Equity research

Lumentum at ECOC indicated that one emerging ELS customer could ultimately represent a larger opportunity than Nvidia because of the optical intensity of the architecture.

$NVIDIA(NVDA.US) $Alphabet(GOOGL.US) $AMD(AMD.US) $Lumentum(LITE.US)

UBS: Nvidia's CPO suppliers

$NVIDIA(NVDA.US) $Taiwan Semiconductor(TSM.US) $Teradyne(TER.US) $Advanced Semiconductor Engineering(ASX.US)

According to the IPO prospectus of Anthropic seen by Reuters.

> Anthropic reported a net ‌loss of $42 billion in 2025, and plans to spend $518 billion on cloud, computing and infrastructure obligations in coming year.

> Revenue grew 12-fold in 2025 to nearly $4.6 billion, even as the company lost more than $8 billion on an operating basis, excluding writedowns of various liabilities mostly tied to previous fundraising.

> The AI lab spent $7.33 billion on compute and infrastructure last year, a threefold surge from 2024, accounting for more than half of its $12.65 billion in total operating expenses.

> The near-$42 billion net loss included a roughly $34 billion accounting charge that reflected an increase in the ‌estimated value ⁠of financing that could eventually turn into Anthropic shares, rather than money the company spent running its business.

> Anthropic had cash, cash equivalents and short-term investments totaling $20.28 billion as of December 31, the prospectus shows.

$Alphabet(GOOGL.US) $Amazon(AMZN.US) $Microsoft(MSFT.US) $Meta Platforms(META.US) $AMD(AMD.US) $NVIDIA(NVDA.US)

Evercore ISI: Marvell's 800G DSP share to remain high and perhaps increase 80+% as Broadcom prioritizes, although, 1.6T share expected to increase in 2027 vs 2026.

$Broadcom(AVGO.US) $Marvell Tech(MRVL.US)

Barclays: total sales for the ethernet market is expected to increase from ~$20 billion in 2025 to $80 billion in 2031. AI scale-out remains the largest segment, while AI scale-up expands fastest, with a 413%/198% yoy unit growth in 2027 and 2028, respectively. AI-related demand is estimated to be 79% of units in 2031, compared to 62% in 2025.

$NVIDIA(NVDA.US) $Applied Optoelectronics(AAOI.US) $Lumentum(LITE.US) $Coherent Corp.(COHR.US) $Credo Tech(CRDO.US) $Macom Tech(MTSI.US)

Nice coverage by @FuturumEquities on Coherent from ECOC 2026.

> Anderson told the ECOC audience that PhotonLink adds roughly $30 billion of integrated optics opportunity to the company’s 2030 addressable market, raising the total from $60 billion for the existing portfolio to $90 billion.

> The company disclosed 2 long-term agreements with leading hyperscale AI datacenter customers for its ultra-high-power laser platform, spanning both CPO and NPO, and put its content opportunity at up to $15,000 per 100 Tbps switch or compute chip on 200G optical lanes when it supplies the complete solution.

> More than 300 million InP lasers shipped, more than 1 billion photodetectors, more than 650 million optical components, and more than 250 million lenses. Capacity is expanding to match, with GaAs lines supporting more than 1 billion VCSEL arrays per year.

> InP capacity doubled last year, a quarter ahead of schedule, and a further doubling is planned through CY2027 on the industry’s first 6-inch InP production ramp.

> Production timing frames the revenue story with scale-out CPO ramping in the December quarter, scale-up CPO and NPO following in 2H CY2027, and chip-to-chip landing in the 2029-2030 window.

$Coherent Corp.(COHR.US)

Jefferies: AI transceiver forecast now pushing to nearly 180 million units in CY27, including 90 million 800G and 80+ million 1.6T units. Growing to 245 million in CY28. Material production volume for Scale-Up CPO has likely pushed to CY29.

$Applied Optoelectronics(AAOI.US) $Lumentum(LITE.US) $Coherent Corp.(COHR.US) $Macom Tech(MTSI.US) $NVIDIA(NVDA.US) $Broadcom(AVGO.US) $Marvell Tech(MRVL.US)

After holding a conference with major memory makers, BofA remains confident in the memory super-cycle continuing through 2028-2030 with no downturn, as high ASPs, strong demand, and supply tightness continue.

$Micron Tech(MU.US) $DRAM $EWY $DISK $SK Hynix(SKHY.US) $Sandisk(SNDK.US)

ABF substrate S/D gap is expected to become progressively worse year over year, reaching a S/D gap of 46% in 2028. Up from just 1% in 2026.

Hello ABF substrate department? Yes, please send us more ABF we need it desperately!

Long-term contracts for 12-inch silicon wafers are expected to rise next year

Industry surveys estimate that the price of new long-term contracts for 12-inch epitaxial wafers will increase by 15% to 25%, while the price of polished wafers may increase by more than 40%, with some second-tier customers seeing increases of over 50%, and there is still room for further price increases in 2028.

