Wolfspeed pitch i put together for prem subs. I think its still very compelling as not much has changed in tone on name since I covered this when it remerged from bk. Think it still has that bk stigma which seems silly at this point considering rapdily rising strategic value of the asset for datacenter infra buildouts. Has had huge move past 30 days like everything else semi related, but that was after falling 50% again. Note earnings are today and this isn't exactly and earnings print levg name as you will get from the piece. In other related news I exited Atlassian yesterday which had been largest ongoing position over this past crazy month.

I am going to try walk through why this is a clean asymmetric setup and that holding, despite the move thus far, makes most sense.

The pitch has three legs: an enterprise value disconnect anchored on September 29, 2025 fresh-start accounting re-emergence valuation that hasn't moved despite the entire AI Datacenter-exposed semiconductor complex getting violently repriced (subjective but very relevant imho/also great relative compares in power here), a tight-float dc AI capex derivative dynamic that the comparable AI power names have already demonstrated will work explosively, and a quantifiable doubling of the AI data center SiC market over the last three quarters that feeds a fundamental narrative that has gotten little to no attention thus far.

I am going to emphasize that as we approach the current earn print this isn’t about operational execution upside this q (i.e mxl). You are also not doing a multi-year dcf on an absurd intrinsic valuation opportunity (i.e team) for this because that's the wrong framing for this asset at this point in time. This thing is more like a hybrid of my MP pitch from last year before Govt/Apple/China 4xd it and relative catch-up to where every other AI-exposed semiconductor name has already been repriced since September 2025. I then would argue you can layer on some interesting mechanical optionality with it having tightest effective float in the entire space. I also think you still have a decent bk stigma overhang which at this point works in your favor as many folks are default cautious on pitching something that just went under even if in many ways created the best parts of the go fwd opportunity. Anyway, here you go, and obviously plenty of supreme intelligence analyst assist in parts of this.

.

Part One: SIC AI FOR Dummies

What Silicon Carbide Actually Does in AI Data Centers?

I’d argue write now that most AI DC generalists bucket SiC is a 'future technology' that 'might' matter when 800V architectures arrive. Not exactly high bottleneck trade levg. They probably think of it as the EV power semiconductor story currently in trough conditions with some little breeze of datcenter ifra in air. Both framings are wrong.

SiC is already deployed across multiple distinct revenue streams in current-generation AI data centers. The architecture transition between 2025 and 2030 expands SiC content per megawatt by 7-10x. The right way to think about this is three phases of architectural deployment, each with materially different SiC economics.

The three-phase deployment framework

The transition from current-generation data centers to AI-native MW-scale rack architecture happens in three discrete phases, each with characteristic SiC content per MW:

•      Phase 1 — Peripheral insertion (2025-2026). SiC sits at specific points in the conversion chain — UPS inverters, PDU rectification, BESS interfaces. Architecture remains 480V AC distribution with 48V intermediate stages. Racks at 100-150 kW. SiC is additive content, not architectural. Content per MW: $20-50K.

•      Phase 2 — Edge insertion (2027). Medium-voltage conversion expands. Early solid-state transformer (SST) deployments. Grid-to-data-center interface starts using SiC. AI clusters pushing rack density toward 500kW+. Architecture is transitional — partial 800V deployment alongside legacy AC. Content per MW: $50-100K.

•      Phase 3 — Architecture flip (2028-2030). 800V+ DC distribution becomes default for greenfield builds. MW-scale racks. SST replaces legacy iron-core transformers entirely. Power moves closer to compute through wafer-scale integration. SiC replaces silicon as the default conversion technology, not just additive content. Content per MW: $145-275K.

The phase shift between 2 and 3 is structural — not 'more SiC content' but 'SiC instead of silicon across the entire conversion stack.' This is what drives the 5-15x expansion in SiC content per MW between 2025 and 2030.

Where SiC sits in current-generation AI data centers (Phase 1)

Walking through where SiC content lives today in a 100MW data center running Blackwell-class GPUs with 100-150kW racks:

•      Uninterruptible Power Supplies ($8-15K of SiC per MW). 1200V SiC MOSFETs in inverter stages, battery charge/discharge management, and static transfer switches. Per direct procurement contacts at Schneider Electric, Wolfspeed has 30% share of UPS sockets at one of the world's largest power equipment OEMs.

