A fresh leak making the rounds on hardware forums and X today is complicating one of the tidiest assumptions in chip marketing: that more cores automatically mean a higher power bill. A specification table attributed to leaker Laurent’s Choice, and now circulating across outlets including Tom’s Hardware, TechPowerUp, PCMag Middle East, SoyaCincau, and Sportskeeda Tech, lists seven unreleased Intel desktop chips from the upcoming Nova Lake generation. The standout detail isn’t the core count. It’s that the chip with the most cores in the table doesn’t carry the highest power rating, and the chip with fewer cores doesn’t necessarily carry a lower one. Intel has not confirmed any of it.
That mismatch matters because for a decade, desktop CPU marketing has leaned on a simple story: bigger core count, bigger power envelope, bigger price tag. The leaked Nova Lake table breaks that pattern in a way that, if accurate, says as much about Intel’s manufacturing and packaging strategy as it does about any single chip’s gaming benchmark. This article walks through what the leak actually claims, why the power-to-core relationship looks unusual, and what it would mean for buyers watching the fight between Intel’s upcoming Core Ultra 4000 series and AMD’s X3D lineup, which has its own 128MB-cache leak in play.
What the Nova Lake Leak Actually Shows
According to the table attributed to Laurent’s Choice, Intel’s Nova Lake desktop lineup is anchored by a chip called the Core Ultra 9 4970K BFC. The leak lists it with 8 performance cores, 16 efficiency cores, and 4 low-power efficiency cores, for 28 total cores, paired with a 32-execution-unit integrated GPU and a 125W base power rating. That 125W figure is the detail drawing attention, because it sits well below what desktop flagship chips have carried in recent generations once you account for boost behavior, and because other SKUs lower in the stack are reportedly rated at the exact same 125W despite having fewer cores.
The reported lineup also includes a Core Ultra 9 4950K at 24 cores and 125W, a Core Ultra 7 4870K BFC at 24 cores and 125W, and a Core Ultra 9 4900 BFC at 22 cores but only 65W. Line those four up and the pattern breaks cleanly from the old assumption. A 28-core chip and a 24-core chip reportedly share an identical 125W rating. A 22-core chip sits at little more than half that. Core count and power class, at least in this leak, are not moving together the way they used to.
None of this is confirmed by Intel. The company’s official product pages make no mention of Nova Lake model names, core splits, cache figures, or power numbers, and leaked engineering tables routinely shift before a chip reaches retail. Multiple outlets are reporting the same seven-chip structure independently, which is why the story is getting broad pickup today, but independent repetition of a leak is not the same thing as an Intel press release.
The BFC Suffix and bLLC Cache, Explained
Several of the leaked SKUs carry a “BFC” suffix that doesn’t match any existing Intel naming convention, a detail tracked closely by outlets like Wccftech. Reporting around the table has not pinned down what the three letters stand for, and Intel has said nothing about it, so treat any explanation of the suffix itself as unverified. What is more consistently described across the coverage is a cache designation called “bLLC,” shorthand reportedly for big last-level cache. That is a stacked or expanded L3 pool, conceptually similar in intent to AMD’s 3D V-Cache approach, though the leak does not establish a specific capacity figure that Intel has confirmed, nor any performance delta versus a non-BFC part.
The timing of a big-cache SKU is not an accident if the leak holds up. AMD built an entire halo-product strategy around stacking extra cache onto gaming chips, and Intel’s own coverage history on this site has already tracked how seriously the company appears to be taking that threat. A Nova Lake variant specifically carrying a cache-focused suffix reads as a direct answer to that pressure, even though the exact competitive gap it’s meant to close remains unverified. For background on how the cache side of this story developed, see the earlier reporting on Intel’s BFC cache bet against AMD’s X3D lead.
Why More Cores Didn’t Mean More Watts This Time
The simplest reading of the leaked numbers is that Intel is no longer treating “base power” as a direct proxy for silicon size. Base power, often labeled PL1, is a sustained-operation ceiling, not a hard cap on what the chip can draw during a boost window. A 28-core part and a 24-core part sharing a 125W base rating doesn’t mean they behave identically under load. It means Intel picked a common thermal and power target for that tier of the stack and let core count, clock speed, and cache size vary underneath it.
