3 stocks. 1 thesis. Cerebras IPO + supply chain leverage.

Don't pay 50-100x revenue for Cerebras. Own the supply chain at 1-3x revenue.

Cerebras Systems IPOs May 14 at a $26.6B SEC valuation — Polymarket-implied closing market cap is $50B+. At those prices, you're paying 50-100x trailing revenue for a company with 86% customer concentration in two UAE entities (G42 + MBZUAI) and a non-GAAP operating loss of $75.7M. The math gets ugly fast. The smarter trade: buy the picks-and-shovels. Three publicly-traded companies have verified Cerebras contracts at dramatically lower valuations: DGXX ($1.1B contract on $483M market cap = 230% leverage), WYFI ($811M market cap, already operational with Cerebras revenue plus an $865M Nscale deal), and VICR (Cerebras' Power-on-Package supplier for years, just posted a Q1 2026 beat). Combined market cap of all three is $12.9B — roughly half of Cerebras' IPO valuation. If you believe in the Cerebras growth thesis, you can own the entire supply chain at half the price.

$24.6B backlog. OpenAI $20B deal. The supply chain captures it all.

To understand why these three names are leveraged to Cerebras specifically, you have to understand the Cerebras growth math. Cerebras converts 15% of its $24.6B backlog in 2026 = $3.7B implied revenue (7x growth from FY25's $510M). To deliver that, Cerebras needs: (1) chips manufactured at TSMC, (2) data center capacity to host them, (3) power delivery components for the high-density compute. Each of these picks captures a specific bottleneck — DGXX provides the data center, WYFI provides additional data center capacity (already running), and VICR provides the power delivery silicon that actually keeps the wafer-scale chips operational.

How Each Pick Captures Cerebras Growth

DGXX captures it via colocation revenue. Once Phase 1 is operational (December 2026), DGXX bills Cerebras roughly $110M/year for the next 10 years. That's 23% of DGXX's current market cap as ANNUAL recurring revenue. Phase 2 doubles that. The $1.4B expansion option adds another $140M/year potential.

WYFI captures it via two operational facilities. MTL-3 is already generating Cerebras colocation revenue. NC-1 starts billing Nscale (a Cerebras customer indirectly, also runs OpenAI workloads) in April-May 2026. The combination is a more diversified take on the same AI inference thesis.

VICR captures it via Power-on-Package ChiP-sets. Every CS-3 system Cerebras builds requires sophisticated power delivery — the wafer-scale chip pulls 15+ kW per cabinet. VICR's PoP technology is one of the few solutions that can deliver that current density at the package level. As Cerebras scales from $510M to $3.7B revenue (7x), VICR's revenue per CS-3 system stays the same but the volume grows correspondingly.

The Cerebras Backlog Conversion Schedule

Digi Power X. $1.1B Cerebras contract on a $483M company.

Digi Power X is the highest-leverage Cerebras supply chain play in the public markets. A penny stock turned AI data center operator with a single contract worth 230% of its market cap. Stock currently $6.63, up 470% over 12 months from a 52-week low of $1.16. Just signed the most transformational deal in the company's history: a 10-year Master Services Agreement with Cerebras Systems for a 40 MW AI data center in Columbiana, Alabama, valued at $1.1B initial / $2.5B with renewals / +$1.4B expansion option.

The Deal Math

The Company

Digi Power X is an AI data center infrastructure operator headquartered in Miami, Florida. Listed on Nasdaq (DGXX) and TSX Venture (DGX). The company pivoted from Bitcoin mining to AI infrastructure in 2025 — a transition that's now validated by the Cerebras anchor contract. Key operational facts:

• 400 MW of secured power across Alabama, New York, and North Carolina
• Tier III data center grade infrastructure with proprietary AI-Ready Modular Solution (ARMS) platform
• Zero debt, $93M liquid as of April 2026 (per company filings)
• ~73.5M shares outstanding — relatively tight float
• NeoCloud GPU-as-a-Service portal — secondary revenue stream beyond colocation
• On-site substation completed for Phase 1 with grid interconnection finalized via Alabama Power
• Owns underlying real property for Columbiana campus — balance-sheet-backed

The Stock Profile

The DGXX Bull Case

• Single contract = 230% of market cap. Most leveraged single-contract trade in public markets right now.
• Revenue starts December 15, 2026. Concrete catalyst date in 7 months.
• 360 MW of unbooked capacity available for additional Cerebras expansion or other tenants.
• Cerebras IPO success creates tailwind — every CBRS milestone validates DGXX anchor contract.
• Real estate ownership means asset value floor — DGXX owns the Columbiana site.
• Hans Vestberg (former Verizon CEO, BlackRock board member) is senior advisor — institutional credibility.
• Self-funding Phase 1 construction — strong financial commitment from management.
• Zero debt + $93M liquid — clean balance sheet for execution.

