PM Portfolio Manager Reports

From Portfolio Manager reports

Scientific Compute And Foundation Models

This page covers the scientific compute and foundation-model layer that lets researchers train, host, and use large biological models. The stack includes GPU and accelerator servers, cloud high-performance computing, custom silicon, model APIs, managed omics workflows, scientific databases, and enterprise research tools. It sits in hyperscale cloud regions, private AI clusters, pharma research environments, and software workflows used by computational biologists, chemists, and lab teams. The layer matters when protein, genomics, chemistry, simulation, and lab-agent workloads move from research pilots into repeat production: hardware ships, cloud usage rises, model services are paid for, and backlog or remaining performance obligations turn into revenue and cash flow. NVDA has the clearest direct biology-workflow evidence through BioNeMo and life-science partners; GOOGL owns the strongest public model lineage through DeepMind and Isomorphic Labs; MSFT and AMZN have credible enterprise and cloud workflow routes; ORCL, AMD, and AVGO are useful but less direct infrastructure exposures.

Report boundary: this node is a tactical report layer. It ranks the current value-chain basket, confirmation triggers, invalidation checks, and chart provenance. Durable research, claim IDs, raw source registries, and sector thesis maintenance stay in the linked knowledge pages. Price and volume context comes from read-only discovery daily_ohlc through 2026-06-12.

Current Setup

Infrastructure monetization Biology model evidence is strongest where it attaches to reported cloud or Data Center revenue.

Use the node for compute and model-platform economics, then route clinical, wet-lab, diagnostics, and TechBio evidence to adjacent nodes.

Best direct bridge NVDA connects biology tools to the largest reported AI infrastructure revenue pool. BioNeMo, Clara, NIM, DGX, and life-science partners sit next to Q1 FY2027 Data Center revenue of $75.2B.
Model ownership GOOGL has the clearest public biology-model lineage. AlphaFold, AlphaGenome, DeepMind, Isomorphic Labs, and Google Cloud define the model route, but direct biology revenue is undisclosed.
Disclosure gap No basket member reports biology-specific revenue. Watch paid production usage, cloud margins, capex, customer concentration, export rules, and whether lab evidence changes workflow decisions.
  1. InputData + wet-lab needs
  2. StackGPU, cloud, models
  3. Revenue routeUsage, RPO, services
  4. GatePaid production proof

Right-rail charts use weekly bars aggregated from discovery daily_ohlc through 2026-06-12. Static setup levels from the older 2026-06-05 table were removed; use the API chart before treating a price level as current trading evidence. Fundamental claims route to the AI Biology Platform Validation concept, AI Capex Cycle concept, Information Technology sector lane, and linked security lanes.

Basket

This basket is inherited from the parent theme and hand-ranked against local knowledge coverage, official result releases, bounded web checks, and read-only discovery coverage. Ranking uses direct biology-model or scientific-compute evidence first, then public-company monetization route and disclosure quality. Recent price strength does not substitute for node relevance. SDGR, RXRX, DNA, TEM, ABCL, ABSI, and CERT stay in adjacent TechBio, data, lab, and clinical-evidence nodes because their evidence is clinical, partner, software ACV, wet-lab utilization, or funding rather than scaled compute and foundation-model infrastructure.

NVDA NVIDIA

NVIDIA is the most direct compute monetization bridge because BioNeMo and life-science partner evidence attach to the largest reported Data Center revenue pool.

Market cap$4.99T
Next earningsNot confirmed
Latest qtr revenue$81.6B

Role in stack

NVIDIA sells GPUs, networking, DGX systems, BioNeMo, NIM, Clara, and cloud-deployable AI software to hyperscalers, pharma companies, biotech model builders, lab-automation partners, and tool vendors. Biology demand becomes company economics when those customers run protein, chemistry, genomics, and lab-agent workflows on repeat NVIDIA compute and software.

Revenue mix

The linked lane frames NVIDIA as primarily a Data Center AI infrastructure company. Q1 FY2027 Data Center revenue was $75.2B, about 92% of total revenue. Biology-specific revenue is not disclosed.

Latest qtr revenue

Q1 FY2027 total revenue was $81.615B for the quarter ended April 26, 2026, from NVIDIA's May 20, 2026 official earnings release. The next event date was not visible on NVIDIA's events page; Nasdaq returned no earnings date during the June 13, 2026 check.

GOOGL Alphabet

Alphabet owns the strongest public biology-model lineage through DeepMind, AlphaFold, AlphaGenome, Google Cloud, and Isomorphic Labs.

