E108: Doxing debate, Nuclear fusion breakthrough, state of the markets & more

The prediction concerns milestones in the 2030s (production-level, commercially relevant fusion pilot plants) and 2040s (grid-scale deployment). As of November 30, 2025, those years have not yet occurred, so the timeline cannot be confirmed or falsified.

What we do know today:

• The National Ignition Facility (NIF) has repeatedly achieved fusion ignition with energy gain in the lab (starting December 5, 2022, and with improved yields through 2023–2025), but NIF is a scientific experiment, not a continuously operating, grid-connected power plant. It does not constitute a commercial or pilot power facility. (annual.llnl.gov)
• Private companies are planning pilot and grid-scale fusion plants, but none is yet operating:
  • Helion Energy has a power purchase agreement with Microsoft targeting a first fusion power plant delivering at least 50 MW to the grid by 2028, and construction on the “Orion” plant site in Washington state began in 2025. This would be a commercial demonstration ahead of Friedberg’s 2030s estimate if it succeeds, but as of now it is only a goal under construction, not a demonstrated, operating plant. (helionenergy.com)
  • Commonwealth Fusion Systems (CFS) plans its ARC fusion power plant in Virginia to deliver grid power in the early 2030s, with the SPARC demonstration device aiming for first plasma around 2026; again, these are future plans, not operating plants. (reuters.com)

Because:

1. No continuously operating, commercially relevant fusion pilot plant is yet online or delivering power to the grid, and
2. The prediction’s key time windows (2030s for demonstration, 2040s for grid-scale deployment) lie fully in the future relative to 2025,

there is currently no way to say whether Friedberg’s specific 2030s/2040s timeline will ultimately prove correct or incorrect. It remains too early to judge, so the appropriate assessment is inconclusive.

The prediction’s time window is 2022–2032, so as of 30 November 2025 we are only about three years into a ten‑year horizon.

Evidence about the underlying trend:

• Non‑state actors (e.g., ISIS, Mexican cartels, Yemen’s Houthis, Ukraine’s volunteer units) have already used weaponized commercial drones with GPS‑based autopilots and improvised munitions, demonstrating that relatively sophisticated precision‑style attacks are possible without being a state military.
• Consumer and DIY platforms like DJI drones and open‑source autopilot stacks (e.g., ArduPilot, PX4) support waypoint navigation using GPS coordinates out of the box; adding an explosive payload is technically straightforward for a determined group, though still non‑trivial and illegal.
• However, the specific scenario Sacks describes—that continuously publishing an individual’s precise GPS coordinates would make it “pretty easy” for non‑state actors to use precision GPS‑guided weapons for assassination—is not clearly documented as a demonstrated or widely accessible capability. The risk is widely discussed in security circles, but it remains more of a plausible threat model than a routinely observed or clearly commoditized capability.

Because:

1. The deadline (2032) has not yet arrived, and
2. The claim is about future feasibility and ease, not a clearly falsifiable event that has either happened or not,

we cannot definitively say the prediction is right or wrong at this point. It is still developing and depends on subjective judgments about what counts as “pretty easy” and “feasible” for non‑state actors.

Therefore the appropriate classification is "inconclusive" (too early to tell).

The prediction is explicitly about costs ~10 years after 2022 (around 2032), based on an IAEA forecast: that grid‑scale solar would fall from about $0.03/kWh to ~$0.015/kWh, and solar + storage LCOE would be around $0.03/kWh.

As of today (November 30, 2025), only ~3 years of that 10‑year window have elapsed. Long‑term LCOE forecasts from agencies (e.g., IEA, IRENA, IAEA and similar bodies) typically give projections for 2030–2050, and current real‑world LCOE numbers for utility‑scale solar plus storage are still well above the $0.03/kWh combined level in most markets; many analyses place current utility‑scale solar without storage in the low‑$0.02 to ~$0.04/kWh range, and with storage significantly higher, depending on region and financing.

However, because the prediction’s main checkable date is ~2032, and we are only in 2025, it is too early to determine whether costs will in fact reach those levels by that time. The partial data we have now cannot definitively confirm or falsify a 2032 cost target; the learning curve for solar and storage costs is path‑dependent and sensitive to policy, supply chains, interest rates, and technology breakthroughs.

Therefore, the correct status for this prediction is “inconclusive (too early)” rather than right or wrong.