The DGXX Bear Case

• Single-tenant concentration risk. Cerebras IS the business. Any Cerebras stumble cascades to DGXX immediately.
• Phase 2 financing not yet secured. Will require capital raise or debt — possible dilution overhang.
• Construction execution risk. 40 MW Tier III data center on aggressive 7-month timeline.
• No revenue from contract until late 2026. Holding cost is real — 7+ months of waiting.
• Stock already up 470% in 12 months. Easy money already made — entry timing matters.
• Insider selling activity in recent SEC filings — pre-pop exits.
• Penny-stock volatility profile. Beta 2.29 — can drop 30%+ on any negative Cerebras headline.
• Bitcoin mining legacy creates partial exposure to BTC price volatility.

WhiteFiber. Cerebras revenue flowing today, plus $865M Nscale anchor.

WhiteFiber is the cleanest Cerebras play in the public markets. Already operational. Already generating revenue. Two anchor tenants (Cerebras + Nscale Global Holdings). Carved out of Bit Digital via August 2025 IPO at $17/share. Currently trading at $21.25 — only +25% from IPO price despite operational launch and major new contract. The diversification reduces single-customer risk that DGXX has, while still offering meaningful Cerebras exposure. FY25 revenue $79.2M. FY25 Adjusted EBITDA $17.3M. Real cash flow, not just contracted backlog.

The Cerebras + Nscale Deal Stack

The Company

WhiteFiber is a vertically-integrated AI infrastructure provider operating high-performance computing data centers and cloud services. Combines specialized colocation, hosting, and cloud services engineered for generative AI workloads. Originally part of Bit Digital's HPC business, carved out in August 2025 IPO. Bit Digital still holds 27M WYFI shares (~$574M at current price) and has committed not to sell in 2026. Key facts:

• Listed on Nasdaq (WYFI) since August 7, 2025 (IPO at $17, +24% to current $21.25)
• FY25 Revenue: $79.2M (full year, with Q4 being $23.6M, +61% YoY)
• FY25 Adjusted EBITDA: $17.3M — already EBITDA-positive
• Q4 2025 Colocation Revenue: $3.9M — driven by MTL-3 ramp post-Cerebras live
• IPO proceeds: $183.3M
• $230M convertible note closed January 2026 — well-capitalized for expansion
• FY25 capex: $268M — building real capacity
• NC-1 has 99 MW Duke Energy capacity pact (expandable to 200 MW)
• Tier 3-equivalent design with 150 kW per cabinet density, N+1 cooling, PUE 1.3

Why WYFI Is Uniquely Positioned

WhiteFiber is the only public AI infrastructure pure-play with two confirmed anchor tenants (Cerebras + Nscale) and operational revenue from Cerebras flowing today. The diversification matters: if Cerebras stumbles, Nscale carries the NC-1 economics. If Nscale stumbles, the Cerebras MTL-3 facility keeps generating revenue. The structure offers Cerebras leverage without single-tenant risk.

The Nscale partnership is also indirectly Cerebras-relevant — Nscale runs OpenAI workloads, and OpenAI is Cerebras' biggest customer. Effectively, WYFI captures Cerebras growth from two angles: directly via the MTL-3 contract, and indirectly via Nscale serving OpenAI demand that Cerebras helps fulfill.

The Stock Profile

The WYFI Bull Case

• Already operational and revenue-generating — not a "future revenue" story.
• FY25 Adjusted EBITDA positive at $17.3M — actual cash flow.
• Two anchor tenants reduce concentration risk — Cerebras + Nscale.
• $865M Nscale deal locked in with billing starting April-May 2026.
• Cerebras has ROFR on additional MTL-3 capacity — growth path embedded.
• NC-1 expandable to 200 MW via Duke Energy capacity pact.
• Bit Digital won't sell shares in 2026 — supply overhang removed.
• $183M IPO proceeds + $230M convertible — well-capitalized for expansion.
• Currently trades only +25% from IPO price despite operational milestones — multiple expansion potential.
• B. Riley $40 PT implies +88% upside from current.