Market cap$4.36T
Next earningsNot confirmed
Latest qtr revenue$109.9B

Role in stack

Alphabet supplies model research, TPU and cloud infrastructure, and drug-design capabilities through Google DeepMind, Google Cloud, and Isomorphic Labs. Biology demand converts through Google Cloud usage, model access, pharma partnerships, and eventual Isomorphic milestones rather than through a separately reported biology segment.

Revenue mix

Google Services funds most company economics, while Google Cloud is the visible AI infrastructure route. Local knowledge and bounded web checks report Q1 2026 Google Cloud revenue of $20.0B. DeepMind and Isomorphic biology economics are not separately disclosed.

Latest qtr revenue

Q1 2026 total revenue was $109.9B for the quarter ended March 31, 2026, from Alphabet Q1 earnings coverage reviewed on June 13, 2026. Alphabet's events page did not show a confirmed next earnings date in the accessible page snapshot.

MSFT Microsoft

Microsoft has the best enterprise software and governed-cloud route into scientific R&D workflows.

Market cap$2.90T
Next earningsNot confirmed
Latest qtr revenue$82.9B

Role in stack

Microsoft sells Azure AI and HPC capacity, Microsoft Discovery, Azure AI Foundry, GitHub/Copilot workflow entry, and governed enterprise research tools to life-science, biology, chemistry, materials, semiconductor, and engineering customers. Conversion depends on Azure consumption, enterprise licenses, private-data workflows, and paid research workspaces.

Revenue mix

Enterprise software and Azure carry the economics. The local lane reports Microsoft Cloud revenue of $54.5B, Azure and other cloud services growth of 40%, and commercial RPO of $627B for Q3 FY2026. Biology-specific revenue is not disclosed.

Latest qtr revenue

Q3 FY2026 total revenue was $82.9B for the quarter ended March 31, 2026, from Microsoft's April 29, 2026 official earnings release. Microsoft's events page was accessible but did not expose a confirmed next earnings date in the June 13, 2026 check.

AMZN Amazon

Amazon is the direct cloud workflow route through AWS HealthOmics, storage, compute, model services, and orchestration.

Market cap$2.57T
Next earningsNot confirmed
Latest qtr revenue$181.5B

Role in stack

Amazon sells AWS HealthOmics, EC2, SageMaker, Bedrock, storage, workflow orchestration, and partner model access to pharma, biotech, diagnostics, and research teams. Biology demand converts through AWS compute, storage, managed pipelines, and model-service usage.

Revenue mix

AWS is a minority of consolidated sales but the main node exposure and profit pool. Q1 2026 AWS sales were $37.6B and AWS operating income was $14.2B in local knowledge and earnings coverage. HealthOmics and biological foundation-model usage are not separately disclosed.

Latest qtr revenue

Q1 2026 net sales were $181.5B for the quarter ended March 31, 2026, from Amazon Q1 earnings coverage reviewed on June 13, 2026. The accessible Amazon investor event route did not provide a confirmed next earnings date.

ORCL Oracle

Oracle is the OCI and life-sciences data-platform route, with relevant capacity exposure and heavier financing checks.

Market cap$0.53T
Next earningsNot confirmed
Latest qtr revenue$19.18B

Role in stack

Oracle provides OCI GPU supercomputing, cloud infrastructure, database, vector search, and Oracle Life Sciences data-platform routes for scientific and life-sciences workloads. Conversion depends on OCI consumption, RPO conversion, database pull-through, life-sciences data workflows, and funded customer commitments.

Revenue mix

Cloud and software dominate the company, with OCI as the fast-growth route. The linked lane has Q3 FY2026 OCI revenue of $4.9B and RPO of about $553B. Post-Q4 coverage says OCI revenue rose to $5.8B while capex and funding needs became a larger constraint.

Latest qtr revenue

Fiscal Q4 2026 revenue was $19.18B, reported in June 2026 earnings coverage after Oracle's June 10 release window. A confirmed next earnings date was not available from the bounded checks, so the card does not use a calendar estimate.

AMD Advanced Micro Devices

AMD is the second-source scientific compute route through EPYC CPUs, Instinct GPUs, ROCm, and high-memory accelerator platforms.

Market cap$0.83T
Next earningsNot confirmed
Latest qtr revenue$10.253B

Role in stack

AMD sells EPYC server CPUs, Instinct accelerators, ROCm software, and high-memory platforms used by scientific HPC and cost-sensitive AI biology workloads. Biology demand converts through Data Center CPU and GPU revenue when life-science users choose AMD as a credible second-source compute platform.

Revenue mix

Data Center is the relevant segment. AMD reported Q1 2026 Data Center revenue of $5.775B, or about 56% of total revenue, and said the segment includes AI accelerators, server CPUs, GPUs, APUs, DPUs, AI NICs, FPGAs, and adaptive SoC products.