Friedberg predicted that ITER, a ~$30B production‑scale fusion demonstration facility in Europe, would be “online” and operating by the end of the 2020s.

Current information shows this cannot happen under ITER’s own official schedule:

• In July 2024, ITER adopted a new baseline in which first plasma is delayed to at least the early/mid‑2030s (around 2033–2034), with deuterium–deuterium operations starting in 2035 and deuterium–tritium (burning plasma) operations in 2039. (en.wikipedia.org)
• A 2024 U.S. Congressional research report likewise notes that ITER announced it would not be fully operational (with burning plasma) until 2039, with substantial cost increases. (congress.gov)
• Independent coverage of the revised baseline (e.g., The Chemical Engineer, DDay.it) also reports research operations beginning in the early 2030s and D–T fusion not before 2039. (thechemicalengineer.com)

Given these officially adopted plans, ITER will not be online and operating as a fusion demonstration facility by 2029; its operational milestones are now placed firmly in the 2030s. Even if further schedule changes occur, they would have to pull a 2030s roadmap forward by nearly a decade, which is incompatible with all current project planning. Therefore, the prediction that ITER would be online and operating by the end of the 2020s is wrong based on the information available today.

Chamath said the Coupa–Thoma Bravo deal was the start of a “tidal wave” of private‑equity‑sponsored deals in tech, especially SaaS, over the following years. Looking at 2023–2025, that broad directional call did play out.

Key evidence that a multi‑year PE wave in tech/SaaS did emerge:

• Coupa itself was taken private by Thoma Bravo in an ~$8 billion all‑cash deal that closed in 2023, marking a major software/SaaS take‑private exactly in the timeframe he was talking about. (thomabravo.com)
• 2023 saw a series of large PE take‑privates of software and SaaS companies, including Sumo Logic (SaaS analytics) by Francisco Partners for $1.7 billion and Qualtrics by Silver Lake and CPP Investments for $12.5 billion. (franciscopartners.com) S&P Global noted that global PE public‑to‑private deals reached 96 by Oct. 25, 2023—the highest annual total in 16 years—with software singled out as a popular target and Qualtrics and New Relic highlighted among the year’s notable software take‑privates. (spglobal.com)
• PitchBook data cited by TechCrunch shows 136 PE‑led take‑privates globally in 2023, up 15% from 2022, and by mid‑2024 there were already 97 such deals, 46 of them in the technology sector alone—on pace to match 2023’s high level. (techcrunch.com) Those 2024 tech take‑privates include many SaaS or software firms such as Adevinta, Alteryx, PowerSchool, Squarespace and Nuvei. (techcrunch.com)
• By 2025 the trend is very clear in software/SaaS specifically. RBC Capital estimates that software M&A activity in 2025 is up 78%, with private‑equity software deals more than doubling versus the prior year as PE buyers hunt for undervalued SaaS and software assets. (businessinsider.com) A Bespoke Partners analysis reports that in just the first half of 2025, enterprise SaaS companies were involved in 405 private‑equity transactions out of 671 SaaS deals overall, underscoring intense PE focus on SaaS. (linkedin.com) Regionally, for example, PE investments in India’s enterprise SaaS sector reached $1.38 billion in the first seven months of 2025, a 66% increase over the entire year 2024. (economictimes.indiatimes.com)
• The activity has clearly extended “into other realms” of tech beyond classic horizontal SaaS: PE firms have bought cybersecurity companies like Darktrace (Thoma Bravo), education software providers PowerSchool and Instructure, and various infrastructure and data‑analytics software firms. (ft.com)

Important nuance:

• Some tech‑specific M&A datasets show that 2023 alone was not an immediate step‑function higher in PE tech deals versus 2021–2022. Cooley’s 2023 tech M&A review finds private equity’s share of tech M&A fell to a six‑year low (27%), with only 16 PE take‑privates of US‑listed tech companies in 2023, down from 21 in both 2021 and 2022. (jdsupra.com) European/US tech‑only data similarly show PE take‑private volume peaking in 2021–2022 and moderating in 2023 before re‑accelerating. (2023.stateofeuropeantech.com)

Putting it together, the multi‑year pattern after late 2022 is exactly what Chamath described: a sustained and then intensifying wave of private‑equity‑backed acquisitions of tech companies, heavily weighted toward SaaS but spreading across software verticals. Even though 2023 was a softer year for tech‑only PE relative to the 2021–2022 peak, the broader 2023–2025 data show a pronounced and growing PE wave in tech/SaaS, with Coupa among the early high‑profile deals in that cycle. On balance, that makes his prediction substantively right, albeit with a less immediate spike in 2023 than the word “tidal wave” might imply.