•      Power Distribution Units ($5-8K of SiC per MW). Front-end rectification, branch circuit protection, and rack-level distribution use SiC at higher power densities. SiC content represents 25-30% of PDU bill-of-materials cost today.

•      Battery Energy Storage Systems ($5-12K of SiC per MW). Data centers are adding on-site battery storage for peak shaving, grid services revenue, and behind-the-meter resilience. Current BESS deployments use SiC inverters at 250kW-2MW scales.

•      Front-End Distribution and Switchgear ($1-3K of SiC per MW). Limited today but rapidly transitioning as solid-state transformers replace traditional iron-core distribution transformers.

Total SiC content per MW in Phase 1 architecture: $20-40K. For a 100MW AI data center: $2-4M of SiC.

How architecture flip transforms the math (Phase 3, 2028-2030)

The 800V DC architecture fundamentally changes how power flows through the data center. Instead of multiple AC-to-DC and DC-to-DC conversion stages, power moves at 800V DC across the facility. The traditional 480V AC distribution and 48V intermediate stages are eliminated. Rack power scales from current 100-150 kW to 600 kW within two years and 1 MW by 2029-2030 (per Wolfspeed CEO disclosure).

Vertiv confirmed on their Q1 2026 earnings call that 800V is 'a 2027 thing' with portfolio launches in second half 2026. Schneider sees 800V ramping more aggressively in 2028. By 2028-2030, greenfield AI data center builds default to full 800V architecture.

Where SiC content explodes in MW-scale 800V architecture:

•      Solid-State Transformer Stack ($50-105K of SiC per MW). Full SST function from medium-voltage grid (13.8kV) to 800V DC bus. Includes medium-voltage rectification ($20-40K), DC bus regulation and protection ($10-20K), 800V output stages ($15-30K), and bidirectional capability for grid services ($5-15K). Wolfspeed has commercially available 10kV products — the highest commercially available SiC voltage in the market. Few suppliers can produce reliable 10kV SiC at scale.

•      Rack-Level Power Conversion at MW Scale ($40-70K of SiC per MW). At 1MW racks, the 800V→48V intermediate bus converters carry $25-40K of SiC per MW. The 48V→GPU rail conversion (SiC and GaN contested) adds $10-20K. Rack-level protection and isolation adds $5-10K. As rack power scales 7-10x from current architecture, rack-level conversion SiC content scales similarly.

•      Battery Storage at Behind-the-Meter Scale ($40-65K of SiC per MW). As data centers transition to behind-the-meter power architectures (Bloom fuel cells, gas turbines, eventual SMRs), battery storage scales materially to handle generation intermittency. 800V DC-native battery systems carry premium SiC content. Bidirectional grid-tie capability for grid services revenue adds additional content.

•      High-Voltage Switchgear and Protection ($15-33K of SiC per MW). MW-scale racks with 800V distribution require solid-state circuit breakers ($5-10K), DC fault protection ($5-10K), isolation switches and disconnects ($3-8K), and surge protection at higher voltages ($2-5K). All SiC-based at full architecture flip.

Total SiC content per MW in Phase 3 architecture: $145-275K. For a 100MW AI data center: $14.5-27.5M of SiC. Approximately 7-10x the current architecture content per MW.

The TAM trajectory through 2030

Combining the per-MW progression with capacity delivery forecasts:

Cumulative data center SiC TAM 2026-2030: $9-21B, with most of the value back-loaded to 2028-2030 when architecture flip is complete. The midpoint of $13-15B reflects the wide range of possible deployment scenarios.

This is on top of $4-6B automotive SiC TAM by 2030 (recovering from current trough), $2.5-3.5B industrial power conversion SiC, $800M-1.2B aerospace and defense, and $3.5-5B renewable and grid SiC. Total SiC market 2030: $20-30B, growing from approximately $5-7B today.

The data center segment alone goes from a small portion of total SiC today to 30-40% of the total market by 2030. It's the fastest-growing segment in the entire power semiconductor complex, and Wolfspeed is positioned in the highest-value applications within it.