Latest qtr revenue

Q1 2026 total revenue was $10.253B for the quarter ended March 28, 2026, from AMD's May 5, 2026 official earnings release. A confirmed next earnings date was not found in the bounded June 13, 2026 checks.

AVGO Broadcom

Broadcom supplies custom AI silicon and networking used inside large AI clusters, but its biology link is indirect.

Market cap$1.81T
Next earningsNot confirmed
Latest qtr revenue$22.187B

Role in stack

Broadcom supplies custom AI accelerators, Ethernet fabric, SerDes, networking, and infrastructure software to hyperscalers and model platforms. Biology demand is indirect: scientific models may run on clusters that use Broadcom silicon, but Broadcom does not own a visible biology foundation-model product or pharma workflow.

Revenue mix

Q2 FY2026 revenue was split between Semiconductor Solutions at $15.009B and Infrastructure Software at $7.178B. AI semiconductor revenue was $10.8B. Biology-specific revenue is not disclosed.

Latest qtr revenue

Q2 FY2026 total revenue was $22.187B for the quarter ended May 3, 2026, from Broadcom's June 3, 2026 official earnings release. A confirmed next earnings date was not available; Broadcom only provided Q3 FY2026 guidance for the quarter ending August 2, 2026.

Market caps are calculated from local discovery adjusted closes on 2026-06-12 multiplied by stored weighted shares outstanding. Next-earnings fields are marked Not confirmed because the checked company event pages and Nasdaq earnings pages did not expose reliable future dates during the June 13, 2026 update. Latest-quarter revenue uses reported quarterly total revenue, not annualized revenue.

What Confirms Or Weakens

Area What confirms What weakens or invalidates Watch next
01 Node thesis Compute becomes production biology
What confirms

Named production deployments where pharma, tool vendors, model builders, or research organizations run biology workloads on BioNeMo, AlphaFold or AlphaGenome, Microsoft Discovery, AWS HealthOmics, OCI, AMD compute, or Broadcom-backed capacity.

What weakens or invalidates

Announcements remain pilots, free research access, demos, or broad AI claims without recurring usage, cloud consumption, model-service revenue, accepted lab decisions, or workflow renewals.

Watch next
  • BioNeMo production use
  • AlphaGenome access
  • HealthOmics pipelines
  • Microsoft Discovery tenants
02 Economics mechanism Usage becomes revenue
What confirms

Compute and model usage converts into Data Center revenue, cloud revenue, model-service usage, RPO or backlog conversion, software attach, margin, and free cash flow after capex.

What weakens or invalidates

Capex, leases, depreciation, power, components, customer financing, or equity dilution rise faster than revenue, utilization, operating income, or free cash flow.

Watch next
  • NVDA Data Center guide
  • GOOGL Cloud margin
  • AWS margin
  • ORCL RPO conversion
03 Customer and workflow Lab decisions improve
What confirms

Model output is tied to measured wet-lab decisions: target selection, protein or molecule ranking, assay design, genomics interpretation, lab automation throughput, candidate packages, or partner option exercises.

What weakens or invalidates

Benchmarks improve but customers do not report better hit rates, fewer design-test cycles, paid renewals, partner payments, or reproducible lab evidence.

Watch next
  • Pharma AI factories
  • Thermo/NVIDIA lab workflows
  • Isomorphic milestones
  • Customer renewals
04 Funding and capex Infrastructure stays funded
What confirms

Cloud providers and AI infrastructure vendors keep customer breadth, funded backlog, power access, GPU supply, export compliance, and gross margin intact.

What weakens or invalidates

Customer concentration rises, export controls tighten, power availability delays clusters, large customers renegotiate economics, or supplier commitments outrun cash conversion.

Watch next
  • China export rules
  • Oracle financing
  • AMD commitments
  • Broadcom customer breadth
05 Policy and data control Private research remains deployable
What confirms

Providers support private-data workflows, regulated research controls, export-compliant compute, pharma partner requirements, secure model deployment, and auditability for scientific decisions.

What weakens or invalidates

Data residency, intellectual-property, model-governance, export-control, or pharma validation concerns push buyers to internalize the stack or slow external platform adoption.

Watch next
  • Private cloud controls
  • Model governance
  • Export-compliant SKUs
  • Pharma validation language
06 Operating and supply constraint Clusters can be delivered
What confirms

GPU, accelerator, networking, memory, power, cooling, and cloud-region capacity improve while service reliability and data-governance controls stay adequate for production research workloads.