It’s too early to know whether this full-cycle prediction is right or wrong.

1. Time horizon issue (core reason for “inconclusive”)
Chamath’s claim is explicitly cycle-wide: that “over the current cycle beginning in 2022” private equity (PE) will buy many private software companies, cut headcount, run them more efficiently, and on average generate roughly 1.2x–1.7x MOIC with few money‑losing deals, in line with historical PE performance.

PE buyout funds typically have holding periods of ~4–7 years before exits, so realized MOIC for deals done starting in 2022 generally will not be known until the late 2020s or early 2030s. Industry data on returns by vintage year show that even funds from the early 2020s are still mostly in the investment/hold phase, with only partial realizations and volatile marks due to changing interest rates and valuations.
Because the “current cycle” starting in 2022 is clearly still ongoing as of late 2025, we cannot yet determine the eventual average MOIC or the fraction of money‑losing deals for that cohort.

2. Observed trends that are directionally consistent, but not dispositive
Even though we can’t fully score the prediction, several parts of the narrative do line up with observed trends since 2022:

• “Buy a ton of private software companies” – There has been a large wave of PE‑backed software take‑privates and buyouts (e.g., Thoma Bravo, Vista Equity Partners, Silver Lake and others remain very active in software and enterprise SaaS). Public reporting and deal databases show that software remains one of the most targeted sectors for PE globally.
• “Fire lots of people” / run them “hyper efficiently” – Numerous PE‑backed software companies (and tech companies more broadly) have announced substantial headcount reductions and cost‑cutting since 2022 as interest rates rose and investors emphasized profitability and efficiency.

These patterns support the mechanics of what Chamath described, but they do not yet tell us if the final fund‑level economics will average 1.2x–1.7x MOIC with few losers for this 2022+ vintage.

3. Why we still cannot score MOIC and loss rates

• Most 2022–2025 software buyouts are still held on PE books, with values based on interim marks; these marks can change materially before exit.
• Exit markets (IPOs, strategic sales, secondary buyouts) have been choppy post‑2022, so ultimate MOIC is highly path‑dependent on future capital‑market conditions.
• Public sources do not yet provide a stable, realized-return dataset specific to the post‑2022 software buyout cohort that would allow us to say, confidently, whether the realized average will fall in the 1.2x–1.7x range with few money‑losing deals.

Because the key measurable components of the prediction (average MOIC and loss frequency over the full 2022+ cycle) cannot yet be observed, the correct scoring as of November 30, 2025 is “inconclusive (too early)”, even though early qualitative evidence on deal activity and layoffs is directionally aligned with his setup.

Available data show that 2023 was clearly a tougher demand year for software and SaaS, but there is no sector‑wide measurement of “new customer business / new bookings vs. 2022” that would allow a clean yes/no on Sacks’s specific 50% figure.

What we can see:

• Sacks’s quote on the episode is indeed: “You should expect your new business to be roughly 50% of what it was. Next year it’ll be 50% of what it was last year. That’s my rule of thumb for most companies. New business down 50%.” (allinpodcast.readablepods.com)
• Gartner’s post‑hoc market‑share analysis shows worldwide enterprise software revenue grew 11.0–11.1% in 2023 to about $785–788 billion, rather than shrinking, despite macro headwinds. (gartner.com) Earlier forecasts for 2023 had also called for high single‑ to low double‑digit growth in enterprise software spend, and those broad numbers were borne out. (saastr.com) This tells us overall software revenue and spend continued to rise; it does not directly tell us how much of 2023’s new bookings were down vs. 2022.
• Bessemer’s Cloud 100 Benchmarks Report notes that for the 2023 Cloud 100 (100 top private cloud/SaaS companies), the average revenue growth rate dropped from ~100% in 2022 to ~55% in 2023 – i.e., growth was “nearly halved” in that cohort. (bvp.com) That is directionally consistent with Sacks’s idea that the pace of new business would fall sharply, but it’s a revenue‑growth statistic for a specific elite subset, not a bookings‑level statistic for “most software companies.”
• Other industry summaries of the SaaS market also report that the market size increased from roughly $237B in 2022 to about $273B in 2023 (~15% growth), even while describing 2023 as a “year of efficiency” with slower growth and more cautious buyers. (fox.agency) Again, these figures speak to overall revenue/market size, not to new‑logo or new‑ARR bookings as a percentage of 2022.