Why this content trajectory matters for Wolfspeed specifically

The phase shift from $20-40K per MW in 2025 to $145-275K per MW in 2030 is not linear — it's an architecture transition where SiC replaces silicon in the conversion stack. The companies positioned for that transition are those with:

•      Differentiated high-voltage SiC technology (10kV MOSFETs commercially available)

•      Vertically integrated substrate-to-device manufacturing

•      Demonstrated 300mm SiC wafer capability for advanced packaging integration

•      US-domestic manufacturing footprint for tariff and policy advantages

•      Customer relationships with Tier 1 power equipment OEMs

Wolfspeed checks every box on this list. The phase shift framework is essentially the bull case for Wolfspeed expressed as architectural transition rather than market share competition. As the market goes through Phase 3 architecture flip, the entire SiC value chain repricers, and US-domestic vertically integrated leaders like Wolfspeed capture disproportionate share of the value expansion.

Part Two: The TAM Has Doubled in Three Quarters  

The AI power semiconductor market has roughly doubled since Wolf emerged from bk based on hard data points that materialized over the past three quarters. This is a combo of macro level hyperscaler spend rising and 800v timeline pull-fwd.

Hyperscaler capex revisions

Q3 2025 consensus 2026 hyperscaler capex (top 5 US): approximately $450-500B

Q1 2026 reported and confirmed 2026 hyperscaler capex: approximately $720B

Magnitude of the revision: +$220-270B in incremental annual capex committed.

At roughly 8-12% power infrastructure share of total data center capex, that incremental commitment alone implies $18-32B of additional power infrastructure spending added to 2026 expectations versus what was knowable at the September 2025 fresh-start date.

Capacity buildout forecasts

Q3 2025 forecasts: approximately 70-80 GW incremental data center capacity 2025-2030

Current forecasts (Goldman, JLL, McKinsey): approximately 92-122 GW incremental

Magnitude: +20-50 GW of additional capacity baked into forecasts.

At $80,000-110,000 SiC content per MW in 800V architecture, an additional 20-50 GW of capacity translates to $1.6-5.5B of incremental SiC TAM added to forecasts in three quarters alone.

800V architecture timeline acceleration

Q3 2025 expectation: 800V deployment 'around 2030'

Q1 2026 actuals: Vertiv confirmed '2027 thing' with H2 2026 portfolio launch, Schneider 2028 ramp, Wolfspeed disclosed Mohawk Valley qualified for 800V

Timeline pull-in: 2-3 years earlier than the September 2025 baseline.

This matters because SiC content per MW expands 7-10x in 800V architecture. Pulling the timeline forward by 2-3 years materially accelerates the SiC demand curve.

Per-rack power consumption validation

Q3 2025: Rubin Ultra rack expectation approximately 600 kW (uncertain)

Q1 2026: Wolfspeed CEO confirmed 600 kW racks 'in two years' with 1MW racks by 2029-2030

Current racks: 100-150 kW

Power density: 4-7x increase per rack, validated by multiple primary sources.

Behind-the-meter validation

Q3 2025: behind-the-meter was a 'concept' if grid challenges emeerged

Q1 2026: Vertiv confirmed 'here to stay,' Bloom $5B Brookfield deal and recent Oracle full commit deal

Impac: behind-the-meter architectures require more sophisticated power electronics thus driving SiC content higher per MW.

The TAM math

Pulling these data points together for cumulative incremental SiC TAM 2026-2030:

September 2025 view (fresh-start PP bk emergence baseline): $5-7B

Current view (Q1 2026 data): $10-12B

The AI power semiconductor TAM has roughly doubled in three quarters.

Part Three: The Enterprise Value Hasn't Budged

Wolfspeed emerged from Chapter 11 restructuring on September 29, 2025. Fresh-start accounting required marking PP&E to fair value at emergence date. The bankruptcy court used a 20.1% WACC and trough cash flow projections through 2034 to establish the enterprise value at exactly $2.6 billion.

That valuation reflected trough coniditions: punitive discount rates capturing emergence-stage risk premium, conservative cash flow assumptions, full bankruptcy stigma(supplier diversificaton/brand equity), and still totally auto anchored with zero credit for the AI capex narrative that subsequently exploded across the entire complex.

Where comparable AI-exposed names traded on the same date

September 29, 2025 is a nice reference date because you can see clearly how different everything else traded.

AAOI and AXTI are illustrative of broader AI capex derivative pricing the retail market has fallen in love with and are not power semi comps. I obviously have addressed AAOI in its pre-optomania window for this potential which has continued far past I think is remotely justified. I can make a similar case for navitas which I also been on the bull side for its nvidia positioning value which it has also tended to exceed lately in terms of incremental real business value progress as the tam is just so comparatively small over the next 3 years even with robust 800v adoption.