That’s a meaningfully different design philosophy than “more cores, more power,” and it lines up with a trend that’s been building since Arrow Lake: Intel’s recent desktop generations have put more emphasis on performance-per-watt than raw TDP escalation, a shift the company has leaned on publicly as Arrow Lake’s efficiency gains became a selling point against both its own Raptor Lake predecessor and AMD’s lineup. If Nova Lake continues that approach, the headline core count becomes less useful as a shorthand for how hot or power-hungry a chip will run, and buyers comparing SKUs by looking at “cores” alone will be comparing the wrong column.
There’s a second explanation worth considering: binning. Chipmakers routinely take the same physical die and sort it into different SKUs based on how it performs during testing. A 28-core die that can’t clock quite as high, or that leaks slightly more current, could be set to the same conservative 125W base power as a 24-core die built to hit higher frequencies. Readers who want the mechanics of how desktop chips get pushed and measured against their power limits can see the testing approach used on current-generation hardware in our coverage of Nova Lake’s USB4 and PCIe 5.0 compliance testing, which draws on the same leak family.
Nova Lake Leaked Lineup: Core Count vs Power Rating
| Leaked Model | P-Cores / E-Cores / LP E-Cores | Total Cores | Reported Base Power | Cache / GPU Notes |
|---|---|---|---|---|
| Core Ultra 9 4970K BFC | 8 / 16 / 4 | 28 | 125W | bLLC, 32-EU iGPU, vPro |
| Core Ultra 9 4950K | Not fully detailed in leak | 24 | 125W | No bLLC variant reported |
| Core Ultra 7 4870K BFC | Not fully detailed in leak | 24 | 125W | bLLC |
| Core Ultra 9 4900 BFC | Not fully detailed in leak | 22 | 65W | bLLC |
Every figure in that table is a leak, not a confirmed Intel spec. The pattern is still instructive: the 28-core and 24-core parts share an identical base-power class, while the 22-core part drops to roughly half that, suggesting power targets are being set by product tier and intended use case rather than scaled strictly off core count.
How This Compares to Arrow Lake and Raptor Lake
Raptor Lake, Intel’s 13th and 14th-generation desktop family, became something of a cautionary tale for chasing peak performance through power, a saga extensively documented by enthusiast outlets like AnandTech at the time. Flagship Raptor Lake chips carried a 125W base rating too, but their PL2 boost ceiling climbed well past 250W, and the generation was later dogged by stability complaints tied to voltage and power delivery that pushed Intel to issue microcode fixes. Arrow Lake, which followed, pulled back from that approach and prioritized efficiency, trading some peak multi-threaded throughput for lower power draw and cooler operation under sustained load.
If the Nova Lake leak is accurate, it reads as a continuation of the Arrow Lake philosophy rather than a reversal back toward Raptor Lake-style power scaling. Sharing a 125W base rating across both a 28-core and a 24-core part suggests Intel is still managing its top desktop tier around a fixed thermal target instead of letting the power number climb in step with every additional core. Whether that holds at the PL2 boost level, where separate and lower-confidence figures in the broader leak ecosystem have cited numbers in the 250-300W range for top single-die parts, is a different question the leak itself does not definitively answer.
Competitive Context: Intel vs AMD’s X3D Strategy
AMD’s X3D line has spent several generations proving that cache size, not just core count or clock speed, moves gaming benchmarks. The company’s recent Ryzen 9 5900X3D leak, which pointed to a chip carrying 128MB of cache, underlines how far AMD is willing to push that formula even on chips built around an older core architecture. Cache, in other words, has become its own competitive battleground independent of core count or raw clock speed, and both companies appear to be leaning into it rather than away from it.
Intel’s reported bLLC-equipped BFC variants look like a direct response to that pressure. If the leak’s cache claims hold up even loosely, Intel would be fielding a cache-heavy SKU at the top of its desktop stack for the first time in this generation, aimed squarely at the gaming workloads where AMD’s X3D chips have built their reputation. The catch is that cache capacity alone doesn’t guarantee a benchmark win. AMD’s X3D advantage comes from the specific placement and latency characteristics of its 3D-stacked cache, not merely from having more megabytes on the die, and nothing in the current Nova Lake leak confirms that Intel’s bLLC implementation matches that latency profile.