What weakens or invalidates

HBM, packaging, networking, data-center power, customer concentration, or supply commitments delay capacity or lower the margin captured by compute and cloud providers.

Watch next
  • HBM availability
  • Power delivery
  • Cloud capacity
  • Reliability disclosures
07 Stale condition Refresh trigger
What confirms

Discovery daily_ohlc remains current, company event pages still show no confirmed next earnings dates, and no material BioNeMo, AlphaGenome, Discovery, HealthOmics, OCI, AMD deployment, or Broadcom AI revenue update has arrived since the cited checks.

What weakens or invalidates

A new trading session, earnings release, customer win, export-policy change, financing update, or cloud-capex disclosure changes the basket evidence before this page is refreshed.

Watch next
  • Refresh discovery
  • Recheck events pages
  • Update source trail
  • Review right-rail charts

Source Trail

Canonical Thesis

  • AI Biology Platform Validation for the biology proof ladder, adjacent TechBio routes, and evidence boundary between platform announcements and durable scientific or commercial proof.
  • AI Capex Cycle for infrastructure capex, cloud, semiconductor, custom-silicon, and funding-conversion evidence gates.
  • Funding Dependency And Duration Tolerance for capex, lease, depreciation, customer-financing, and free-cash-flow checks used in the compute economics rows.
  • Information Technology sector lane for AI infrastructure, cloud, semiconductor, software, export-policy, and capex-return context.
  • Health Care sector lane for the pharma, biotech, diagnostics, data-governance, and regulated-workflow context behind the biology customer claims.

Security Lanes

  • NVDA security lane for Data Center revenue, margin, customer concentration, export controls, commitments, and AI infrastructure durability.
  • GOOGL security lane for Google Cloud, capex, financing, regulatory, and Alphabet-level AI infrastructure evidence.
  • MSFT security lane for Microsoft Cloud, Azure, RPO, Copilot, capex, and FCF conversion.
  • AMZN security lane for AWS sales and operating income, capex, commitments, and FCF recovery.
  • ORCL security lane for OCI growth, RPO, debt, leases, power, and FCF risk.
  • AMD security lane for Data Center, Instinct, EPYC, ROCm, customer milestones, warrants, and accelerator margin proof.
  • AVGO security lane for custom AI silicon, AI networking, customer concentration, TSMC dependence, software cash flow, and FCF.

Bounded Web Checks

Discovery And Chart Provenance

  • Discovery status checks and read-only DuckDB queries showed local daily_ohlc coverage through 2026-06-12 for NVDA, GOOGL, MSFT, AMZN, ORCL, AMD, and AVGO.
  • Market caps were calculated from 2026-06-12 adjusted closes multiplied by stored weighted shares outstanding in the read-only discovery instruments table: NVDA $4.99T, GOOGL $4.36T, MSFT $2.90T, AMZN $2.57T, ORCL $0.53T, AMD $0.83T, and AVGO $1.81T.
  • Weekly bars were derived from daily_ohlc using first open, maximum high, minimum low, final close, and summed volume by calendar week. EMAs were computed from weekly closes using full available history before reading the latest three-year chart window through the report API.
  • Right-rail charts use /api/securities/{ticker}/chart?frequency=weekly&window=3y&as_of=latest and should render in basket order: NVDA, GOOGL, MSFT, AMZN, ORCL, AMD, AVGO.
  • Review validation started the local report API and verified the representative NVDA route at /api/securities/NVDA/chart?frequency=weekly&window=3y&as_of=latest. The response resolved as_of=latest to 2026-06-12 and returned weekly chart rows from reports.weekly_ohlc+discovery.daily_ohlc.

Known Gaps

  • Biology-specific revenue is not disclosed for any basket member. The page treats biology evidence as workflow and product adoption unless a segment, RPO, ACV, or revenue bridge is explicitly cited.
  • SDGR, RXRX, DNA, TEM, ABCL, ABSI, and CERT are not excluded from the theme; they are routed to computational-design, TechBio, lab, data, and clinical-evidence nodes where clinical, partner, software ACV, wet-lab utilization, or funding evidence matters more.
  • ANET, MRVL, and ALAB are valid AI cluster suppliers, but the current parent node does not own networking and interconnect as a separate biology basket. They remain optional infrastructure watch names for future parent expansion.
  • Price and volume context supports right-rail chart review only. It does not establish why a stock moved without same-window earnings, filing, customer, policy, or supply evidence.
  • Static setup thresholds from the prior 2026-06-05 table are stale relative to 2026-06-12 discovery coverage and were intentionally removed from the article body. Refresh the API chart and source trail before using specific price levels as current trading evidence.

Categories