The WYFI Bear Case

• Carve-out structure means Bit Digital still owns 33%+ of float. Future selling pressure when 2026 ends.
• $268M capex in FY25 on $79M revenue — heavy infrastructure investment phase.
• Nscale revenue ramp dependent on hyperscaler customer execution.
• NC-1 Phase 2 financing needed for full 40 MW deployment.
• Tier 3 vs Tier 4 positioning — premium customers may demand higher resilience.
• Stock already up 25% from IPO — base hit, not multi-bagger from here.
• Convertible note dilution overhang at conversion price.
• Competition from established colocation (EQIX, DLR) at scale.

Vicor Corporation. Quiet large-cap with Cerebras supplier relationship since 2023.

Vicor is the surprise on this list — a $11.62B large-cap power semiconductor company that's quietly been Cerebras' Power-on-Package supplier since the 2023 era. Per SemiAnalysis: "Vicor went from a supplier of commodity power components to entering advanced datacenter power applications with design wins in various hyperscaler datacenter rack-level power solutions and AI accelerators from Nvidia, Google, AMD, Cerebras, Tesla, and Intel." Q1 2026 just printed a clean beat: revenue +20% YoY, EPS $0.52 vs $0.41 est (+27% surprise), backlog +70% to $300.6M, book-to-bill above 2.0. Stock up 413% in 12 months. Different position-size profile than DGXX/WYFI — this is a blue-chip large-cap, not a small-cap pure-play.

The Cerebras Connection

Vicor's Power-on-Package (PoP) ChiP-set is one of the only solutions that can deliver the high-current density required by wafer-scale processors at the package level. The CS-3 system needs power delivered with extreme precision because the WSE-3 chip is roughly 21.5cm on each side and pulls 15+ kW per cabinet. Conventional 12V multi-phase voltage regulators can't keep up. Vicor's 48V→1V conversion at 1,000A peak current with current multiplication factors up to 40 is one of the few solutions that works at this scale.

The Q1 2026 earnings call had an explicit Cerebras-related question from Justin Clare (Roth Capital): "We saw a large transaction announcement between OpenAI and a wafer-scale supplier last week." Management's response acknowledged engagement with HPC customers for second-generation VPD solutions. Translation: Cerebras is real revenue for Vicor, and the OpenAI deal is the catalyst for Vicor's Fab Two planning.

Q1 2026 Financial Highlights

The Capacity Story

Vicor management outlined a clear path on the Q1 2026 call: Fab One can scale to $1.5B annual revenue run rate with targeted process outsourcing. Plans to acquire existing buildings to accelerate Fab Two ramp. The capacity expansion is the key bull case — Vicor isn't just riding existing demand, they're building infrastructure to capture the next wave of AI accelerator launches (including future Cerebras WSE-4 generations).

The Stock Profile

The VICR Bull Case

• Multi-customer hyperscaler exposure — Nvidia, Google, AMD, Cerebras, Tesla, Intel design wins.
• Q1 2026 backlog +70% QoQ = strongest demand signal in years.
• Book-to-bill above 2.0 = orders coming in 2x faster than billing.
• Path to $1.5B revenue run rate from existing Fab One alone (3x current).
• Fab Two planning underway — capacity expansion already in motion.
• Patented technology + licensing income — Qualcomm-style royalty model.
• $404M cash, $50M total liabilities — fortress balance sheet.
• Analyst consensus PT $282.50 implies +10% from current.
• Cerebras OpenAI deal = direct catalyst for Vicor demand acceleration.
• Royalty income inflection from licensing partners (Sony, others) provides margin upside.

The VICR Bear Case

• Already up 413% in 12 months — massive run-up. Easy money made.
• Trading at 20x forward revenue — premium multiple even for AI infrastructure.
• Forward P/E 85-130x — pricing in significant execution.
• Competition from MPWR (Monolithic Power) — historically took share from Vicor at Nvidia.
• Not exclusive Cerebras supplier — Cerebras could dual-source.
• Single-fab risk — Fab Two not online for years.
• Management commentary "selective adding customers" — capacity-constrained which is good for pricing but limits upside.
• Insider selling — VPs Tuozzolo ($1.7M) and Fendelet ($1.4M) recent sales.
• Litigation settlement payment $28.6M in Q1 — historical legal overhangs persist.
• Large-cap not small-cap — different position-sizing profile than DGXX/WYFI.

Same thesis. Three different conviction tiers.

The three picks all play the same Cerebras growth thesis, but at fundamentally different risk-reward profiles. Understanding which one fits your conviction level and risk tolerance is essential to position correctly.

How They Stack On Specific Catalysts

Real exposure to Cerebras growth. Real downside risks if it stumbles.

Asymmetric upside on the leveraged play. Steady compounding on the diversified ones.