All of the AI euphoria pricing is still ahead of WOLF. The stock move so far is just the bankruptcy unwind playing out and we have snapped back to those levels reached on my post re-emergence pitch. What hasn't happened yet is the market recognizing SIC in AI power infrastructure value the way it has for VICR, NVTS, and countless broader material/component others like AAOI, and AXTI.

Part Four: The MP/Real Asset Value Below Replacement Cost Template

This is the part of the trade that stands out the most for me. Wolfspeed at $2.5B EV is similar template that worked for MP Materials pitch. I.E irreplaceable US-domestic critical infrastructure trading below replacement cost while the Trump administration assigns a strategic premium to the exact profile.

We did the MP trade last summer. It’s a good comp because it reminds you what catalyst structure repriced the asset, how long it took, and what magnitude of move resulted when the strategic value got properly recognized.

The MP COMP

MP Materials owns Mountain Pass in California, the only operating rare earth mine in the United States. The setup then:

•      Asset value below sunk cost. Mountain Pass cost $1.7B+ to bring back to operational state. MP traded below that asset value for an extended period because the market priced it as a commodity miner with China-driven price compression.

•      Strategic value the market wasn't pricing. US-only rare earth supply chain. Critical for defense, EVs, wind turbines. China controlled 60-70% of global supply and 85% of processing. The strategic value was real but invisible in commodity miner multiples.

•      Administration policy as catalyst. DPA Title III designation, defense contracts, Pentagon investment, eventual Apple supply agreement. Policy made the strategic value bankable.

•      Time-to-rebuild moat. You cannot open a new rare earth mine in 18 months. Permitting alone takes 5-10 years in the US. Existing operational assets at trough valuations were structurally underpriced because new entrants couldn't compete.

•      The re-rating happened violently when strategic value was recognized. MP went from sub-$15 to $80+ when Pentagon investment and the Apple deal validated the strategic premium. Asset value framework collapsed into strategic-asset framework. Notably, multiple expansion was non-linear.

Wolfspeed's actual sunk capex: ~$5B in cash spent over the past five years

USA based cash capex is a good starting point for real-asset value in this decoupling obsessed ai infra world.

Cumulative cash capex by fiscal year (purchases of property and equipment per cash flow statements):

Cumulative cash capex for the current Wolfspeed footprint: approximately $5.1 billion gross. Net of long-term incentive reimbursements from state and local agreements: approximately $4.6 billion in actual shareholder cash. The big spend was concentrated in FY2024 ($2.27B in a single year) when Mohawk Valley Phase 2 expansion and Siler City construction were at peak burn.

Plus pre-FY2020 historical investment: Durham operations and legacy R&D infrastructure represent decades of additional capex that is now fully depreciated on the books but operationally functional. Estimated cumulative pre-FY2020 SiC-related investment: $1-2B.

Total cumulative sunk capex for current Wolfspeed footprint: approximately $6-7 billion in actual cash spent.

Current enterprise value: approximately $2.5 billion.

Commodity cycle investing, which is what this is AI DC shit is right now, always focuses on replacement cost for these types of assets. By that measure, the market is valuing Wolfspeed at roughly 30-40c  on the dollar of actual cash already invested to build a leading edge domestic SIC manufacturing footprint. Seeing as how all things physical material/semi now getting tighter, that sunk cash cost is gonna understate present replacement cost notably.

Replacement cost: what it would take to build it today based on some external comps

Asian competitor capex per 200mm SiC fab (announced or spent):

•      ROHM Kunitomi (200mm SiC fab): $2.8B

•      Mitsubishi Kumamoto (200mm SiC): $1B+

•      Bosch/TSI semiconductors (200mm SiC retrofit): $1.5B+

•      Infineon Kulim Module 3 (200mm SiC): $5B+ for full buildout

•      STMicro Catania (integrated SiC): $5B program

Wolfspeed's full footprint replacement cost: $7-11B realistic. The high end captures the integrated substrate-to-device vertical integration that few competitors match.

Comparable transaction validation: Coherent paid $7B for II-VI in 2022 for materials and substrate manufacturing capability that was smaller than Wolfspeed's current operations.

Wolfspeed's current EV is below what Asian competitors are paying just to build single 200mm SiC facilities from scratch. That's the asset value disconnect thesis in relative comp form.

