Diagnostics sits at the center of precision medicine, yet, in the U.S., the industry is being pulled in two opposing directions. On one side, outsourced reference labs like Caris, Tempus, Guardant, and Foundation Medicine are consolidating share, building comprehensive assay menus and data platforms at scale. On the other, we see a countercurrent of decentralization: hospitals, academic centers, and even pharma partners looking to bring testing closer to patients, integrate it into EMRs, and capture more of the value locally.

The strategic question is not whether the market will expand; it will. The question is who captures the value as testing shifts across these two models.

In this article, we share our perspective on decentralization based on internal DeciBio databases and tracker, layering the perspective of Joseph Annunziata, a KOL we interview for many of our engagements.

TL;DR: Hospitals want to decentralize, but IT and staffing —among many other barriers listed below— often make it a mirage outside top (academic) centers. Reference labs aren’t just winning on scale; they’re winning on operational excellence and simplicity. This will remain the case for the next decade unless hospitals fundamentally rewire their operating model.

‍

Reference Labs: Scale, Breadth, and Data Gravity

A recent survey with 100 oncologists that we conducted shows the reference lab model is still dominant: Foundation, Tempus, Caris, and Guardant together account for >60% of genomic profiling in the U.S. Their pitch is compelling: breadth of assays, strong salesforces, faster turnaround times, and increasingly, unmatched databases that attract pharma partners. Analysts reinforce this, with Caris leading in tissue CGP and Tempus setting the pace with its $900M+ data backlog.

The reference lab model thrives on outsourcing economics: hospitals and oncologists avoid the cost and complexity of running high-complexity assays in-house, while labs benefit from volume, scale efficiencies, and the ability to monetize data across customers. Economies of scale and network effects from strong variant databases is a clear competitive advantage.

‍

Decentralization: Control, Integration, and Clinical Proximity

Yet the decentralization trend is real. Many oncologists (especially at academic centers) expect to increase use of in-house assays or local solutions over the next 12 months. The drivers are clear:

• Integration into EMRs and care pathways, making testing results more actionable at the point of care.
• Faster reflex testing, especially where tissue is available in-house.
• Economic capture, as institutions seek to avoid sending high-value samples (and revenue, now that reimbursements are increasingly available) to external labs.

Tools players like Illumina and ThermoFisher are positioned to benefit from this decentralization, equipping local labs with instruments and reagents that can deliver parts of the CGP (or MRD workflow) closer to the bedside. However, as Dr. Annunziata highlights, many barriers exist:

• IT integration: LIS/EMR integration (e.g., Epic Beaker) can take 6-12 months in community settings, delaying test launches more than capital does.
• Capital intensity & infrastructure gaps: High-complexity assays (CGP, MRD) require costly instruments, bioinformatics pipelines, and quality systems that many hospitals aren’t equipped to build or maintain.
• Reimbursement uncertainty: Even as coverage expands, payer contracts are typically negotiated at the reference lab level. Local labs face longer timelines and more friction in securing consistent reimbursement.
• Billing complexity: In the same vein, navigating coding, prior authorizations, denials, and appeals is resource-intensive. Reference labs have large, specialized billing teams, while hospitals or academic labs often lack this sophistication, creating revenue leakage.
• Operational complexity: Running CLIA labs with NGS, bioinformatics, and variant interpretation requires specialized staff. Academic centers may manage this, but community hospitals often cannot.
• Regulatory risk: With LDT reform and FDA oversight intensifying (despite the recent vacating of the FDA LDT ruling), decentralized labs face higher compliance burdens that reference labs are better resourced to navigate.
• Data fragmentation: Reference labs create value not just from running tests, but from aggregating and monetizing data at scale. Decentralized testing risks producing siloed datasets that lack pharma utility.
• Physician engagement & training: Outsourced labs deploy large, specialized salesforces that engage oncologists daily. Local labs struggle to replicate this commercial infrastructure, which slows adoption.
• Operational staffing economics: Hospital technologists are often paid on the same scale regardless of specialization, so molecular techs don’t get rewarded for higher skill / lower error risk. Commercial labs, by contrast, can pay more competitively and reroute applicants, letting them build specialist teams faster. On top of that there is a “competing with ourselves” issue within hospitals that can’t reassign chem tech applicants to molecular roles, which exacerbates the talent crunch.