The key problems in judging the prediction are:

1. No unified bookings metric: New‑customer bookings are not reported consistently across vendors, and there is no global dataset aggregating “2023 new customer business vs. 2022” for “most software companies.” Public sources focus on revenue, ARR, and sometimes total billings, not the specific “new customer bookings” slice.
2. Level vs. growth: Where we do have broad data (e.g., Cloud 100), the main story is that growth rates roughly halved, not that new‑business levels themselves dropped to 50% of the prior year. That’s qualitatively similar but not the same as “new bookings are 50% of 2022 in absolute dollars.”
3. Heterogeneity: The evidence we have is heavily skewed toward large public vendors and top‑tier private cloud companies. Sacks’s comment was about “most companies,” including a long tail of smaller and mid‑market software firms for which we lack systematic bookings data; operator anecdotes vary widely.

Putting this together: the direction of Sacks’s call – that 2023 would see a pronounced slowdown in new business for many software/SaaS vendors – is supported by multiple sources. But his specific quantitative claim that, in calendar 2023, new customer bookings for most software companies would fall to “roughly 50%” of 2022 can’t be rigorously confirmed or falsified from public data. Because of the absence of a comprehensive, comparable bookings metric, the prediction’s accuracy has to be rated as ambiguous rather than clearly right or clearly wrong.

Public data on SaaS churn over 2023–2024 is mostly (a) aggregate across all customer segments and (b) reported as broad benchmarks, not precise time‑series for small‑business logo churn.

• KeyBanc’s widely cited Private SaaS surveys show median overall annual logo churn around 13% in both their 2022 and 2023 data, with no evidence of an industry‑wide spike to 25–30% logo churn during that period.(snohat.com) These figures are not broken out specifically for small‑business customers. • For SMB segments specifically, a Bessemer Venture Partners guide (about two years old) already described typical SMB churn ranges like 15–20% annually as “good,” with 10–15% or lower considered better, implying mid‑teens churn was already normal for SMB SaaS prior to or around the start of the prediction window, not a low baseline that later spiked.(bvp.com) • Newer benchmark articles (2024–2025) often say small‑business / consumer SaaS churn of roughly 20–30% annually is common, again as a structural norm rather than as a recently observed jump traceable to 2023–2024 macro conditions.(tomorrowdesk.com) • Company‑specific commentary supports the idea that SMB churn is structurally higher, not that it suddenly surged from ~15% to 25–30%. For example, ZoomInfo’s finance leadership in 2024 talks about natural SMB churn “close to 20%” largely because many small customers eventually shut down, but frames this as an inherent feature of the SMB segment, not as an unusually elevated rate unique to 2023–2024.(in.marketscreener.com)

Because (1) the best quantitative sources aggregate all customer sizes, (2) SMB‑only churn benchmarks are presented as broad ranges without clear pre‑ vs post‑2023 deltas, and (3) no sector‑wide evidence shows SMB logo churn specifically rising from a stable ~15% to a sustained 25–30% in 2023–2024, we cannot confidently say whether Sacks’s specific forecast about magnitude and change in small‑business logo churn came true. The timeframe has passed, but the available data are too coarse and inconsistent to judge the prediction as clearly right or clearly wrong, so the fairest rating is ambiguous.

Evidence from 2023–mid‑2024 shows that software and broader tech companies did face sustained macro headwinds: venture funding into business software fell sharply (for example, European SaaS investment dropped about 59% in 2023 versus the prior year) as rising rates and the Silicon Valley Bank crisis created a “difficult financing climate” and lengthened sales cycles and tightened customer budgets for cloud/SaaS firms.(forbes.com) Large public software companies saw revenue growth and “growth efficiency” fall markedly between 2021 and 2023 even as sales and marketing spend kept rising, indicating it was harder to generate high growth without heavy spending.(mckinsey.com) Industry‑wide, overall IT‑spending forecasts for 2023 were cut roughly in half to low‑single‑digit growth amid economic uncertainty and CIO “change fatigue,” while 2024 was expected to improve but still described as constrained.(gartner.com) At the same time, tech companies executed unusually large layoffs—about 263,000 tech workers were laid off in 2023, with more than 120,000 additional cuts in 2024—as firms adjusted to higher interest rates and slower demand.(marketwatch.com) Analysts in 2024 still characterized the near‑term software outlook as positive but tempered by “heightened macroeconomic uncertainty” through 2024.(morningstar.com) Altogether, this supports Sacks’s qualitative claim that the roughly 4–6 quarters after December 2022 (early 2023 through mid‑2024) would be a period of major economic headwinds that made sustaining very high software growth without aggressive spending significantly harder, so the prediction is best judged as right.