Honest scenario map for each name over the next 12-18 months. Position-sizing should match each name's volatility profile, not the upside potential.

DGXX — High Beta Single-Tenant Play

WYFI — Medium Beta Operational Play

VICR — Lower Beta Diversified Compounder

The Combined Catalyst Calendar

How does this trio stack up against alternative ways to play the AI infrastructure thesis?

There are dozens of ways to play AI infrastructure. The question is which approach offers the cleanest risk-reward at current valuations. We picked five peers that represent different angles on the same theme: CBRS itself (direct exposure), CRWV (cloud operator), APLD (other AI data center pivot), CRDO (specialty AI silicon), and AVGO (custom ASIC). Each gets you AI infrastructure exposure differently — and the math says the DGXX + WYFI + VICR combination is structurally more attractive than most.

The Trade-Off Analysis

vs CBRS Direct — Why Buy The Supply Chain Instead

Buying CBRS at IPO ($26.6B) or post-Polymarket ($50B+) means paying 50-100x trailing revenue for direct exposure. The combined three-stock supply chain at $12.9B gives you the same thesis at half the valuation, with the bonus that VICR alone has multiple AI customers (Nvidia, AMD, Tesla) so it's not pure Cerebras risk. If Cerebras stumbles, VICR still wins. If you bought CBRS, you wouldn't.

vs CRWV — Different Layer Of The Stack

CoreWeave is a cloud operator, not a Cerebras supplier. They rent Nvidia GPUs. Different business entirely. CRWV at $70B trades at 13x forward revenue — also rich. The DGXX+WYFI+VICR combination at $12.9B is dramatically cheaper for similar AI infrastructure exposure with direct Cerebras leverage that CRWV doesn't have.

vs APLD — The "Other" Crypto-To-AI Pivot

Applied Digital is the most analogous comparison to DGXX — also a Bitcoin-mining-pivot to AI data centers. APLD has hyperscaler customer exposure but NO confirmed Cerebras anchor. APLD at $2.5B vs DGXX at $483M means DGXX has 5x the leverage to a single contract. If you want pure-play data center pivot exposure, DGXX is more leveraged. If you want diversified data center pivot exposure, APLD is the safer bet at higher valuation.

vs CRDO — Specialty AI Silicon Premium

Credo Technology is in similar territory to VICR — specialty AI silicon at 12x forward revenue. CRDO is a high-speed connectivity play (SerDes, retimers). VICR is power delivery. Both benefit from AI infrastructure buildout. CRDO has a slightly cleaner growth trajectory but trades at similar multiples. The pair-trade thesis: own VICR + CRDO together for diversified specialty AI silicon exposure.

vs AVGO — The Diversified Custom ASIC Play

Broadcom designs custom ASICs for hyperscalers (Google TPU, Meta MTIA, ByteDance). $1.3T market cap. Trades at 16x forward revenue. The most diversified AI silicon play but at a much larger size. AVGO doesn't supply Cerebras directly — they compete for hyperscaler dollars. If you want one-stop AI silicon exposure, AVGO is the safest. If you want Cerebras-specific leverage, the trio is the answer.

Three names. Three position sizes. One thesis at half the valuation.

The Cerebras IPO is one of the most important AI infrastructure listings of 2026. The technology is real (verified 2-21x faster than Nvidia Blackwell on multiple benchmarks). The customers are real (OpenAI $20B deal, AWS term sheet). The backlog is real ($24.6B). But the IPO valuation prices in nearly perfect execution at 50-100x trailing revenue. The supply chain offers the same thesis at dramatically better entry points.

The defensible read: DGXX, WYFI, and VICR each capture a specific layer of Cerebras' growth. DGXX provides anchor data center capacity at a single-contract leverage of 230% of market cap. WYFI is already operational with Cerebras revenue flowing today plus Nscale diversification. VICR is the blue-chip large-cap with multi-customer exposure that benefits from any AI accelerator vendor's success — Cerebras, Nvidia, AMD, or Tesla. Combined, they offer exposure to the same AI inference thesis at half the valuation of CBRS itself.

The honest read: Each name has real risks. DGXX is single-tenant concentrated and beta 2.29 — it can drop 50% on any negative Cerebras headline. WYFI has a Bit Digital lockup overhang at end of 2026 and is in heavy capex mode. VICR has already run 413% in 12 months and trades at 20x forward revenue — most of the easy money has been made. Position-sizing matters more than direction here.

This is a setup with verifiable catalysts + multiple risk tiers + reasonable valuations vs the CBRS IPO directly. The trade isn't betting on one — it's owning a basket sized to your conviction in Cerebras' execution.