Why Wolfspeed Triggers the MP Reflex Trade 4 Me..

 Asset value below sunk cost

MP traded below the $1.7B sunk cost of bringing Mountain Pass back to operational state. Wolfspeed at $2.5B EV trades below the $5.1B in sunk cash capex actually spent over the past five years (and below the $6-7B all-in including legacy operations). Same setup, larger absolute numbers.

Strategic value the market isn't pricing

US-only large-scale 200mm SiC manufacturing. Critical for:

•      AI data center power infrastructure (the binding chokepoint on hyperscaler buildout)

•      EV powertrain (administration EV preference plus tariff dynamics)

•      Defense applications (DPA Title III designation makes this explicit)

•      Grid infrastructure (federal grid modernization priorities)

•      Aerospace and space applications (Wolfspeed has long-standing customer base)

China controls roughly 30-40% of global SiC substrate supply and growing. The exact same supply chain concentration framework that got MP repriced applies to SiC. The administration is increasingly explicit about wanting US-domestic priorities (not just alternatives) across all critical materials and components.

Administration policy as catalyst (already happening just not getting front-center billing yet)

Unlike MP where the policy catalyst was prospective when the asset value framework first developed, Wolfspeed's policy catalysts are largely already in place:

•      April 2026 DPA Title III determination. SiC explicitly designated as essential to national defense and grid infrastructure. This is the same designation that anchored the MP strategic premium.

•      Section 48D credit raised to 35% from 25%. Wolfspeed already received $700M in tax refunds. Continuing capex qualifies for premium tax treatment.

•      May 2026 EU auto tariffs at 25%. Reinforces 'domestic production preference' theme. Foreign competitors face structural disadvantages on US-bound product..

•      Federal procurement preference. US-domiciled critical technology companies get structural preference in defense and infrastructure contracts. SiC for grid hardening explicitly qualifies.

The administration policy stack is more developed for Wolfspeed today than it was for MP at the equivalent stage of the trade. The difference is the market hasn't pounced on it yet.

Time-to-rebuild is the Bottleneck MOAT

This is the thing that I’m channeling the most from past commodity investing bull markets. The moat that capital alone cannot solve..time! Bottlenecks get love and Mohawk utilization today is not gonna have those arguing this is not a “bottleneck”. Think that’s an easy counter considering how much of a given the SIC demand is going fwd IF THESE DATACENTERS ARE BUILT.

•      200mm SiC fab construction: 24-36 months minimum. Equipment delivery, clean room construction, process qualification. Can’t really compress this with money.

•      Customer qualification cycles: 18-30 months for new SiC parts. Hyperscalers, automotive OEMs, and Tier 1 industrial customers require extensive qualification before production design wins. Even if you build the fab, you don't get the revenue for two years.

•      Process IP and yield curve: 3-5 years of accumulated knowledge. SiC manufacturing yield curves are notoriously difficult. Wolfspeed has 35+ years (Cree predecessor since 1987) of accumulated process knowledge. New entrants start at low yields and ramp slowly.

•      Substrate supply: vertically integrated control. Wolfspeed produces its own SiC substrates at Durham and Siler City. Most competitors buy substrates from limited suppliers (Wolfspeed itself, Soitec, Coherent, Resonac). Substrate supply is the binding constraint for the entire industry.

Even with unlimited capital, building a full Wolfspeed equivalent from greenfield takes 4-6 years and substantial capital risk. The replacement cost framework understates the time-to-rebuild moat because it doesn't capture the qualification and yield ramp.

Channel Intel? From broken commodity capital sink to USA strategic asset and champion

Worth noting because it's the closest analog for what AI capex narrative recognition does to a US-domestic semiconductor manufacturer.

For years, Intel was treated as a commodity capital sink and tech process loser..

That framework collapsed when:

•      CHIPS Act passed and US-domestic semi manufacturing got explicit policy premium

•      China-Taiwan geopolitical risk became concrete in market pricing

•      AI capex created urgent need for non-Asian advanced semi manufacturing

•      Foundry strategy gave Intel optionality on hyperscaler and defense customer relationships

Intel went from being valued as broken foundry non-ai moat silicon to being valued as a strategic US-domestic critical infrastructure asset with optionality. Yea, wolfspeed is not in the intel league of problem, but you in many ways its actually far easier and cleaner already on a tech moat position.

