data layer handles ingestion and normalization and is where deployment work sits; Auger claims value in weeks, against the multi-year data-cleansing programs the suites are known for. The agentic execution layer fuses S&OP and S&OE into a continuous decision loop, running agents alongside conventional optimizers. The governance layer is the glass box: configurable human-confirmation thresholds, full traceability, hard limits at execution.

Distribution has one notable channel: since March 2026 Auger has been a premier supply chain partner on Microsoft Fabric, with Microsoft reps able to earn commission on Auger deals — real reach into Fortune 500 accounts, though Clark conceded in July 2026 it is still early. The target buyer, per his October 2024 framing, is the Fortune 500 outside the top 50: big enough to have the Franken-stack problem, too small to have built their own Amazon.

Business model and pricing

Auger is enterprise SaaS: annual contracts with large shippers and manufacturers, priced against the operating layer rather than per seat. It publishes no rate card and has disclosed neither revenue nor average contract value as of July 2026. Whether it charges a platform fee, a per-decision rate, or a share of the working capital it claims to unlock is simply not public — normal at this stage, and a real gap in the diligence.

Two inferences are safe. The “new employee” framing and the half-of-GDP ambition both point toward volume- or value-based pricing rather than seats, because seats are what an autonomy product destroys. And with ~$150M raised against ~130 people and a handful of production customers, Auger is spending far ahead of revenue on a winner-take-most assumption. Clark said he raised the Series B early, before he needed it, to avoid fundraising during a busy onboarding autumn — a reasonable explanation that also means the company chose dilution over testing whether revenue could carry it.

Traction over time

Three customers in three unrelated supply chains — consumer electronics, licensed sports merchandise spiking on live outcomes, thin-margin CPG — is not a normal eighteen-month result and is hard to fake. But the pipeline is the tell: eight to ten companies in pilots or negotiation is a design-partner-stage business, with no ACV attached.

Market analysis

Gartner (April 2026) forecast that spend on SCM software with agentic AI capability grows from under $2B in 2025 to $53B by 2030 — the aggressive framing of the opportunity Auger is selling into. Gartner’s earlier forecast has the broader SCM software market going from roughly $29B in 2023 to about $62B in 2028, a ~16% CAGR. Either supports a company of Auger’s ambition; neither supports half of US GDP by 2030, which is a rallying cry, not a forecast.

The structural forces are real. Two decades of ERP, WMS and TMS purchases produced stacks that record everything and decide nothing, with humans stitching them together in spreadsheets and chat threads. Tariff volatility, shorter product cycles and post-COVID network redesign have raised the cost of slow decisions. And Gartner expects annual renewal price increases from SCM leaders to roughly double — precisely the budget frustration that lets a challenger in the door.

Competitive intel

The strategic problem is that the diagnosis is not proprietary. SAP and Oracle are turning systems of record into systems of action from a position where they already own the data, the contract and the renewal — SAP does not have to win, only bundle. Blue Yonder, bought by Panasonic in 2021 at an ~$8.5B enterprise value, has marketed an autonomous supply chain for five years and now ships agents across forecasting and fulfilment. o9 Solutions, last marked at $3.7B in July 2023, runs composite agents on an enterprise knowledge graph — architecturally the nearest thing to Auger’s ontology. Kinaxis ($2.9B market cap, ~$581M trailing revenue, mid-2026) propagates changes across a concurrent network model instantly and has embedded Maestro Agents inside it. Palantir sells the identical shape of product with vastly more credibility on mission-critical decisioning.

Where Auger wins: no system-of-record migration, no golden-record project, weeks not years, and a team that has run a supply chain at scale rather than sold software to one. Where it loses: sitting on someone else’s data is a rentable position. If incumbents reach good-enough agents inside their own suites, the integration advantage evaporates and the moat must become cross-customer decision history — a network effect on operational judgment that needs far more than three deployments to compound.

History and evolution

Amazon 1999-2022 (exits as CEO of Worldwide Consumer). Flexport CEO announced June 2022, sole control March 2023, out by September 2023 after the Petersen falling-out. Auger founded summer 2024; $100M Series A announced 8 October 2024; an eleven-strong executive team named December 2024, drawing immediate skepticism because there was still no product to show. Quiet hiring through 2025. Meta Reality Labs 4 February 2026, Microsoft Fabric March 2026, the Eclipse-led Series B 9 July 2026. No pivots — but a long stealth period that cost real credibility with the practitioner community Auger needs.