Market Impact: What a Flatter Power Curve Means for Buyers
For buyers and system builders, a power rating that doesn’t scale predictably with core count changes how shopping by spec sheet works. Historically, a buyer could use TDP as a rough proxy for cooling requirements and power supply headroom, with higher-core chips assumed to need beefier cooling. If Nova Lake’s top two tiers really do share a 125W base rating regardless of whether they’re running 24 or 28 cores, that shorthand stops working. Builders will need to look past the core count and check the specific SKU’s rated power, boost behavior, and cache variant before sizing a cooler or power supply.
It also affects how retailers and reviewers frame these chips once they launch. A 28-core flagship that doesn’t demand a dramatically bigger power budget than a 24-core sibling is an easier upsell for Intel’s partners, since it removes one of the traditional objections to buying the bigger chip (that it runs hotter and needs pricier cooling). Whether that framing survives contact with real-world boost behavior, where PL2 figures can run well above the base rating, is something only independent lab testing after launch will settle.
Historical Context: Intel’s Power Rating Confusion
Intel’s relationship with the term “TDP” has been messy for years. The company has long distinguished between a base power figure (PL1) and a maximum turbo power figure (PL2), but marketing materials and retail listings have frequently blurred the two, leaving buyers to assume the lower, headline number represents real-world power draw. That confusion peaked during the Raptor Lake era, when flagship chips rated at 125W base routinely pulled well over double that under full multi-core load, prompting enthusiast outlets to publish their own measured power figures alongside Intel’s official specs.
The Nova Lake leak’s 125W figures are already being described in coverage as base power rather than a true ceiling, which is the correct framing but also the same framing that caused confusion two generations ago. If Intel wants buyers to actually understand what these chips draw, the company will need to be more explicit about PL2 figures at launch than it has been in the past, rather than letting a clean base-power number do the marketing work while boost behavior gets buried in a footnote.
What Remains Unconfirmed
It’s worth being precise about what this leak does and doesn’t establish, because leaked tables have a habit of hardening into assumed fact by the time a chip actually ships. Unconfirmed: whether these are official Intel specifications at all, since the company has not acknowledged the table. Unconfirmed: any release date, retail pricing, or final product positioning for any of the seven SKUs. Unconfirmed: what the BFC suffix stands for. Unconfirmed: whether the final shipping chips will carry exactly the core counts, cache sizes, and power ratings listed here, given that Intel has changed specs between leak and launch on past generations.
What is consistent across the reporting is the shape of the table itself. Multiple outlets, independently citing the same leaker, are describing the same seven-chip structure with the same naming scheme, which is a stronger signal than a single-source leak would be. That consistency is why this is being treated as a credible leak rather than speculation, even though “credible leak” and “confirmed specification” remain two different things.
Predictions: Where Nova Lake’s Power Story Goes Next
- Intel will likely clarify or push back on at least some of the leaked figures once Nova Lake moves closer to an official unveiling, following its usual pattern of letting leaks circulate before confirming a narrower set of real specs.
- Expect independent reviewers to focus heavily on PL2 boost power once engineering samples surface, since base power alone has repeatedly understated real draw in past Intel generations.
- If the bLLC cache claims hold up in any form, AMD’s X3D marketing will likely respond with earlier or more aggressive cache-capacity messaging on its next desktop refresh rather than waiting for a head-to-head benchmark cycle.
- Motherboard partners will probably treat the flatter power curve as a selling point for mid-range boards, marketing them as capable of running higher-core-count Nova Lake SKUs without the premium VRM hardware that Raptor Lake’s power spikes once demanded.
- Expect continued leak-table churn between now and launch, with at least one more round of “corrected” specs, since that has been the pattern across Intel’s last three desktop generations.
How to Read Leaked CPU Spec Tables Responsibly
A few habits help separate signal from noise when a table like this circulates. First, check whether multiple outlets are citing the same primary source or independently corroborating it; a single screenshot repeated by ten blogs is not ten confirmations. Second, distinguish base power from boost power explicitly, since conflating the two is the single most common error in early coverage of any new CPU generation. Third, treat suffixes and naming schemes as provisional, since Intel has renamed SKUs between leak and retail launch before. Readers tracking the broader hardware landscape, including recent pricing moves on competing silicon, can see how quickly leaked figures can shift once official benchmarks land in our coverage of Snapdragon X2 versus Intel Core Ultra benchmarks and the ongoing AMD EPYC Venice comparison against Xeon and Nvidia’s Vera CPU.