Wolfspeed is currently being priced as a commodity SiC manufacturer with ev overhang uncertaintiy/drag. It’s getting zero AI Datacenter moat premium love yet or US strategic asset replacement value. The framework shift that prices it as strategic US-domestic AI power infrastructure with optionality is the trade with a lot more legs. Same template that worked for Intel and MP, applied to a smaller, cleaner execution story, and extremely tight float stock.

The TSMC interposer optionality and 300mm SiC leadership ASSET OPTION

Beyond the clear asset value today and strategic domestic value pillars, Wolfspeed has some serious IP-driven optionality that doesn't exist in most names.

300mm SiC wafer demonstration

Wolfspeed announced 300mm SiC wafer capability which is an industry first. No other competitor has demonstrated 300mm SiC at scale. ROHM, Mitsubishi, Infineon, ST, Onsemi all are still on 200mm or below.

Why 300mm matters:

•      Cost reduction. 300mm wafer reduces per-die cost roughly 30% versus 200mm, which itself was 30% versus 150mm. Compound cost advantage on a percentage basis.

•      Standard silicon equipment compatibility. 300mm is the standard silicon wafer size. Equipment is plentiful, mature, and cost-effective. SiC at 300mm enables integration with standard silicon manufacturing flows.

•      Advanced packaging integration. 300mm SiC enables SiC interposer manufacturing for advanced packaging. This is the bridge between SiC power devices and silicon compute dies in heterogeneous packaging.

The TSMC SiC interposer roadmap

This is the optionality kicker that nobody else in the power semi space has.

TSMC's CoWoS advanced packaging roadmap is the foundational technology for AI accelerator manufacturing (NVIDIA H100, B100, B200, GB200 all use CoWoS variants). Current CoWoS uses silicon interposers.

SiC interposers offer several advantages over silicon for AI accelerator packaging:

•      Higher thermal conductivity (3x silicon) — better heat dissipation for high-power AI chips

•      Higher voltage handling — enables integrated power delivery on the interposer

•      Lower coefficient of thermal expansion mismatch with high-power dies

•      Higher mechanical strength — enables larger interposer sizes for multi-die integration

TSMC has publicly indicated SiC interposer development as part of their roadmap. Wolfspeed is the only company with demonstrated 300mm SiC wafer capability, which is the prerequisite for SiC interposer manufacturing at scale.

If TSMC commercializes SiC interposers for AI accelerator packaging, Wolfspeed is structurally positioned as the substrate supplier. The TAM for advanced packaging substrates is in the multi-billion dollar range, and SiC interposers would be one of the most high-value, technology-differentiated AI narrative driven product lines.

This is option value that the market is currently pricing at zero. The TSMC interposer outcome is uncertain in timing and probability (more like 3yrs out), but the option premium for Wolfspeed should be material given that no other power semi company has the prerequisite technology.

Putting it together: the layered asset value framework

WOLF current EV: approximately $2.5B

What you're getting at $2.5B EV:

You're paying $2.5B EV for cash plus a manufacturing footprint that cost shareholders $5.1B+ to build over the past five years. The market is valuing the operational footprint at less than 30 cents on the dollar of actual cash already spent before counting the strategic premium, time-to-rebuild moat, IP optionality, or business value. Every layer above the cash and sunk capex line is currently priced at zero, and so is arguably the AI product portfolio differentiation.

Hard to argue this isn’t similar to the MP setup. Asset value below sunk cost plus strategic premium plus policy tailwind plus time-to-rebuild moat plus IP optionality, all priced at zero. The catalyst that closes that gap is the same one that worked for MP, market recognition that the strategic value is real and bankable. For MP, it was Pentagon investment plus Apple. For Wolfspeed, it could be an acquisition, a hyperscaler design win disclosure, a TSMC interposer announcement, additional federal funding, or a Renesas action that validates strategic positioning.

Any one of those catalysts could trigger the asset value framework collapse into strategic asset framework and attract commiserate investor interest..

Part Five: Where Wolfspeed Is Actually Differentiated AI Power Tech

Wolfspeed color I have seen from others basically takes a uniform commodity approach to SIC and sprinkles on AI to the rescue on EV oversupply view. Basically, it will be slow moving recovery with not much differentiation in the portfolio, but some as rising tide/ cool breezes here and there. This is not accurate.