‍

MRD as a Battleground

Minimal residual disease (MRD) is perhaps the clearest arena for this outsourcing vs. decentralization tension. The field is now largely centralized. Guardant’s tumor-naïve Reveal and Natera’s Signatera are dominant and scaling via a reference lab model. But both Tempus and Natera are hedging, with both tumor-naïve test and a tumor-informed assay available, with the ambition of becoming a one-stop shop. Meanwhile, academic centers are piloting their own MRD assays, raising the question: will MRD become another test hospitals “own,” or will it consolidate under outsourced players who can generate payer coverage and industrialized throughput? In the short term, we believe that this relatively complex assay will remain centralized, but decentralization forces are real. This message was echoed by Dr. Annunziata: “MRD is especially unlikely at community hospitals for a decade, since we’re still climbing the ladder of hotspot panels to heme panels to CGP, and only then will be get to liquid biopsies”.

Therefore, MRD adoption stands as a stark illustration of centralization’s advantage in the U.S.: commercial labs are years ahead not just because of science, but because hospital IT and staffing economics make them structurally unable to prioritize MRD until they’ve fully digested earlier panels. This reframes MRD as less about science risk, more about system inertia.

‍

Salesforce Reality: Outsourcers Still Win on Influence

Another insight: reference labs dominate physician interactions. Guardant ranks #1 in sales rep engagement across both community and academic oncologists, with Tempus, Foundation, and Natera not far behind. In contrast, NGS vendors are trailing far behind. In a market where adoption hinges on awareness, training, and trust, the ability to consistently engage physicians is a moat that decentralized solutions struggle to replicate.

‍

A Hybrid Future?

Ultimately, the U.S. market may resolve into a hybrid equilibrium:

• Reference labs as the “platform hubs”, providing the breadth of testing, payer contracting, and data services pharma needs. That makes reference labs the de facto backbones of precision medicine for the next decade in the U.S.
• Decentralized assays as the “last mile”, integrated into local care, enabling faster answers for common cases and expanding access.

This model however assumes that hospitals address the barriers listed above, so that decentralization doesn’t remain aspirational. And if it happens, the strategic winners will be those who navigate both worlds! Importantly, markets outside the U.S. may follow a different trajectory, with many European countries favoring a more decentralized approach from the start.

‍

Conclusion: Power Is Shifting, Not Disappearing

The tension between outsourcing and decentralization is not a zero-sum game, but it will reshape the power dynamics of diagnostics. Reference labs may retain scale and data gravity, but decentralization is a check on their dominance, keeping hospitals and payers engaged in the value chain.

Diagnostics is no longer just about running tests. It is about designing systems of care that balance scale, access, and data. The sector is finally at an inflection point, and the strategic choices made now will determine who builds the backbone of precision medicine in the decade ahead.

Unless hospitals reinvent their models, the backbone is already being built by reference labs. The strategic question is not whether they will dominate; it’s who, if anyone, can challenge their momentum. While this article makes the case for reference labs to remain dominant in the U.S. for the decade to come, we acknowledge the possibility that the barrier we list may be alleviated and that other “blackswan events” may disrupt the space. What if AI-driven workflow automation makes LIS integration a commodity? Would that collapse the moat of reference labs?

‍

‍

‍

_______________________

Note: Some of the companies listed in this article may be DeciBio Consulting clients.

Author: Stephane Budel, Partners at DeciBio Consulting, LLC

Connect with Stephane on LinkedIn