Currently, the spot price of 12-inch silicon wafers continues to rise. Analysts predict that the spot price may surpass existing long-term contract prices in the fourth quarter of this year, becoming an important support for the renegotiation of new long-term contracts in 2027.

From the demand side, analysts believe that AI will continue to drive demand for advanced logic chips. Overall CoWoS capacity is projected to increase by 70.9% year-on-year in 2027, while shipments of GPU and AI ASIC chips are projected to increase by 44.9% year-on-year. Overall shipments of Intel, AMD, and NVIDIA CPUs are also projected to increase by 36% year-on-year, which will simultaneously boost 12-inch wafer starts and silicon interposer demand.

Memory capacity expansion is another source of demand support. Analysts predict that DRAM bit supply will only increase by 22% year-on-year in 2027, which will still be insufficient to fully support the growth in demand.

With both prices and shipments growing in tandem, GlobalWafers' operations are expected to expand significantly starting in 2027. Analysts predict that its ASP for 12-inch silicon wafers will increase from US$92 in 2026 to US$122 in 2027, an annual increase of approximately 32.6%, and further rise to US$166 in 2028, an annual increase of approximately 36.9%.

Shipments of 12-inch silicon wafers are estimated to increase from 11,907 kwpq in 2026 to 16,174 kwpq in 2027 and 21,237 kwpq in 2028.

$GlobalFoundries(GFS.US) $Micron Tech(MU.US) $Intel(INTC.US) $AMD(AMD.US) $NVIDIA(NVDA.US) $Taiwan Semiconductor(TSM.US) $Alphabet(GOOGL.US)

Rothschild & Co: Nebius and CoreWeave ecosystem

$Nebius(NBIS.US) $Coreweave(CRWV.US) $Microsoft(MSFT.US) $Amazon(AMZN.US) $NVIDIA(NVDA.US) $AMD(AMD.US) $Alphabet(GOOGL.US)

Samsung Securities: Key Terafab Beneficiaries

Terafab initially plans to leverage Intel's 14A process technology and supply chain, as a result, these global and domestic equipment vendors are positioned as key early beneficiaries.

$ASML(ASML.US) $Applied Materials(AMAT.US) $Lam Research(LRCX.US) $KLA(KLAC.US) $Intel(INTC.US) $SpaceX(SPCX.US)

BofA: expect U.S. top hyperscalers (Amazon, Microsoft, Oracle, Alphabet) to grow cloud revenue to $975 billion in 2028, up from $290 billion in 2025. Meanwhile, China's top hyperscalers (Alibaba, Tencent, ByteDance, Baidu) cloud revenue to increase to $94 billion, up from $28 billion.

$Alphabet(GOOGL.US) $Amazon(AMZN.US) $Microsoft(MSFT.US) $Oracle(ORCL.US)

General Manager of Microsoft Taiwan says the demand for AI computing power will continue to exceed supply for the next 2-3 years, as a result, the company is expanding the construction of data centers in Taiwan from 2 to 4.

$Microsoft(MSFT.US) $NVIDIA(NVDA.US) $AMD(AMD.US) $Alphabet(GOOGL.US) $Amazon(AMZN.US)

BofA: We do not expect “de-spec” or lower memory content in new GPUs/ ASICs; instead, we assume a further increase, to 1,000GB per Rubin Ultra or 384GB for the new Rubin, vs 288GB currently.

$NVIDIA(NVDA.US) $AMD(AMD.US) $Alphabet(GOOGL.US) $Amazon(AMZN.US) $Microsoft(MSFT.US) $Meta Platforms(META.US)

DS Investment & Securities: SiC & GaN

Market Growth & Opportunities:

> The adoption of 800V architectures creates new growth opportunities for 8inch foundries, challenging traditional Si power semiconductors.

> SiC Power Market: Expected to grow at a CAGR of +20%, scaling from $3.4 billion (2024–2025) to $10.3 billion by 2030F.

> GaN Power Market: Expected to experience even faster growth at a CAGR of +44%, rising from $0.6 billion to $3.0 billion by 2030F.

Role Breakdown & Material Specialization:

> SiC: Excels in high-voltage, high-power, and high-temperature environments, serving as a partial substitute for Si IGBT. Main applications include EV inverters/OBC, 800V EVs, AI datacenters (AI DC), and energy storage systems (ESS).

> GaN: Excels in high-frequency, high-power-density, and high-efficiency power conversion, serving as a partial substitute for Si MOSFETs. Key application areas include chargers, adapters, power supply units (PSU), AI datacenters, robots, and solar power.

Infrastructure & Trends:

> Datacenter Penetration: The penetration rate of SiC/GaN within datacenter power systems is projected to increase from 17% in 2026F to over 30% by 2030F, driven heavily by new high-frequency, high-density conversion needs (such as 800V to 54V/12V transitions in AI datacenters).

> Manufacturing Shifts: SiC is currently transitioning from 6-inch to 8-inch fabrication lines, while GaN relies heavily on 8-inch GaN-on-Si processes. Long-term transitions up to 12-inch wafers are also being evaluated.