Where Wolfspeed is genuinely differentiated

•      10kV SiC MOSFETs (highest commercially available voltage). Wolfspeed has commercially available 10kV products. Infineon and STMicro compete at 6.5kV. Few suppliers globally can produce reliable 10kV SiC at scale. This is the binding application for Solid-State Transformer function in 800V architectures.

•      200mm SiC wafer manufacturing leadership. Wolfspeed Mohawk Valley was the first 200mm SiC fab to reach commercial production (April 2022). Three-year lead on Asian competitors who are still ramping. 200mm wafers reduce per-die cost approximately 30% versus 150mm.

•      300mm SiC wafer demonstration (industry first). Wolfspeed announced 300mm wafer capability December 2024. No other competitor has demonstrated 300mm SiC at scale. Critical for next-generation interposer applications and integration with mainstream silicon manufacturing. Optionality for SiC interposer thesis with TSMC.

•      Vertically integrated substrate to device manufacturing. Wolfspeed produces its own SiC substrates (Durham, Siler City). Few competitors are fully vertically integrated. Substrate supply is the bottleneck for the entire industry — Wolfspeed controls its own supply.

•      35+ years of SiC process IP. Wolfspeed (Cree predecessor) has been working on SiC since 1987. Process knowledge embedded in operations, customer qualifications, and yield improvements. Difficult to replicate even with capital investment.

Where Wolfspeed is not differentiated

•      General-purpose 1200V automotive SiC MOSFETs (highly competitive with ST, Infineon, Onsemi, ROHM, Mitsubishi)

•      Standard discrete power devices (commodity space)

•      Industrial general-purpose applications (Tier 2 at ABB, not Tier 1)

•      Cost competitiveness on commodity products

Why this differentiation framing supports the trade

The high-value SiC applications in AI data centers are exactly where Wolfspeed stands out vs the EV after-thought shift angle:

•      10kV SST function ($25-50K per MW) — Wolfspeed very differentiated according to customers, few competitors

•      200mm wafer cost structure — Wolfspeed three-year lead

•      Vertically integrated substrate supply — Wolfspeed controls bottleneck

•      300mm wafer optionality — Wolfspeed industry first, potential TSMC interposer thesis

Expert Call reference fit/violation

The 30% UPS share at Schneider validates the incumbent position in the highest-value AI data center socket. Eaton/Vertiv support Schneider differentiation and share takes. The Tier 2 status at ABB reflects commodity industrial positioning is irrelevant to the AI data center battle. The Soitec competitive threat affects Materials business but not the device-level differentiation in 10kV applications.

Wolfspeed's value proposition is being the only company with both differentiated technology in high-value AI applications AND vertical integration of the substrate supply chain. Every other competitor either lacks the technology lead (ST, Onsemi) or lacks the vertical integration (most Asian competitors).

Figure pure end AI DC SIC demand is at least 1.5bl in 2028 and you shud get the path ahead clearly.

Part SIX: The Float Dynamics That Make This Intriguing Trade Wise

Here's what distinguishes Wolfspeed from every other AI-exposed name. The effective free float is extraordinarily tight, and the cap structure verifies as 93% aligned holders.

Verified cap structure (post-CFIUS, May 2026)

Total common shares outstanding: 48.3M

Holder breakdown:

•      Apollo / Senior Noteholder Block: 28M shares (57%) — mostly locked, some registration rights on approximately 5M

•      Renesas common: 16.85M shares (35%) — locked, board seat through Aris Bolisay, strategic relationship

•      Legacy pre-petition equity: 2.2M shares (4.6%) — tradeable

•      Other (prior converts, management): 1.3M shares (2.7%)

Plus dilutive instruments not in basic count:

•      Renesas warrant ($23.95 strike, ITM): 4.94M shares

•      Renesas 2L Convertible ($203.6M, due 2031): 11.1M shares

• Non-Renesas 2L Convertible ($331.4M, due 2031, $18.35 strike): 18.1M shares

Fully diluted basis depends on convert settlement form. With cash-settleable converts treated as debt: ~55M shares. Maximum theoretical dilution if all converts share-settle: ~103M shares. Realistic share count at meaningful stock appreciation likely settles in the 70-85M range as Wolfspeed mixes cash and share settlement."

True tradeable float and short interest

Of the 48.3M common shares outstanding, approximately 93% are held by aligned holders (Apollo, Renesas, pre-petition equity). The remaining shares constitute the true tradeable float.