Nova Lake in the Broader Desktop CPU Timeline
| Generation | Flagship Base Power | Design Emphasis | Status |
|---|---|---|---|
| Raptor Lake (13th/14th Gen) | 125W base, high PL2 boost | Peak multi-threaded performance | Shipped, later patched for stability issues |
| Arrow Lake (Core Ultra 200) | 125W base, efficiency-focused | Performance-per-watt | Shipped |
| Nova Lake (leaked, Core Ultra 4000) | Reportedly 65-125W depending on SKU | Unverified; leak suggests flatter power-to-core scaling | Unconfirmed, not yet announced |
The table above frames Nova Lake’s reported numbers against Intel’s two most recent desktop generations. The emphasis column for Nova Lake is explicitly marked unverified because it is drawn entirely from the leak and reflects an inference from the leaked numbers, not an Intel statement.
What Enthusiasts and Builders Should Watch For Next
The next real checkpoint for this story isn’t another leak, it’s Intel’s own roadmap disclosures, which typically narrow from a wide leaked table to a smaller set of confirmed SKUs a few months before launch. As tech coverage elsewhere has noted with past Intel leaks, builders planning a Nova Lake upgrade should treat every number in today’s leak as a placeholder rather than a shopping-list spec, and specifically avoid buying cooling or power hardware sized around leaked figures that could shift before retail availability. The cache story is worth tracking separately from the power story, since a bLLC-equipped part that underperforms AMD’s X3D chips in gaming benchmarks would undercut the competitive logic of adding the cache in the first place, regardless of how the power numbers shake out.
Frequently Asked Questions
Is the Intel Nova Lake leak officially confirmed?
No. Intel has not confirmed the model names, core counts, cache figures, or power ratings in the leaked table. The information is attributed to leaker Laurent’s Choice and has been republished by multiple hardware outlets, but it remains unverified until Intel issues an official statement.
What is the Intel Core Ultra 9 4970K BFC?
It’s the name attached to the reported Nova Lake flagship in the leaked table, listed with 8 performance cores, 16 efficiency cores, and 4 low-power efficiency cores for 28 total cores, a 32-EU integrated GPU, and a 125W base power rating. None of these figures are Intel-confirmed.
Why don’t more cores mean a higher power rating in this leak?
The leaked table lists a 28-core and a 24-core part sharing an identical 125W base-power rating, while a 22-core part drops to 65W. That pattern suggests Intel may be setting power targets by product tier and binning rather than scaling the rating strictly with core count, though the exact reasoning behind the leaked numbers is not confirmed.
What does the BFC suffix mean on Nova Lake chips?
It’s unconfirmed. Coverage of the leak has not established what the three letters stand for, and Intel has not commented on the naming scheme.
What is bLLC cache?
Reporting around the leak describes bLLC as “big last-level cache,” an expanded L3 cache pool reportedly featured on select BFC-suffixed SKUs. No official capacity figure has been confirmed by Intel.
How does this compare to AMD’s X3D cache strategy?
AMD’s X3D chips use 3D-stacked cache to boost gaming performance, and a recent leak pointed to a Ryzen 9 5900X3D variant with 128MB of cache. Intel’s reported bLLC approach looks aimed at the same competitive space, but nothing in the current leak confirms it matches AMD’s latency profile or gaming performance gains.
When will Nova Lake actually launch?
No release date has been confirmed. Any specific launch window circulating alongside this leak should be treated as speculation until Intel provides an official timeline.
Should I wait for Nova Lake instead of buying a current CPU?
That depends on your timeline and needs, but given that none of the leaked specs, pricing, or release date are confirmed, it is premature to delay a purchase solely based on this leak. Builders with an immediate need should evaluate currently shipping chips on their own merits.
Related
- Intel Nova Lake Leak: 28 Cores, 144MB Cache [2026]
- Intel’s Nova Lake ‘BFC’ Cache Bet Takes Aim at AMD’s X3D Lead [2026]
- AMD Ryzen 9 5900X3D Leak Reveals 128MB Cache Chip [2026]
- Intel Nova Lake Clears USB4 80Gbps, PCIe 5.0 Tests [2026]
- Snapdragon X2 Beats Intel Core Ultra by 31%, Costs $750 Less [2026]