HSBC: Server DRAM prices should continue rising

amid strong demand for AI and general servers.

$Micron Tech(MU.US) $DRAM $EWY $SK Hynix(SKHY.US)

KB Securities: with higher HBM4 shipments, we expect Samsung Electronic's HBM market share to jump from 33% in 2Q26 to ~40% in 4026.

$Micron Tech(MU.US) $DRAM $DISK $EWY

Funda AI: "Our interviews suggest that frontier labs remain committed to training and advancing model capabilities internally, while additional evaluation and remediation may slow the pace of public releases ONLY."

One of the employees they interviewed, including the head of sales at one of the labs said:

"A slowdown aimed at improving safety is actually making overall compute demand increase, not decrease."

$Alphabet(GOOGL.US) $Meta Platforms(META.US) $Microsoft(MSFT.US) $Amazon(AMZN.US) $Oracle(ORCL.US) $NVIDIA(NVDA.US) $AMD(AMD.US)

TrendForce: Global Power Demand Capacity

> Global data center power demand capacity is expected to expand from 122.9 GW in 2025 to 161 GW in 2026.

> The share of AI servers is likely to exceed 40% by 2027, whereas the share of general servers is expected to decline from around 40% historically to 25.6% by 2027.

> AI servers accounted for approximately 25% of total data center power demand capacity in 2025, and are projected to rise to 33.4% in 2026.

> The share of AI servers is likely to exceed 40% by 2027, whereas the share of general servers is expected to decline from around 40% historically to 25.6% by 2027.

> The global data center power supply-demand gap in 2030 is estimated at 268 GW.

> Based on current grid data, the U.S. data center power supply-demand gap is estimated to exceed 170 GW by 2030.

$NVIDIA(NVDA.US) $GE Vernova(GEV.US) $AMD(AMD.US) $Alphabet(GOOGL.US) $Amazon(AMZN.US) $Microsoft(MSFT.US) $Dell Tech(DELL.US)

Samsung Securities: Hyperscaler Compute Business Models

Hyperscalers operate through two primary methods: 1) Computing rental and 2) Model inference services, both of which can achieve a Gross Profit Margin (GPM) of 68% to 81%.

> Model Inference Advantages: Model inference services feature a large revenue base and high margins, which is why Meta focuses on deploying its capacity toward in-house services rather than plain rentals.

> Value of Compute Scarcity: As seen in examples like SpaceX, simple rental services based on compute scarcity can achieve revenue and margin levels comparable to model inference services.

> Widening Gap: The flexibility and profitability gap between operators capable of deploying and servicing massive compute infrastructure versus those that cannot is expected to widen further.

Chart Breakdown

> Compute Rental Service GPM (Left Chart): Shows $21.7B in revenue, $6.9B in cost of goods sold (COGS), resulting in $14.8B in gross profit, yielding a 68% GPM.

> Open-Weight Model Inference Service GPM (Middle Chart): Shows $36.6B in revenue, $6.9B in COGS, resulting in $29.7B in gross profit, yielding an 81% GPM.

> Hyperscaler GPM Comparison by Compute Business (Right Chart): Compares compute rental (GPM 68%, $14.8B gross profit), model inference with 35% training (GPM 71%, $16.9B gross profit), and model inference with 0% training (GPM 81%, $29.7B gross profit).

$Alphabet(GOOGL.US) $Amazon(AMZN.US) $Meta Platforms(META.US) $Microsoft(MSFT.US) $SpaceX(SPCX.US) $Oracle(ORCL.US) $Nebius(NBIS.US)

Samsung Securities: Hyperscalers to have a temporary FCF deficit period spanning from 2Q26 to 3Q28, followed by a dramatic expansion in FCF from 2029 through 2030.

> CapEx Growth as a Rational Investment for RPO: The increase in capital expenditure is driven by rapid growth in Remaining Performance Obligations (RPO)—long-term signed contracts—requiring front-loaded CapEx to fulfill future revenue.

> Strong Future Visibility: For instance, Amazon highlighted that a substantial portion of its additional capacity for 2027 is already contracted, with considerable portions of 2028 capacity also pre-booked.

> Break-Even and Payoff Timeline: Server and network investment break-even occurs in under 3 years on average, allowing for meaningful FCF generation for the subsequent 2 to 3 years.

$Meta Platforms(META.US) $Amazon(AMZN.US) $Dell Tech(DELL.US) $NVIDIA(NVDA.US) $Microsoft(MSFT.US) $Oracle(ORCL.US) $Alphabet(GOOGL.US)

Goldman Sachs: companies we visited see pricing competition remaining healthy, with 800G optical module pricing decline within single digits in 2027, and 1.6T optical module price to sustain at $700+.

$Coherent Corp.(COHR.US) $Lumentum(LITE.US) $Applied Optoelectronics(AAOI.US) $Macom Tech(MTSI.US)