True tradeable float: approximately 7M shares

Components: 2.2M legacy pre-petition equity plus approximately 5M tradeable noteholder portion with registration rights.

FINRA-reported short interest: approximately 11M shares

Short interest as percentage of true tradeable float: 157%

Short interest exceeds the entire tradeable float by 57%.

Float comparison to peer set

Having preached extensively about gaming others behaviors vs fundamentals you kind of need to look at this table and appreciate the setup still. Wolfspeed's effective tradeable market cap is approximately $257M as of publishing which is basically a penny stock float in this animalistic ai capex environment. Now put that  against quarterly revenue of $168M just to remind yourself this is a biz with some real material scale already and not just slide decks. It’s important context when you look at where SIC business is heading in dc and what others have gotten on far less so far.

•      Enterprise value below trough fresh-start book on $2.5B vs $5-15B implied at current power semi multiples

•      Tightest effective float in the complex (7M shares vs $1.8B basic market cap)

•      Real business floor underneath ($168M quarterly revenue, more than NVTS and AXTI combined)

•      Real strategic value (DPA designation, Renesas anchor, 10kV technology, 300mm wafer, interposer optionality)

•      Real political tailwind (Trump admin domestic preference)

•      AI data center direct beneficiary across multiple SiC applications

•      Capital structure reset with T. Rowe and Fidelity anchor positioning

The asymmetry is the cleanest setup since NVTS at $4 and VICR at $30. Better underlying substance than either, at a fraction of the EV, with tighter float dynamics, and against a market that has doubled in size during the period where WOLF specifically did not get repriced.

Part Seven: What Could Go Wrong

AI capex narrative crack

If hyperscaler capex revisions go negative, the entire AI-exposed semiconductor complex de-rates simultaneously. Not happening this month clearly, but its always out there in terms of risks to consider and weight.

Specific narrative break for Wolfspeed

Seeing as I generally think this just gets better till the stock has gone too high to reassess it; it’s not top of mind right now. But any operational event that breaks the asset value relative catch-up narrative, is something to keep an eye out for.

Float Relief

If Renesas decides to monetize their 35% stake into strength, the float dynamics flip and the technical trade setup weakens. That being said hard to see how you are not already winning bigly if that issue materializes.

Part Eight:  Sizing and Time Frame

The risk-reward at $25-37 entry is asymmetric in a way that still supports meaningful position size for traders who understand the AI power capex derivative dynamics. The downside is asset-value-protected at trough replacement cost levels with a far firmer balance sheet and months of improving visibility ahead. The upside is comparable re-rating dynamics that have already played out in adjacent names, amplified by the immediate tightest float in the complex.

Immediate (Q3 FY26 earnings May 5)

Doesn't need to be amazing, just needs to not break the operational narrative and the slowly building AI interest. On the flip side, AI data center catching some sort of more broader coverage in terms of how incrementally robust the thesis is gonna be q to q for next several q’s could be a surprise.

Near-term (6-18 months)

Asset value catch-up plays out as the market recognizes the disconnect with comparable names. L1 refinancing executes successfully (mostly already done). Specific hyperscaler design wins emerge. Vertiv  Investor Day May 19-20 reinforces 800V demand thesis.Operational improvement continues. Interposer optionality becomes more bankable as TSMC roadmap clarifies. EV’s rebound faster than expected.

Recap

Been here done that vibe for us in power semi. Better risk setup than  NVTS at $4 and VICR at $30 if you consider both the macro picture and stock specific info we had at time on those. Wolf has real assets at trough ev kind of like vicor plant then, real revenue floor at $168M per quarter (vs nvts level leap of faith on nvda), and good laggard rationalization as emerging from bankruptcy and auto ev sector stigma have been prime focus vs dc.

The bankruptcy court priced WOLF at $2.6B EV using 20.1% WACC and trough cash flows on September 29, 2025. Today the EV is $2.6B, while the AI data center SiC market doubled, hyperscaler capex revised up $220-270B, the 800V architecture timeline pulled in 2-3 years, and every comparable AI-exposed name re-rated 2-20x.

Effective tradeable float is approximately 7M shares at $257M tradeable market cap vs peers with less scale/ ai infra strategic value at 10x that.

You don’t need to nail operational inflection on the print in this. It's catch-up to where the rest of the complex already moved, amplified by the tightest float dynamics in the entire AI power semiconductor space. I think that's appealing.