Available data from 2022–2024 show that private equity dealmaking evolved exactly in the way Friedberg described.

1. Bolt‑ons/add‑ons became the dominant form of PE buyouts over these years.

   • PitchBook- and law‑firm analyses report that add‑on acquisitions made up roughly three‑quarters of PE buyouts in 2022 and 2023 and remained around 76% of all PE buyouts in early 2024. (cbh.com)
   • McKinsey and Brookfield/Evalueserve similarly note that add‑ons accounted for about 70% of total PE deal count in 2023, up sharply from prior years. (mckinsey.com)
   • BDO reports that 76% of PE deals by volume in Q3 2023 were add‑ons, explicitly tying this to buy‑and‑build strategies. (bdo.com)
   • Other industry commentary (e.g., KPMG cited via business‑sale.com) describes “around two in three” PE deals being bolt‑ons in the current environment, reinforcing that bolt‑ons are now the typical PE transaction. (business-sale.com)

2. Tech/software is a leading sector for these add‑on strategies.

   • Cherry Bekaert’s 2024–2025 PE outlook notes that technology and healthcare are the top two industries for add‑on deal activity, and that the software segment is driving the tech sector’s rebound, with 926 software PE deals and $134.8B of deal value in 2024. (cbh.com)
   • Evalueserve’s Brookfield-sourced analysis gives concrete software examples where a sponsor used multiple add‑on acquisitions to boost EBITDA and exit valuation, illustrating that this buy‑and‑build/bolt‑on model is actively used in software platforms. (evalueserve.com)
   • While not every large, headline software deal is a bolt‑on (many are new platforms or take‑privates), the data above show that most PE deals by count are add‑ons, and tech/software is one of the most active sectors within that add‑on activity. It is therefore reasonable to infer that most PE acquisitions of software/tech businesses over this period are indeed structured as bolt‑ons to existing PE‑owned platforms.

3. The PE “playbook” has shifted from pure cost‑cutting toward synergy‑building, especially via add‑ons.

   • McKinsey’s 2024 paper notes that traditional PE value‑creation levers like financial engineering and simple operational efficiency have become harder, and that funds are turning to “transformational strategies” and add‑on deals to drive significant growth—i.e., revenue and operational synergies rather than just cutting costs. (mckinsey.com)
   • Brookfield/Evalueserve explicitly describes add‑on acquisitions as a way to realize revenue, cost and capital synergies, citing software cases where cross‑selling and expanded offerings drove higher EBITDA and exit multiples. (evalueserve.com)
   • Other market commentaries on add‑ons (e.g., Coldstream, BDO, Cherry Bekaert) consistently frame them as core to buy‑and‑build strategies—averaging down multiples, building scale, and accelerating growth and synergies—rather than as mere cost‑cutting exercises. (coldstreamcapitalpartners.com)

Assessment: Friedberg predicted that over the “next couple of years” after late 2022, the standard PE playbook in software/tech would center on bolt‑on/add‑on deals to existing platforms, with a focus on cross‑selling and synergy‑building in addition to cost‑cutting. Between 2022 and at least late 2024, add‑ons in fact made up the clear majority of PE buyouts, tech/software was a leading sector for such add‑ons, and industry analyses repeatedly describe these transactions as vehicles for revenue and operational synergies, not just cost reductions. Although deal databases do not break out an exact percentage solely for software add‑ons in public summaries, the combination of (a) overall PE statistics, (b) sector‑specific commentary, and (c) software case studies strongly supports his characterization. On balance, the prediction is best classified as right.

Chamath’s prediction specified a 4–5 year period from 2022 for the true performance of the 2015–2022 venture vintages to become clear, i.e., roughly 2026–2027. The current date is November 30, 2025, which is only about 3 years after the prediction was made in December 2022. Since we have not yet reached the end of the 4–5 year window he gave, it is too early to determine whether his timing estimate (that it will take 4–5 years “to really sort out”) is accurate. Therefore, the correctness of the prediction cannot yet be evaluated.