What people say

The case for. The strongest endorsements are commercial, not editorial: Meta’s Reality Labs, Fanatics and Kimberly-Clark put an eighteen-month-old company inside operational workflows, and Eclipse repriced the business at roughly 2x within twenty months (GeekWire, 9 July 2026). Oak HC/FT’s public thesis is that the stack has systems of record but no system of action, and that the team that built Amazon’s is the one to build it. Microsoft naming Auger a premier Fabric partner in March 2026 and letting its own reps carry the product is a meaningful third-party signal. Glassdoor reviews (small sample, ~10 as of mid-2026) skew toward a clear, ambitious leadership vision, real data-science investment and competitive pay. And the 85%-autonomy figure at Fanatics, whatever its scope, is a more specific claim than most agentic vendors will put a number on.

The complaints. The supply chain practitioner community was openly hostile at launch, and the criticism has not been retracted. The Global Supply Chain Council’s trade newsletter The Chain ran a piece in May 2025 asking whether Auger was a $100M PowerPoint, cataloguing LinkedIn commentary that the single-pane-of-glass pitch is twenty years old, that AI trained on garbage data in disconnected systems will produce garbage, and — most pointedly — that a former Amazon big-shot’s CV should not by itself command a $100M round with no product. Adrian Gonzalez of Talking Logistics (December 2024) made the more durable version of the objection: supply chain dysfunction is largely a trust and incentive problem between functions and trading partners, and no amount of capital or software solves that. The Flexport shadow is explicit in the criticism — the tenure is described as short, messy, and a warning that looking brilliant inside Amazon is not the same as building outside it. Glassdoor’s negative reviews, few as they are, cluster on familiar early-stage themes: false promises from management, expectation of hours beyond finish time, clique-ish leadership, and at least one reviewer alleging the five-star reviews are internal. On the technical side, one Bellevue engineer flagged friction from the strategic bet on Azure when the engineering team’s roots were in AWS — a detail worth watching, since the Microsoft Fabric dependency is now both the distribution channel and the architectural commitment. And nobody outside the company has validated the half-of-GDP claim as anything but marketing.

Outlook: the open question

The question resolves in the next four quarters, and logo velocity is the tell. Auger has three things that matter: a founder with more supply chain scar tissue than almost anyone alive, an architectural bet (orchestrate, don’t replace) that is probably correct, and production customers that are hard to fake. It has three unresolved problems. The moat is unproven — if the value is AI reasoning, that is commoditizing; if it is integrations, incumbents own the data; if it is cross-customer decision history, three logos compound nothing. The autonomy metric is narrower than it sounds: 85% of decisions in one process at one customer is a pilot statistic, not a category proof point. And founder risk cuts both ways — the Amazon record is unimpeachable, the Flexport record a live warning about how Clark performs when the operating system is not one he designed.

The bull case: the Microsoft channel plus operator credibility converts those eight-to-ten pilots into seven-figure ACVs before SAP, Blue Yonder and Palantir ship good-enough agents, and Auger becomes the default execution layer for the upper-middle Fortune 500. The bear case: this is an extremely well-funded feature, and the suites bundle it before Auger reaches escape velocity. Watch logo velocity, and watch harder for the first disclosed ACV. Until Auger puts a revenue number on the table, the $150M is a bet on a résumé.

Sources and further reading

• GeekWire — Supply chain startup Auger raises $50M and lands big customers (Todd Bishop, 9 July 2026)
• Fortune — Ex-Amazon exec Dave Clark raised $100 million for AI supply-chain startup Auger (8 October 2024)
• Bloomberg — Ex-Amazon retail chief signs up Meta as customer for startup Auger (4 February 2026)
• Fortune — Dave Clark hires Amazon chief scientist and former Amazon execs (9 December 2024)
• CNBC — The inside story of Dave Clark’s tumultuous last days at Flexport (2 October 2023)
• FreightWaves — Flexport founder says ousted CEO lost customer focus, spending discipline (September 2023)
• The Chain / Global Supply Chain Council — Is Auger AI the future of supply chain tech, or just a $100M PowerPoint? (7 May 2025)
• Talking Logistics — The supply chain problem $100 million and smart executives can’t solve (Adrian Gonzalez, 11 December 2024)
• Gartner — SCM software with agentic AI forecast to reach $53B in spend by 2030 (7 April 2026)
• PR Newswire — Auger chosen as the premier supply chain autonomy platform on Microsoft Fabric (20 March 2026)